S&T - 2019 Working notes titled 'Research ecosystems need structural diversity' - learning from past sci/tech revolutions at Bell labs/PARC/LMB to design new institutes

A collection of rough notes & thoughts on the structures of scientific research, with emphasis on the common features of Bell labs/LMB/Xerox PARC

Oct 31, 2022

NB: endnote links inside doc are broken and pointing to a google doc atm, will try to fix… endnotes are at the bottom.

NB2 [added Dec 2022]: Part 2 [click to jump to section] of these notes are the research notes underlying the Lovelace labs vision document, which outlined a set of organisational features common to the LMB/PARC/Bell Labs, and suggested a broad ‘new’ model of research lab learning from that.

Prior to entering 10 Downing Street in April 2020 I spent 7 months in my parents rural house re-reading my notes and studying other books from the past 6 years of researching what we now call applied metascience. I made several ‘working notes’ documents that I thought would be useful to myself and others when I entered government (which was delayed due to the Brexit chaos).

The notes are largely my own words but it was based on work with my brother, and also discussions and input with a range of people including Alexey Guzey (who spent quite a lot of time improving and tidying the notes and suggesting new things to include). The notes sit somewhere between a reference list and an essay, which we intended to mine to turn into separate papers. Due to COVID that only partly happened.

I was recently rereading this one and thought it might be useful to post it. I haven’t redacted anything from the version we circulated in 2020 (which was a little redacted to avoid pissing too many people off). I know it was read carefully by several people behind some of the new institutes in the US. Its blunt in places and I would moderate some of it now, but it shows some of what we were thinking before entering no10.

Part 1 discusses some broad challenges in the science and technology space. I’d spent the years before having a hobby of trying to think and understand how to improve the processes of R&D, particularly in a UK context. I spoke to hundreds of people globally in a kind of ‘focus grouping’ of global science to see what issues people cared about, and what a government should be looking at.

Part 2 [click to jump to] is a set of notes on the common elements between Bell Labs, Xerox PARC, and the Cambridge LMB, based on the reading citing and also my discussions with figures from these institutes like Sydney Brenner, Eric Betzig, and Alan Kay. I’ve since found what I think is a much more compelling way of explaining this argument, based on a talk I gave at Metascience 2022 in California, so I’ll be doing a second blog post on this. But that one will be a lot shorter and have much less detail than I put here. So if you want a reference dump on how those organisations worked, read on. The basic logic here is that rather than studying a single outlier laboratory, you look at the intersection of the Venn diagram to see what might be causal to their unique results.

This second study was the genesis of a ‘twin for ARIA’ program that didn’t make it through the Whitehall machine - the Lovelace program - which aimed to setup a network of research labs operating on fundamentally different principles to academic labs to diversify the structures. (ARIA was only one of several new things we felt needed to be created in UK R&D).

This might be of interest to some people to see what we were thinking prior to entering government, what we thought some key issues were, and it contains quite a lot of resources/citations that could be of use to people thinking about how to reinvent science. The endnotes are full of quotes/refs that could be of use.

Again, it is not comprehensive of what we were thinking/considering nor was it intended as a linear essay, its just the only one that’s legible enough to circulate. There is nothing in here about the practicalities/policies of implementation - its more of a historical analysis. I’ll blog some notes on that later. It also doesn’t cover UK ARPA (now ARIA) much because we’d covered that elsewhere.

Research ecosystems need structural diversity: a collection of rough notes & thoughts on the structures of scientific research, with emphasis on the common features of Bell labs/LMB/Xerox PARC

Unfinished resource/group wiki (not intended to be read linearly) - last updated April 2020

Table of contents:

Some quotations that cover the major themes as a ‘summary’

The structure of the scientific unknown, and a science ecosystem crisis

Is there a slowing of scientific progress?
Finding secrets: the scientific unknown and an entrepreneurial, start-up science
Taking the hidden paths: start-up dynamics and disruption of the prevailing views 1
A science ecosystem climate crisis: central management of research, perverse incentives, and an increasingly Potemkin academy
Wellcome’s 2020 Research Culture report: ‘[research] is about to collapse’

Common features of three of the most successful research laboratories in history

1. Fund scientific communities not scientific factories: “focussed ecosystems pursuing tech visions”
Learn lessons from three of the most transformative research institutes in history
Xerox PARC: the creation of personal computing.
Cambridge MRC laboratory of molecular biology (1950s-70s): the foundations of molecular biology
AT&T Bell Labs: ‘One policy, one system, universal service’
An anti-bureaucracy silver bullet: institutes should be fully internally funded at an elite level
Identify unique and long-term visions
These long-term visions become talent attractors
Form scientific ecosystems, not factories, by recognising the social nature of science
2. Invest long-term in promising world class scientists instead of individual projects
Allow researchers to focus entirely on research and mentoring, not bureaucracy
Invest in exceptional scientists rather than individual projects
Distant central edict cannot control innovation: Give researchers great freedom to pursue ‘heretical’ ideas and realise that ‘impact’ actually means ‘well-received by current opinion’
Use internal evaluation to allow assessment of an individual’s ecosystem contribution, with external evaluation of the institute
Balance internal and external hiring to maintain culture and reward contributions (do not ban internal hires)
Facilitate research via strong core support, technical assistance and resources
A brief summary
3. Organise as self-organising start-up style small teams, not hierarchical bureaucracies
Age balance in research societies: balancing youthful ignorance and senior wisdom
Flatten the hierarchy: positive feedback in academic power and “The slavery of graduate students”
Remaining dynamic: do not give tenure to researchers
Credit assignment & ‘pianists without fingers’: we need research supervisors
Be people number limited: prevent empire building and promote cross-disciplinary collaboration via enforcement of small team sizes
Empower promising young researchers

Some further resources

Appendix - Bureaucracy vs start-up

Some quotations that cover the major themes as a ‘summary’

“Every now and then I receive visits from earnest men and women armed with questionnaires and tape recorders who want to find out what made the Laboratory of Molecular Biology in Cambridge (where I work) so remarkably creative...... creativity in science, as in the arts, cannot be organized. It arises spontaneously from individual talent. Well-run laboratories can foster it, but hierarchical organization, inflexible, bureaucratic rules, and mounds of futile paperwork can kill it. Discoveries cannot be planned; they pop up, like Puck, in unexpected corners.”
First MRC Laboratory of Molecular Biology Director Max Perutz

‘The ARPA/PARC history shows that a combination of vision, a modest amount of funding, with a felicitous context and process can almost magically give rise to new technologies that not only amplify civilization, but also produce tremendous wealth for the society. Isn’t it time to do this again by Reason, even with no Cold War to use as an excuse?’
Alan Kay, Xerox PARC researcher and pioneer of personal computing

‘I am afraid that there will be little tangible left in a later age to remind our heirs that we were men, rather than cogs in a machine.’ John Pierce, Bell Labs researcher and a key player in the 20th century communications revolution (Gertner 2012, pg. 360).

“Heroic lone-wolf entrepreneurs may be the preferred heroes of narratives spun by the media, but history has shown us that teams—and the networks that come from them—are the true engines behind innovation in Silicon Valley and far beyond. No one understood this better than Bob Taylor.” Obituary of ARPA/PARC pioneer Bob Taylor in Wired

‘I learnt very quickly that the only reason that would be accepted for not attending a committee meeting was that one already had a previous commitment to attend a meeting of another organization on the same day. I therefore invented a society, the Orion Society, a highly secret and very exclusive society that spawned a multitude of committees, sub-committees, working parties, evaluation groups and so on that, regrettably, had a prior claim on my attention. Soon people wanted to know more about this club and some even decided that they would like to join it. However, it was always made clear to them that applications were never entertained and that if they were deemed to qualify for membership they would be discreetly approached at the appropriate time.’
Sydney Brenner, pioneer of molecular biology

‘The important breakthroughs come from loonshots, widely dismissed ideas whose champions are often dismissed as crazy.’
Safi Bahcall, ‘Loonshots’

‘If the work you propose to do isn’t virtually certain of success, then it won’t get funded’—Nobel Laureate Roger Kornberg (Lee, May 28th 2007, Washington Post)

‘Even God wouldn’t get a grant today because somebody on the committee would say, oh those were very interesting experiments (creating the universe), but they’ve never been repeated. And then someone else would say, yes and he did it a long time ago, what’s he done recently? And a third would say, to top it all, he published it all in an un-refereed journal (The Bible).’
Sydney Brenner, pioneer of molecular biology

"I recently learned a disagreeable fact: there are influential scientists in the habit of putting their names to publications in whose composition they have played no part. Apparently some senior scientists claim joint authorship of a paper when all that they have contributed is bench space, grant money and an editorial read-through of the manuscript. For all I know, entire scientific reputations may have been built on the work of students and colleagues! I don't know what can be done to combat this dishonesty.”
Richard Dawkins, introduction to the selfish gene [we emphasise that since this was written decades ago, what he describes has in many ways become the system and the norm in biomedical science and it would look odd and harm their ability to get grants for a senior scientist not to put their name on their junior scientists papers regardless of contribution. It is therefore now not ‘dishonesty’ in many cases. A Nobel Laureate told us he puts his name on papers in order to help his students get published in flashy journals at their request, despite feeling deeply uneasy about it. We suggest that rather than authorship, in many cases the lab affiliation should be listed alongside department affiliation]

“... there is a serious problem with incumbents hoarding opportunity. Just make them compete with their grad students and postdocs. Make them get back into the lab. Make them do the analysis. That would sort things out pretty quickly.”

A comment on this piece by a researcher from a free-standing institute

“I strongly believe that the only way to encourage innovation is to give it to the young. The young have a great advantage in that they are ignorant. Because I think ignorance in science is very important. If you’re like me and you know too much you can’t try new things. I always work in fields of which I’m totally ignorant.”
Sydney Brenner, pioneer of molecular biology

Google AI’s view of Academia, [edit: see new 15th Jan Wellcome report for similar]

The structure of the scientific unknown, and a science ecosystem crisis

These notes concern disruptive research and impediments to them.

Is there a slowing of scientific progress?

It has been persuasively argued that technological progress has slowed or at least become far less efficient relative to the early/mid 20th century, when World Wars, the Space Race and the Cold War compelled government investment in high-risk technology beyond that which private capital could provide (see this Kasparov speech [‘reviving the spirit of innovation’, 2013, Oxford] and footnote1)2. This correlates with a historic shift in mindset, from wanting vertical growth (doing new things) to horizontal (more of the same, ie, one to many), beginning in the late 1960s. Quarterly GDP change now matters more than long-term results, and we are living with the long-term consequences of this chronic short-term thinking. At the national level, reaching the moon (1960s) vs a focus on GDP per year (1970s onward) represents this shift in attitude.

Cowen and Southwood (2019) highlight a decline in economic growth, and a decline in general purpose technologies (GPTs)3. For example:

Reproduced from Cowen/Southwood. They write: “Overall, consider Robert J. Gordon’s simple take: ‘U.S. economic growth slowed by more than half from 3.2 percent per year during 1970-2006 to only 1.4 percent during 2006-2016.’”

The physicist turned-researcher-into-research Michael Nielsen4 and Stripe founder Patrick Collison also wrote an essay (“Is science stagnant?”in the Atlantic) in 2018 suggesting that scientific innovation is slowing down since the 1970s. They focussed upon Nobel Prizes, surveying scientists and finding that recent work is not judged by fellow scientists to be of the same transformative type as earlier work5. They find physicists rate the importance of earlier work more highly, with the 1990s and 2000s lacking sufficient Nobel Prizes to be included (1900s included a lot of eccentric Nobel Prizes when the award was new):

Discovery impact scores for physics nobel prizes from the survey of Collison/Nielsen. Note that the 1990s/2000s did not get enough nobel prizes to be included in the analysis.

Nielsen/Collison found similar though slightly better results in other fields6.

A related point is made with respect to the biotechnology industry by Peter Thiel: ‘Despite dramatic advances over the past few centuries, in recent decades biotechnology hasn’t met the expectations of investors—or patients. Eroom’s law—that’s Moore’s law backward—observes that the number of new drugs approved per billion dollars spent on R&D has halved every nine years since 1950.’ (p75, Thiel, Zero to One).

These views are not conclusive as it is difficult to assess innovation (they may even underestimate the problem), but we are certainly not living in the tech future we predicted in the 1950s (Mars habitation, cures for cancer, fast speed of plane travel, etc...—see footnote for historical example7). Almost all progress has been in ‘bits’ not ‘atoms’.

One extreme explanation is that the scientific slowdown is due to science getting harder, with low hanging fruit being gone. This may be the case in part8. Another view is that the slowdown is due a severe shift in the balance of power within the scientific system from a ‘start-up’ style mode to a risk averse, bureaucratised system with power increasingly concentrated in the hands of a small group of more established, senior individuals and bureaucratic staff. As Cowen and Southwood highlight, researchers are increasingly tied into Kafka-esque bureaucracies9. This suggests great riches could be found by enterprises that can create alternatives to this system.

Finding secrets: the scientific unknown and an entrepreneurial, start-up science

Albert Einstein, quoted in The Pittsburgh Press, December 28th 1934.

To understand how we should create new research systems, we should first consider the structure of the landscape of the unknown, and how, historically, such unknowns have become known.

The secrets trap: there is always a sense that we know the landscape of the scientific unknown, that majority opinion knows where to look, and that top-down and senior individuals know what is to come. Yet the established scientific experts at any given moment in history have usually been entirely wrong about the future of their fields. Newton was unexpected in 1663. Darwin’s Origin of Species was a revolution. The papers of Einstein’s annus mirabilis was so unexpected and outside the norm that they were difficult to publish10,11. The ‘leaders’ of today’s science historically have typically not provided tomorrow’s insights, and are unlikely to in the future (some do, but it is by far the minority).

In this way, the origin of molecular biology came in part from physics at the Cavendish, Cambridge—Steve Hsu writes that “many molecular biology departments were established (originally with names like BioPhysics!) against the wishes of "real'' biologists of the era.” Invaders, as Hsu calls them. Likewise, today the cure for Alzheimer’s is likely lying hidden in the unknown, not where the herd is looking. The cure for Alzheimer’s, based on historical precedent, will not come from funding the prevailing view of what constitutes ‘Alzheimer’s research’. See “How an Alzheimer’s ‘cabal’ thwarted progress toward a cure’ by Sharon Begley - are we are about to repeat the mistake with clean energy by flocking to the ‘most likely options’?

As Thiel writes: ‘The best place to look for secrets is where no one else is looking’. A beautiful recent (2015) UK example of apparently blue skies research with major implications for dementia comes from studying hibernating animals such as hedgehogs, finding a potential new brain protecting chemical working in a very unexpected way (see BBC and Nature, and a clue from 1989!)! Completely outside the prevailing orthodoxy.

In Zero to One, Thiel highlights a number of reasons why we are now much less able to find secrets and thus transform our world (in footnotes)12. We shall see that the university system, with tenure and central approval of funding, suffers significantly from these incentive problems that limit its ability to do some kinds of research [see this recent 2020 NBER paper Bhattacharya and Packalen, 2020]. The UK should strongly factor these points into how we fund research, encouraging different systems to collectively explore diverse parts of the unknown rather than low-risk herding (which likely is just replicating other countries' current efforts anyway). We need mechanisms to consciously drive such ‘outside the box’ consensus breaking research. Instead all current pressures and incentives align with satisfying current peers.

Taking the hidden paths: start-up dynamics and disruption of the prevailing views

We need to nurture a diversity of research endeavours. An example of different styles of research is explained in a 2001 speech 13 from then DARPA director Fernando ‘Frank’ Fernandez, who said “People like Dr. Clayton Christensen of Harvard have argued that there are really two kinds of innovation (Fig. 1). There’s the well-planned, evolutionary, sustaining innovation, which is where you have a road map and, based on this road map, you change the way you do business in a fairly predictable manner. Most good organizations have such road maps. The second kind of innovation is radical, disruptive innovation, where something new happens that initially doesn’t fit into and probably doesn’t improve upon the current way of doing business. But if this radical innovation can be nurtured, it will eventually far exceed the performance of the standard business model.”

Figure 1 referred to by former DARPA Director Fernandez - Note how the radical, disruptive research requires accepting a performance drop before a performance increase, and so cannot be found via a locally greedy optimisation.

In the world of business such considerations are much more common: the world of start-ups realises that there are always unconventional yet transformative ideas lurking in the shadows away from clear view, and that large institutes are unlikely to find them. Peter Thiel writes ‘Every one of today’s most famous and familiar ideas was once unknown and unsuspected’ (Thiel, 93).

Despite this diversity of research style, there is very little consideration of how the career structures and power organisation allocation in research systems alters the style of research they produce. There is essentially no consideration of whether PhD->Postdoc->Tenure track etc is the right one for all types of government funded research. Careful, incremental scholarly research has different requirements than, say, development of new experimental methods for biology.

A science ecosystem climate crisis: central management of research, perverse incentives, and an increasingly Potemkin14 academy

‘If there is the tiniest difference between the behavior you want and the incentives you offer, people will find it.’15 Paul Graham, Y Combinator co-founder amongst other achievements, on Twitter 17th Feb 2020

Modern approaches to science bureaucracy often ignore a) the health of the scientific ecosystem and b) the nature of the scientific ecosystem as a complex system with complex incentives, feedback loops and many agents with different priorities. Rather than hand power to the scientists themselves, science bureaucracies often instead think that science can be micromanaged from afar in a top-down manner, selecting for ‘impact’, ‘translatability’ and ‘market potential’ but history and our current predicament suggest at least some caution here. Much of the growth of research policy has arguably harmed research because they try to control something they do not, and arguably cannot, understand (as complex systems prediction is very much in its infancy) 16. We need a more systems-level research policy, evaluating the relationship between parts not summed integration of isolated metrics.

Central management’s reliance on metrics is necessary because a distant bureaucracy can only ever have a scattered, low resolution image of an environment it seeks to control. It is very difficult to assess creative science over the short-term, and attempts to do so focus on what can be remotely quantified (citations, journal impact factor, short term competitor/peer assessment of impact), rather than on what is important (truth, long-term impact, benefit to science ecosystem health (good mentoring etc)17). When central management rewards such metrics, it creates an incentive, meaning people work to optimise the metric rather than what you are intending to reward (this is Campbell’s law18). It is far easier and better rewarded to optimise for short-term attention grabbing than it is to do lasting and important scholarly work, which eventually corrupts the culture of the system as a whole. The system therefore leads not just to what Smaldino and McElreath (2016)19 call the ‘natural selection of bad science’, but actually directly incentivises bad science and promotes more of it (and thus promotes bad scientists) . This, in a nutshell, is why we risk having an increasingly Potemkin academy, and why productivity has become a cheap substitute for ‘excellence’. (One could ask, only part facetiously, whether the UK’s Research excellence framework should more accurately be called the ‘bureaucratic short-term appearance of productivity framework’?)

When we reward being productive over being correct or creative, phenomena such as empire building, taking credit for others work, exploitative practices, poor quality control, misleading analysis and other such problems rise (see part 2). Though they can be detected locally, central bureaucracy is necessarily blind to these phenomena, revealing a strength of local control. We have unintentionally incentivised a large amount of bad science and bad behaviour, usually by good, well-intentioned people.

The long-timescales over which the research system operates mean the symptoms of this take a long time to develop, but are severe - a climate change of the academic ecosystem. Unlike the financial system, there is no way of the system ‘blowing up’ to reveal how rotten it has become, because collective interest amongst scientists mean they are reluctant to speak out (the recent UK government report contains no mention of any of these problems, or any significant problems whatsoever - try to find a single mention of young scientists in the report titled ‘Ensuring 20 a successful UK research endeavour’) (an excellent analogy between modern research and the financial crisis is found from Scientific American blog in 2019 21. The difference is that broken research cannot completely ’collapse’ and reset, it just lingers on, as it is not subject to competitive displacement).

Some unfortunate consequences of this:

From Monya Baker, 2016, Nature - 1,500 scientists lift the lid on reproducibility
“Survey sheds light on the ‘crisis’ rocking research”

Modern academic research is substantially non-replicable: as productivity is rewarded over accurate scholarship (see Ed Yong articles - As Ed Yong wrote in 2019 “They’re an almost inevitable product of an academic world that rewards scientists, above all else, for publishing papers in high-profile journals—journals that prefer flashy studies that make new discoveries over duller ones that check existing work…. People are rewarded for being productive rather than being right, for building ever upward instead of checking the foundations.”
- A large replication studies suggests only half of studies in psychology can be replicated (Ed Yong 2018, Atlantic))
- Ed Yong reported in ‘A waste of 1,000 research papers’, the Atlantic 2019, that ‘Decades of early research on the genetics of depression were built on nonexistent foundations.’22, with the paper suggesting that the foundation of an entire field of medical genetics, and the contents of approximately 1,000 research papers, is entirely statistically spurious.

From Nature’s news website by Brian Owens (2011): “Bayer halts nearly two-thirds of its target-validation projects because in-house experimental findings fail to match up with published literature claims...An unspoken industry rule alleges that at least 50% of published studies from academic laboratories cannot be repeated in an industrial setting.”
An influential paper by Ioannidis from 2005 in PLOS, called ‘Why most published research findings are false’, summarising it as blaming ‘investigator prejudice, incorrect statistical methods, competition in hot fields and publishing bias, whereby journals are more likely to publish positive and novel findings than negative or incremental results’. The opening line of that paper summarises the problem well: ‘There is increasing concern that most current published research findings are false.’23,24 (see25 for a slate star codex piece on this claim).
[Edit feb 2020] Particularly disturbing is the suggestion from a journal editor in chief 26 at molecular brain, a subsidiary of nature, that a substantial fraction of papers cannot provide raw data on request - ‘more than 97% of the 41 manuscripts did not present the raw data supporting their results when requested by an editor, suggesting a possibility that the raw data did not exist from the beginning, at least in some portions of these cases.’ The ‘41’ manuscripts are just under a quarter of all submitted manuscripts handled by the editor.

A luxury journal system which has led to a damaging magazine style of science (see Nobel Laureate Randy Schekman in Guardian), with ‘stories not studies’, high impact paper chasing and competitive (antagonistic) peer reviewing (for solution suggestions see27).
Training far more PhDs and postdocs than there are jobs for (by some estimates a 10:1 ratio, Ben Barres 2017, Nature). - (top graduate students and postdocs are not trainees: they are the real scientists).
Ever-lengthening PostDoc “training” as people need more and more papers to arrive at job interviews with higher metrics, like rainforest trees outcompeting each other for light. This has put particular pressure on families as academics lack stable jobs until their forties and the long-hours training period corresponds to child bearing ages.
Increasing use of graduate students as cheap technicians (What Sydney Brenner called ‘The slavery of graduate students’). In large parts of biology many professors rarely actually do significant amounts of research, instead using large numbers of graduate students to churn out papers. One consequence of this is that the standards for entry to programs have dropped and the systems are not designed for training the most exceptional talent anymore. Brenner makes this point well his interview with Elizabeth Dzeng, available online here.
Operating large labs in order to maximise the number of papers a senior author can take credit for whilst crowding out junior scientists (the declining value of authorship). There needs to be a very fundamental rethink of scientific publishing - the current culture of a senior author, with many in the lab, taking effective intellectual ownership of other’s work via ‘last authorship’. This could be changed with government level ambition (wellcome open science is pushing similar)28.
Professors competing with their trainees and preventing them continuing their research outside (Ben Barres, 2017, Nature).

This matters beyond the academy29: for an example of how the social dynamics of well intentioned people can halt a field's progress, see “How an Alzheimer’s ‘cabal’ thwarted progress toward a cure’ by Sharon Begley, July 2019, perhaps delaying a viable treatment for decades by stifling exploration diversity.

Some of these problems arise from trying to make the academic system do modern research without reforming its career structure—we need more support staff (programmers, engineers, technicians, potentially with accompanying training degrees) and fewer postdocs (see Alberts et al 2014 in PNAS for postdoc overtraining issues). But this is also due to central management, and is a critique of having incentives when there is too big a gap between a) what you want to incentivise (long-term benefits) and b) what you can in practice incentivise (citations # etc)30.

To repeat a core theme, these problems are in many cases an unavoidable result of distant central management attempts, because central management ends up producing systems that they can understand and exert control over rather than actually improving the original system (this is known as legibility, from ‘Seeing like a state’ by James Scott31). As a blog title puts it, ‘every attempt to manage academia makes it worse32’. Central bureaucracies can be blind to much of what really matters in innovative societies.

From Edwards and Roy 2017
‘Academic Research in the 21st Century: Maintaining Scientific Integrity in a Climate of Perverse Incentives and Hypercompetition’.

It will also relatively drive out those who seek scientific excellence as opposed to those seeking scientific fame/power via playing politics. Edwards/Roy 2017 quote Dunn, December 2013 in the Chronicle of Higher Education as saying a major reason for academics leaving the academy is the presence of ‘‘a perverse incentive structure that maintains the status quo, rewards mediocrity, and discourages potential high-impact interdisciplinary work’’ (Dunn, 2013). Edwards/Roy conclude: ‘If a critical mass of scientists become untrustworthy33, a tipping point is possible in which the scientific enterprise itself becomes inherently corrupt and public trust is lost….’. We have placed excerpts from their piece into the footnotes34.

Central bureaucracies are good for scale, not for speed and flexibility and outside the box thinking. In correspondence to Nature in 2012 titled ‘Bureaucracy bypass let research flourish’, Min- Liang Wong highlighted:

Four of the greatest discoveries of the twentieth century — the structure of the atom, quantum mechanics, the theory of relativity and the structure of DNA — were made without project reviewers or grant-giving agencies (B. Gal-Or Cosmology, Physics, and Philosophy p. 493; Springer, 1983).

The bureaucratic process requires a substantial burden of time on their researchers, such as elaborate grant application forms, and it is challenging to find expertise in the areas to centrally evaluate the proposals (Nurse report, 2015)35. In order to get past review panels, work must play into the current established thinking, otherwise it will likely get voted down by someone who thinks it won’t work (see below and Guzey 2019b for peer review being a disaster in a hypercompetitve field36). Further, the peer review process may operate only little better than luck in many cases 37. Recently funders have acknowledged this issue 38 and in several cases begun incorporating random selection in handing out resources 39 (D Adam, November 2019, Nature, ‘Science funders gamble on grant lotteries’).

As Alan Kay wrote of science bureaucracy in a comment on a blog post: ‘Their problems come from feeling that something is wrong if only (say) 35% of the results are breakthroughs. They think this is “not efficient”, but I think this is a hangover from school where students are given easy problems and are expected to solve most of them. The reality in *edge of the art* research is that the problems are not easy, but the ones that get solved then provide enormous leverage to civilization.’ This approach/mindset means that edge of the art research, as Kay calls it, simply don’t get supported. This in turn has led to a drift toward bureaucracy and incrementalism, which has become a vicious cycle.

A simple cautionary tale from Ionnadis 2011, Nature40, ‘.....any system that demands high-risk innovative goals, and requires results, generates potential for exaggeration.’41

Douglas Adam captures the problem well in the Hitchhiker's Guide to the Galaxy: the human urge for the creation of little green pieces of paper42. Likewise for rules and control.

Wellcome’s 2020 Research Culture report: ‘[research] is about to collapse’

Whilst editing these notes (15th January 2020), the Wellcome Trust published the results of a large survey43 about research culture, matching this view of ecosystem dysfunction at a cultural level. Wellcome director Jeremy Farrar said: “These results paint a shocking portrait of the research environment.” A subset of the reports findings:

‘78% of researchers think that high levels of competition have created unkind and aggressive conditions.’
‘Only 14% of researchers agree that current metrics have had a positive impact on research culture….’There was a prevalent idea that funding criteria were a core driver of research misconduct. Many argued that the nature of these criteria, which rewarded researchers who had published in higher-impact journals, encouraged negative research behaviours, such as deliberate embellishment or distortion of data.’
“Interviewees felt that funding was consistently granted to those with impactful research, and therefore teams that conducted rigorous research to exemplary ethical standards only to discover null data were often overlooked for future funding.”
Only 34% of respondents had received career advice or guidance from their supervisor, which is odd for a ‘training’ position(Figure 9). The same figure was true for mentors having ‘Connected you to others within or outside your field’.
43% believe that their workplace puts more value on metrics than on research quality [we suspect the notion of ‘quality’ itself has become degraded also].
‘69% of researchers think that rigour of results is considered an important research outcome by their workplace. However, one in five junior researchers and students (23%) have felt pressured by their supervisor to produce a particular result.’
‘Just over half of researchers (53%) have sought, or have wanted to seek, professional help for depression or anxiety.’
‘Many respondents described a system of patronage and power.’

The consequence, as a Russell group postdoc quoted in the report said “Really, I think [research culture is] about to collapse. Huge things need to change, otherwise they’re going to find everybody’s going to have left academia….”

Despite their ‘shocking’ nature, perhaps most shocking is that these cultural issues have been essentially completely absent from the priorities of scientific leadership: see for example the 2015 government review of science, titled ‘Ensuring a successful UK research endeavour’, which was written from the perspective of senior people and created more bureaucracy44. The entirety of the reports comments on scientific culture are in this footnote45. The report does raise important issues (see example46) but proposes no solutions (the creation of a new overarching bureaucracy is not a solution in and of itself and in view of many is the problem), and resorts to statements of aspiration devoid of any concrete proposal47. By contrast, deep problems with the current system such a lack of focus on young scientists have been highlighted in a bottom up way (for example, see Phillips/Phillips 2018, Telegraph).

Due to flaws in the Haldane principle48, designs around science funding are often made by and from the perspective of the very senior individuals sitting around tables in London. Such individuals have no direct incentive to address fundamental problems (since for them, all is well) (edit Feb 2020 - see new piece by Peter Thiel in ‘first things’49). There is an unhealthy power asymmetry in the research ecosystem and a lack of a corrective mechanism (such as public discontent)50. Because many Professors (in biology at least) often no longer do much research but rather manage laboratories, edit journals and serve on committees, societies such as the royal society increasingly reflect collections of formerly practicing scientists, rather than their original role as bubbling cauldrons of science. Their membership is an award for important and accomplished scientists, ‘Fellowship of the Royal Society has been described by The Guardian newspaper as “the equivalent of a lifetime achievement Oscar”. They therefore often unintentionally advocate as instruments of the status quo, rather than serving as an active forum for debating, facilitating and improving the collective scientific endeavour.

Recognising the contribution of distinguished scientists is of course good, but we additionally need a ‘royal society’ for practicing scientists ‘at the bench’, an elite national society for brilliant and promising minds to link up the nation’s innovators and disruptors through informal bottom-up conferences, workshops and generous long-term fellowships.

These factors combine to mean that consulting only the usual voices will likely unintentionally produce a very distorted sense of how to increase the budget. For an example of how money might be mis-spent, we can look at the effective doubling of the US NIH budget during late 1990s/early 2000s. This simply led to people who wouldn’t have been good enough to get funded being funded, a glut of extra postdocs and PhDs being trained, without a clear increase in the quality of the product. Indeed, the subsequent overtraining arguably contributed to the hypercompetitive atmosphere in academia (see Alberts et al 2014, ‘rescuing US biomedical research from its systemic flaws) (see footnote for example detail51). A doubling of the science budget is not at all guaranteed to meaningfully improve the system.

This suggests that a much more diverse range of opinions should be sought including those internationally and across career stages, with special emphasis on ‘outlier’ institutes and systems that have returned a disproportionate amount of innovation and discovery, but that are not present in the current system to a significant extent.

Part 2 - Common features of three of the most successful research laboratories in history

One explanation for this slowing is that we no longer have some kinds of scientific structure that we used to have. Government has announced an intention to create a UK ARPA, but there are other examples from that period.

These looked quite different to universities and are largely gone. We will now examine what are probably the three most significant laboratories of the twentieth century: Xerox PARC, the Cambridge Laboratory of Molecular Biology, and AT&T Bell Labs. Together they gave rise to the fundamentals of personal computing, molecular biology and communications technology, and created international networks of science with their founding institutes at the center.

Though it is tempting to view them as somehow magical curiosities, they were, as Bell Labs president Kelly said of his own lab, ‘No house[s] of magic’. Instead they followed a set of structural principles, from which great work emerged in a long-term and relatively unplanned and unforeseen manner (though they had many short-term successes also). Our goal is to examine them and work out how to replicate them.

What is most interesting to us is that when one examines these institutes, we find their culture and organisational principles share a number of striking structural commonalities at odds with how modern research is organised, and contrary to common assumptions of how research is best organised. Recent attempts to learn from these commonalities, in particular HHMI Janelia, have become highly successful.

Paper 2 examines whether it is likely they can be replicated effectively with similar order of magnitude results.

This section is structured in three overlapping parts:

The first part explores how, rather than a university department like ‘department of chemistry’, or of ‘computer science’, the institutes were instead focussed essentially upon what can broadly be termed visions.
The second part explores how researchers were selected and funded within this system.
The third part explores how power was allocated and interacted within the system (junior vs senior, for example), and how teams emerged.

Where points are ‘obvious’, our intent is to show how they were fundamental features of these laboratories, not some desirable add-on, and that they had myriad hidden benefits52.

Going forward it is important to note there were big differences these campuses, in terms of size, career path, duration etc. Bell, for example, appears to have had greater top-down guidance. But importantly, the top-down guidance was internal and so could be more intelligent (adaptable and using local control). If anything, that they varied along such dimensions and this and still had the commonalities above makes those commonalities more important, not less.

Note: none of these institutes were funded by competitive, for-profit companies: all were either government money or effective tech monopolies.

1. Fund scientific communities not scientific factories: “focussed ecosystems pursuing tech visions”

Learn lessons from three of the most transformative research institutes in history

Xerox Palo Alto Research Center (PARC) (early 1970s)

Xerox PARC: the creation of personal computing.

People view computers with an inevitability but it could have been much delayed had Xerox PARC not existed. Mitchell Waldrop (The Dream Machine (TDM)) highlights that in the 1950s Wall Street showed very limited interest in technology, just as technology was about to become of high importance. IBM itself totally missed the personal computing revolution even once all of its constituent parts were in place (p164, 187, TDM).

It was instead the visionary government funding agency ARPA that funded a network of university academics (including PhDs) that saw the promise of computing for the world beyond the lab (ARPA is one example that proves ‘visionary government funding’ is not always oxymoronic).

Xerox PARC was financially independent from the original ARPA but it grew from ARPA’s computing program directly and its culture largely came from it. Xerox PARC is widely credited (including by Gates/Jobs) for being a dramatic accelerator of personal computing. PARC was exclusively funded by Xerox’s effective monopoly in the area of photocopying.

Xerox PARC had a vision that built directly on JCR Licklider’s ARPA vision (and personnel): ‘Computers are destined to become interactive intellectual amplifiers for everyone in the world universally networked worldwide’ (Licklider).’ ARPA/PARC were effectively successors in the personal computing story, with researchers such as Kay extending the vision53.

From Wired54: “the 1970s and early 1980s, the essential cluster was at Xerox PARC—and its leader was Bob Taylor. Taylor's lab was of such high caliber that at the time Stanford professor Donald Knuth called it "the greatest by far team of computer scientists ever assembled in one organization." Taylor's Computer Science Laboratory, working with its sister Systems Science Laboratory, pioneered or perfected many of the innovations we associate with modern computing: the graphical user interface, icons, pop-up menus, cut-and-paste techniques, overlapping windows, bitmap displays, easy-to-use word processing programs, and Ethernet networking technologies, among others. When Steve Jobs famously visited Xerox PARC in 1979, these were the innovations he witnessed with awe and later incorporated into the Lisa and Macintosh computers.”

Xerox PARC won three Turing Awards for work in the early 1970s in large part guided by Licklider and later Kay’s vision (the Turing Award is the Nobel Prize equivalent in computer science) from a core research staff of barely 50 (It was smaller than the other institutes included here). More importantly than arbitrary prizes, it created personal computing, as noted above both Gates and Jobs credited their fundamental work. PARC invented the Xerox Alto, the first computer to resemble those we use today, with graphical interfaces. Many everyday features and underlying technologies of personal computing/modern offices were invented there, including object oriented programming, bitmap graphics, laser printing, what you see is what you get software, the ethernet and much more. It was, as Waldrop titles his book, a ‘dream machine’. It did most of this in approximately five years. After the mid-1970s the relationship with the Xerox bureaucracy broke down

How did PARC’s magic happen? It didn’t just happen to have the 50 best researchers in the world - Google almost certainly has many more brilliant people today than PARC had. As we shall see the key features emphasised by alumni are its organisation, structure and resulting culture. For an example of analysis of key features of creative research environments from someone inside xerox PARC, see this recent essay by Alan Kay.

A cautionary note about PARC - it came directly out of an existing program (ARPA’s IPTO) and it did not last. Alan Kay and others ascribe this to a breakdown in the relationship with Xerox after around 5 years ((Bob Taylor had a contract with Xerox in the early days to give it freedom from Xerox). Had this not happened, would PARC have continued at a same-order-of-magnitude rate, or would it have stagnated? Brett Victor + others including Kay note much of Licklider’s vision remains unfulfilled.

The first Cambridge MRC Laboratory for Molecular Biology (LMB) building

Cambridge MRC laboratory of molecular biology (1950s-70s): the foundations of molecular biology

For the MRC laboratory of Molecular Biology at Cambridge, the implicit vision was that the understanding of cellular life would come from understanding interactions between molecular structures, a near heresy at the time (its original name was the ‘MRC Unit for Research on the Molecular Structure of Biological Systems’).

James Watson, co-discoverer of the double helix, described it as “the most productive center for biology in the history of science” (Pennisi, 2003), with twelve Nobel Prizes amongst its staff as of 2019 (several more for work done by visitors at the LMB). By this flawed measure it is the most successful research institute in history.

As with Xerox PARC, LMB’s unique culture was critical:

“A British newspaper once described [the Cambridge LMB] as a Nobel factory. But Klug [Nobel Laureate] takes issue with that characterization: ‘It’s more like a plantation, where you plant the seed.’ The fertilizer came in many forms—money, equipment, collegiality, to name a few.” (Pennisi 2003)

It has delivered many great advances, such as the structure of DNA, the genetic code, messenger RNA, much of modern developmental biology (C. elegans), genetic sequencing technologies, several major advances in broader structural biology, the cause of sickle cell anaemia, monoclonal antibody technology (critical for much modern immunotherapy and basic research) and much more. The LMB laid the foundations for molecular biology that underlies modern medicine. It also made Cambridge (UK) a global hub of molecular biology, with its alumni leading the world of biotech even today (a very underappreciated point).

The LMB is now much bigger with a much broader focus.

We will see that it shared its fundamental structural features with Xerox PARC and Bell.

AT&T Bell Labs Murray Hill campus 1942

AT&T Bell Labs: ‘One policy, one system, universal service’

‘My first stop on any time-travel expedition would be Bell Labs in December 1947’

55Bill Gates (Gertner, p. 4)

For the industrial Bell Labs, the vision was its founder Theodore Vail’s philosophy of a communications system delivered as and providing ‘One policy, one system, universal service’ for communications (p.20, Gertner). They were originally focussed on telephones in the US but the technology created generalised and scaled dramatically.

The Bell vision of a connected world would be delivered through technological process - as Gertner writes in his classic history of Bell: ’Eventually it came to be assumed within the Bell System that there would never be a time when technological innovation would no longer be needed.’ (p18, Gertner). It was tech progress that was inherent to the betterment of the system as a whole. Their task? ‘Our job, essentially, is to devise and develop facilities which will enable two human beings anywhere in the world to talk to each other as clearly as if they were face to face and to do this economically as well as efficiently’ (Oliver Buckley, Bell Labs vice president (1930s),Gertner p45).

Through this vision, Bell Labs laid the foundations of communications technology across almost the entire 20th century. To restate: the core enabling technologies of modern communications technology almost entirely come from one laboratory. Nine Nobel Prizes and four Turing Awards were given for work done there, with many discoveries not winning due to there not being an equivalent prize in the field (such as Shannon). More strikingly, only one institute exists that is structured even slightly similarly (Janelia, where three of us have worked).

Fundamental inventions include vacuum tubes, transistors, core parts of the laser, UNIX programming, C programming, cellular phone technology, fiber optics, photovoltaic cells, information theory, radio astronomy… there are many more. Many of these advances made telephones and later the internet possible. But it was also blue skies - cosmic background radiation was found there, and a Nobel Prize was given for microscope innovations.

Like PARC and the LMB, Bell had a distinctive environment:

‘[Claude Shannon] would later tell an interviewer that the institution of Bell Labs (its intellectual environment, its people, its freedom, and, most importantly, the Bell system’s myriad technical challenges) deserved a fair amount of credit for his information theory [arguably the foundational theory of digital computing].’ P135 Gertner

It was a culture tweaked by management for creativity, not simply productivity56.

Gertner’s 2012 book ‘The Idea Factory’ is by far the largest evaluation of Bell Labs. There has been surprisingly little other analysis of what made it work. Note how tricky it is to find a diagram or explanation of a Bell Labs managerial structure. The lack of interest in the organisational structure is telling.

Bell Labs is particularly interesting because it did not suffer the relative decline thought to be inevitable for institutes, at least until it was killed by antitrust regulators. Bell lasted many decades and if anything it improved in the first few decades. It did not regress to the mean and did not become stagnant/bureaucratised. For this reason I focus more on Bell than on the other two institutes.

For a critique of Bell labs see the footnote57.

This paper highlights common features of these funding structures that are essentially absent from modern research. The common features between them and their difference from modern science bureaucracy is, in our opinion, a major under recognised point in current research policy. In essence, the UK can learn from Bell/PARC/LMB in ways that will positively transform our world.

An anti-bureaucracy silver bullet: institutes should be fully internally funded at an elite level

The foundational difference in funding between these institutes versus Universities is that unlike the vast majority of basic research institutes in the world today, these institutes were all fully internally funded. They largely could not even apply for outside grants though there were exceptions. The institute as a whole was supported, and researchers inside the system had no need (and were essentially forbidden from) to apply for external grants. Instead, resources were allocated internally. This point about internal funding underlies many if not all of the points which come below58.

Funding came from a single source, greatly simplified negotiations, reduced the chaos that arises from competing interests, and allowed decisiveness. Each extra funder would have come with new requirements and complexities. Instead: one source, one approach, independence, and simplicity.

Further, for two of the institutes their funding came from tech monopolies, which gave them greater funding freedom than is possible in competitive markets (see Thiel’s Zero to One, and also footnote59). Given that Britain lacks any tech monopolies, it is incumbent for the government to fill this void and create the kind of blue skies industrial research labs such as Bell Labs. They should be deliberately funding things that cannot be done without government money, in the style that monopolistic companies like AT&T were able to do.

This internal funding freed PARC, Bell and the LMB from the perils of relying on a distant, central bureaucracy and gave them independence to operate ‘outside the norm’. This is in marked contrast to how most University research is funded. Modern universities instead have a strong reliance on these external funding sources which are universally slow and suffer from common problems.

Vannevar Bush’s report that founded the modern US research endeavour

The current bureaucratic situation is very different to University research funding’s founding spirit. When President FDR asked for the war-time funding of science to be extended to peace time, he prescribed the following: “New frontiers of the mind are before us, and if they are pioneered with the same vision, boldness, and drive with which we have waged this war we can create a fuller and more fruitful employment and a fuller and more fruitful life.’60

Vannevar Bush

War time research in the US referred to by FDR came in large part from the Vannevar Bush’s World War 2 OSRD, which was created rapidly and outside existing channels (which led to push back, for quite amusing history see bold in 61). Strikingly, within 6 months, Bush’s war-time funding agency OSRD had awarded and delivered 126 research contracts. The entire OSRD system was up and running, from scratch, much faster than a US NIH allocation funding round (the main US biomedical grant), which recommends 8 months planning before even submitting the proposal, let alone it going through the peer review and the bureaucratic machine (see62 and 63):

A 2019 tweet on the pre-submission phase of the US NIH grant only, longer than the entire startup time of the original government research agency OSRD

Adopting full internal funding would protect research centers from critical problems in central management of science, and allow local control and decision making as is found in startups. In contrast, in Universities researchers are sometimes hired as much if not more for their ability to extract money from external central bureaucracy as they are for their ability to do truly groundbreaking science and innovation and/or help their scientific community.

Identify unique and long-term visions

The research visions uniting each institute allowed them to be a) known for something specific which in turn led them to be b) hubs and attractors for talent: they were ‘the place to be’.

Critically, these visions were not a goal for direct ‘impact’: this is completely the opposite of how these institutes worked. The institutes were not about direct ideas to change today: they were each aiming over 20-30 year timescales (though Bell/LMB/PARC often had big impact much, much faster - aim for the stars and you will quickly reach the moon!). Bell Labs had a thirty year timescale from invention through to application (Gertner p305). ‘The Vail strategy [founder of Bell Labs], in short, would measure the company’s progress “in decades instead of years”’, yet many advances came much sooner (p18, Gertner). Alan Kay’s part of the PARC revolution came by imagining what would be possible in computing in twenty to thirty years time, and beginning the journey to that summit. At the early LMB Cambridge, Sydney Brenner, seeing Crick and Watson’s double helix model before its publication, saw the future of a subject that was a long way from the real world (yet largely birthed the UK’s bioscience sector and the UK’s huge influence on the global molecular biology network). Few of the advances had obvious immediate, real-world application at the time they were planned and their impact was often highly unclear, though advances were often quickly made use of in unexpected ways once they had been made.

An underappreciated point for these visions was that they were relatively unique at the time - they were emerging fields of technology. They were not just researching ‘cancer’ or ‘climate change’ (but had influence on them64): They were doing something outside the mainstream, pushing ‘think different’ not incremental/horizontal development. The institutes therefore had near monopolies in their area of research at least at their foundations. This is in contrast to academic departments between different Universities.

Sydney Brenner with the early ‘heretics’ at the Cambridge LMB, including Crick and Perutz

With hindsight their focus seems obvious. But at the time much of what they were doing seemed crazy to the mainstream. Sydney Brenner, on the early days of the LMB and of molecular biology itself65 (Interview w/Elizabeth Dzeng, 2014):

‘What people don’t realise is that at the beginning, it was just a handful of people who saw the light, if I can put it that way. So it was like belonging to an evangelical sect, because there were so few of us, and all the others sort of thought that there was something wrong with us.
‘I remember when going to London to talk at meetings, people used to ask me what am I going to do in London, and I used to tell them I’m going to preach to the heathens
‘I think it would have been difficult to keep going without the strong support we had from the Medical Research Council66. I think they took a big gamble when they founded that little unit in the Cavendish [the predecessor to the actual physical institute built later in the 50s - this is a model to replicate]. I think all the early people they had were amazing. There were amazing personalities amongst them.

This is very different to how central bureaucracies usually fund institutes: these institutes are typically either very broad, or they are funding something which has already become mainstream (otherwise it cannot get past a committee of people - rare visions get voted down and regression to the mean occurs). But if you want to make a big difference there are advantages to not looking where everyone else is looking globally.

These long-term visions become talent attractors

‘[A] great vision acts like a magnetic field from the future that aligns all the little iron particle artists to point to “North” without having to see it... The pursuit of Art always sets off plans and goals, but plans and goals don’t always give rise to Art. If “visions not goals” opens the heavens, it is important to find artistic people to conceive the projects’
Alan Kay of Xerox PARC

Having unique, focussed visions allowed these institutes to become talent attractors, building a critical density of excellent people focussed on a problem. Of the early cambridge LMB:

‘The increasing international reputation of the group’s work brought talented young scientists to Cambridge, including physicist Francis Crick as a research student and biologist James Watson as a postdoc.’ (Bynum, 2012)

As Nobel Laureate Horvitz wrote of the cambridge LMB:

“In the budding field of molecular biology, “his [Perutz’s] operation became known as the place to be,” says John Sulston, who shared.. [the]... Nobel Prize in physiology or medicine with Brenner and H. Robert Horvitz. (Pennisi 2003)”

Or Brenner in the King’s review:

This [the early LMB] was not your usual university department, but a rather flamboyant and very exceptional group that was meant to get together. An important thing for us was that with the changes in America then, from the late fifties almost to the present day, there was an enormous stream of talent and American postdoctoral fellows that came to our lab to work with us. But the important thing was that they went back. Many of them [UK LMB members] are now leaders of American molecular biology, who are alumni of the old MRC.’67 Brenner, Dzeng interview

Form scientific ecosystems, not factories, by recognising the social nature of science

These laboratories became communities of scientists, not simply collections of people working separately. It allowed a “critical mass” of people to appear, a phrase Bell Labs management specifically used and recognised the need for (p 153, Gertner). Bell became more than the sum of its parts, with Bell Labs president Kelly describing Bell as a ‘living organism’ (p 151, Gertner) - ‘physical proximity was everything’.

Having a focussed research vision means almost every talk is relevant to everybody in the building. Spontaneous interactions were far more likely to lead to positive results than in a university department where everything was working on very different things (though university departments bring other benefits of greater diversity of research focus). These interactions were so critical to the Bell Labs mode that there was a rule that nobody could decline to help a colleague, and all office doors had to be kept open (only Claude Shannon broke this rule)(p151, Gertner). Bell Labs architecture was consciously designed with this in mind68. Similar rules existed at the LMB in its heyday - ‘Per Perutz’s order, there were no doors, no locked cabinets— no secrets among scientists.’69 (Pennisi 2003)

Brenner himself described the Cambridge LMB environment as ‘absolutely the most marvelous place to work in’ (Friedberg 2010, pg 117), and highlighted its social aspect. Comparing the LMB to a newer institute: ‘‘In the LMB, we were more tightly packed… It may be just a romantic sentiment, but I feel that when you are living in more of a warren, you may be more productive…. What’s terribly important in a lab is keeping it open and keeping the conversation going all the time.’ (Sydney Brenner in the 2013 HHMI winter bulletin). In his biography Friedberg describes Brenner routinely holding court on a Saturday morning, ‘talking for an hour or more with the young trainees’ and anybody who happened to pass through70,71. This social interaction is widely remembered as a hallmark of the LMB72. These environments teach without the need for courses, you get a real education by doing, by being, not by bureaucratic pedagogy.

This kind of freewheeling, cross-lab interaction is rarer in a University department, where the range of problems being worked on is vast, and other structural issues (such as being drowned in bureaucratic distractions, see below) can limit such workplace dynamics.

Recognition of community was also present at the very origins of ARPA visionary Licklider’s vision when at MIT. As Licklider’s wife Louise said: ‘Cambridge [US] was like an anthill. Everybody was getting involved with everybody else - finding different challenges, taking up different ideas. I use the word cross-fertilisation because there was an awful lot of that going on. And quite a lot of socializing too.’ (P66, The Dream Machine). These interactions are a hidden cost of, and so are killed by, hierarchy, paperwork and lack of free time. Consider that modern grants and bureaucracy often prohibits scientists using their grant money even to take invited guest speakers travelling from abroad to dinner.

2. Invest long-term in promising world class scientists instead of individual projects

How were individuals in these scientific ecosystems selected (broadly part 2) and self-organised (broadly part 3)?

Allow researchers to focus entirely on research and mentoring, not bureaucracy

The ability to focus on research without distraction, not only total hours of research, is critical to doing great scientific work but largely ignored by science central management. Feynman73, Newton74, Cajal (modern neuroscience founder)75 and many others have said this76. Turing Award winner Richard Hamming, from Bell Labs, advised in his classic ‘You and your research’:

``Knowledge and productivity are like compound interest.'' Given two people of approximately the same ability and one person who works ten percent more than the other, the latter will more than twice outproduce the former. The more you know, the more you learn; the more you learn, the more you can do; the more you can do, the more the opportunity - it is very much like compound interest. I don't want to give you a rate, but it is a very high rate. Given two people with exactly the same ability, the one person who manages day in and day out to get in one more hour of thinking will be tremendously more productive over a lifetime….. Most great scientists are completely committed to their problem. Those who don't become committed seldom produce outstanding, first-class work…..Keep your subconscious starved so it has to work on your problem, so you can sleep peacefully and get the answer in the morning, free.’

This advice is impossible to follow even loosely in any modern academic research environment77. In a publish or perish world, there is little time even to read other’s research, destroying scholarship, cross-fertilisation between fields and, critically, enjoyment. Creative work requires lots of ‘free time’.

The 2020 Wellcome trust report on research culture explored below found that more people agreed with the statement that “My institution/workplace’s expectations of me to undertake a number of roles leave me little time for research” than disagreed (44% vs 33%). We pay scientists to research but they do paperwork instead.

Maher and Anfres 2016, Nature - Researchers spend less than half of their time doing research and this is worsening.

A recent estimate is that researchers at all levels spend less than half of their time actually doing research, and that almost 40% felt this bureaucratic burden has worsened substantially just in the last 5 years (Maher and Anfres 2016, Nature). Given the fractionation of time and focus this entails, this likely substantially underestimates the harm (research output is nonlinearly related to time spent on it). Much of this distraction is inherent to the current University system and likely impossible to reduce without fundamental reform. This all occurs despite, or even because of, the huge increase in University bureaucrat number.

Much of this time wastage goes into the process of applying for and reviewing research grants, an inherent requirement of University. From a 2017 paper by some of our authors, slightly edited:

To illustrate how costly the growing bureaucratic burden upon scientists is, consider the grant writing process. Most grant proposals take months to prepare, and yet have very little chance of success. A study in the British Journal of Medicine (Herbert et al 2013) estimated that for a single round of applications, 550 years of researcher’s time was spent preparing proposals, with only a 25% success rate (which is higher than many grant success rates), equating to 412.5 years of wasted time. When multiplied across all funding applications, this is a truly extraordinary waste of our most highly trained minds’ time. Even if this is an upper estimate (though we suspect it is not), this waste is absurd… we have a system in which a medium sized lab, with approximately three to four projects, ends up writing very lengthy grants with a success rate of 20-30%. Thus, every 2-3 months (assuming 3 year projects), the lab head has to write an extensive grant application, with pilot data and numerous other requirements78. The incentive for real innovation in this system is low.

As a university vice president for research from the US recently remarked despairingly: ‘its amazing anybody gets anything done…’.

Individual universities themselves are also highly bureaucratic79. An everyday example of the absurdly slow and convoluted bureaucracy is that of a set of 5 LEDs, costing a total of under £10, has to go through a nine-step bureaucratic review process at a leading UK institute (see Phillips/Phillips 2018). In the UK, a neuroscientist friend recently had to wait three months after a grant was delivered and approved simply for the grant to go through UCL’s bureaucracy, three months during which he was therefore out of work (after several months of waiting for the grant to be approved). He spent more time waiting in total to get into the job than Bush’s War-time OSRD agency (above) took to be created and be fully functional80.

These problems are likely inherent to all large bureaucracies, but many bureaucracies think the issues are somehow solvable with more paperwork. When the new UK ARPA was announced there was predictable resistance to it being outside the UKRI bureaucracy (‘bizarre’ in the words of one prominent research policy researcher in New Scientist 81), but one of the founding points of ARPA was that it was a way of getting around bureaucracy, like Lockheed’s Skunk Works82. The early ARPA visionaries realised the following:

“Bureaucracy destroys initiative. There is little that bureaucrats hate more than innovation, especially innovation that produces better results than the old routines. Improvements always make those at the top of the heap look inept. Who enjoys appearing inept?” — Frank Herbert, Heretics of Dune, as quoted by Jake Wilder in this piece.

As Alan Kay says, you cannot have great ideas inside the beltway.

A large fraction of this time wastage is removed by relying solely on internal funding. Without this step, an institute is inevitably locked into the problems inherent to external bureaucracies. This is why HHMI’s research campus Janelia declined both external NIH money and affiliation with nearby universities: the hidden costs of bureaucracy greatly exceeded the $10’s of millions they stood to gain.

This is the type of waste of taxpayers money we should be paying attention to - not worrying about mis-spending a pound or two on some LEDs.

Invest in exceptional scientists rather than individual projects

‘There’s no exploration any more except in a very few places….. [the researcher] would need to do something that’s important to advance the aims of the people who fund science.’ Sydney Brenner in Dzeng Interview, Kings Review

Universities are largely funded by each researcher applying to various central funding agencies for funding of individual projects. This project funding system sounds like a fair thing to do, but it is fraught with inherent difficulties (see below).

Rather than micromanaging projects, the institutes examined here focussed on bringing the best in the world to one place united by a vision, then giving them great freedom once admitted. The rest would sort itself out83. They invested in people not projects.

Contrastingly, researcher salaries in the UK are notoriously internationally low which limits our ability to attract top talent - in particle physics “The current rate is ~£17k for a student in London, this compares to $39k for a student at a leading US university and 29k Euros at a leading German university.”84

As Pennisi writes of the LMB:

The lab’s recipe for success dates back to its early days, when leaders such as Max Perutz had the luck and insight to pick the best and the brightest (among them some quite unorthodox choices) and secure them almost unlimited support, both financial and collegial. [Similar quotations in the footnote85].

Gertner’s book about Bell Labs is replete with similar tales of the need for talent at Bell Labs, with Bell bringing the best in and paying them ‘for their imaginative abilities’ (p3 Gertner)86.

For a useful comment on why unorthodox choices work see footnote87.

When Bell Labs sensed opportunities in lasers, Kompfner, a deputy of Pierce, simply began travelling the world to bring the best to Bell:

“He went around the world at that time, in 1960, trying to find good people - that's all he wanted, good people88, and he would try to persuade them to switch their disciplines to take on what he called ‘laser and optical communications research’”. P256 Gertner

Even when projects failed, Bell Labs worked to keep the best scientists: ‘Bell Labs wasn’t going to fire us, they were going to tell us to find a job within Bell Labs.’ (p294, Gertner). This freed scientists to pursue risky work, knowing that their next project grant did not depend on definite success. In the University system, applying for a grant after a barren spell without successful papers is a largely futile endeavour, forcing collective low-risk, low-payoff research to ensure you get the next grant.

But these were no socialist utopias - surviving at Bell was highly competitive, it was simply that at Bell, like LMB/PARC, the individual as a whole was assessed, not the individual projects success - if you failed whilst pursuing a difficult target well, you would be kept on (see example in89). Ie, membership of the laboratory was the key cut off. This ‘people not projects’ is the direct opposite of the project-based University funding system.

These were ‘sink or swim’ institutes where you were solely kept on due to merit (‘Technically competent management, all the way up’). This was not publish or perish, as it wasn’t the short-term output that was used to assess people.

Distant central edict cannot control innovation: Give researchers great freedom to pursue ‘heretical’ ideas and realise that ‘impact’ actually means ‘well-received by current opinion’

"You did what they said could not be done, you created things that they could not see or imagine" From the final email of ARPA/PARC pioneer Bob Taylor, sent to his colleagues, quoted in wired.

A critical issue is that though we might want to fund research into curing dementia or promoting clean energy, we do not know how to because we haven’t discovered it, and so attempts to do so will often simply inflate existing views rather than promote exploration (see the Alzheimer’s example above).

Isaac Asimov writes in ‘How do people get new ideas?’ that “It is only afterward that a new idea seems reasonable. To begin with, it usually seems unreasonable. It seems the height of unreason to suppose the earth was round instead of flat, or that it moved instead of the sun, or that objects required a force to stop them when in motion, instead of a force to keep them moving, and so on.”

Gertner (p.184) cites a 1956 book, The Organisation Man by Whyte, which suggests why Bell Labs (and a similar industrial lab) had been so successful:

‘Of all corporation’s research groups these two have been the most outstandingly profitable…. Of all corporation research groups these two have consistently attracted the most brilliant men [sic]. Why?…. Of all corporation research groups these two are precisely the two that believe in ‘idle curiosity’90.

The three institutes trusted the instincts of their hand picked researchers.

As Bell Labs Unix pioneer Kernighan said: ‘‘I was never told what I should be working on in thirty years!’ (Nottingham interview). He highlighted especially that ‘Most people worked on something that was long-term, or at least whose immediate application was not obvious’. A reminder that Bell Labs largely built the fundamentals of modern communications technology, a higher impact contribution than anything UK R&D has produced for 50 years.

Even in Bell’s early days the first lab president Frank Jewett and architect of the first US east-west coast phone line Harold Harnold agreed on a need for what Gertner calls a certain ‘indistinctness about goals’ (p32, Gertner). Harold Arnold also said of scientific research: ‘Of its output, inventions are a valuable part, but invention is not to be scheduled nor coerced.’ As Ian Ross, who worked in the Bell Labs department doing transistor development in the 1950s, said ‘So often the original concept of what an innovation will do frequently turns out not to be the major impact.’(p250 Gertner). In the current government system, unpredictable discoveries and serendipity must be scheduled and have predicted impact or you won’t be funded!

Whilst modern scientific funding and publication systems force consensus, these institutes fostered crazy thinking that allows ground breaking discoveries by heading to different parts of the scientific unknown.

Impact was actually often consciously disregarded, even (especially?) in the genesis of the most fundamental and impactful advances, such as those of Claude Shannon91.

The LMB was similar, giving individual researchers the freedom to think differently92. An example comes from John Walker: this is what happened when he wanted to do the loonshot project at the LMB for which he would later win the Nobel Prize.

‘Walker never needed to write a proposal about this new research direction [which led to the Nobel Prize]. At the time, MRC relied on the lab chiefs to decide how to spend the money it allotted to the lab; they, in turn, trusted their colleagues to come up with good projects. Thus, Sanger merely asked a few questions before saying, “ ‘Why don’t you get on with it?’ ” Walker recalls….. At the outset, the project did not garner much support. “Quite a number said I was a fool and that I was going to wreck my career,” says Walker. Instead, he shared the 1997 Nobel for determining the structure of ATP synthase. He then went on to figure out how this key membrane protein works.’ (Pennisi, 2003)

Why is the focus on impact by central bureaucracies deadly? The essence of the problem is that, as Alexey Guzey paraphrased our argument, “by focusing on ‘impact’ we're actually focusing on ‘current impact’, but almost all high impact research starts from low impact research which grows and is being recognized over time”. Truly important work is rarely recognised at the time it is born, and almost by definition is not accepted by the peers of the time. By requiring that ‘impact’ be recognisable by the average peer today, we are killing most genuinely impactful work and instead selecting for the consensus view rather than funding work that leads to unpredicted--and therefore likely more valuable--discoveries.

Bahcall has a useful Francis Bacon quote in his book Loonshots: ‘As the births of living creatures are at first ill-shapen, so are all Innovations, which are the births of time.’ Bahcall highlights the fragility of nascent original ideas, which lack the protective clothing of more mainstream thoughts - his book uses a description of them as being ‘ugly babies’.

Bahcall’s loonshots has important analysis of how structural organisation alters the dynamics of innovation, an extremely under analysed point in research policy which usually assumes a ‘University focussed’ policy.

Bahcall summarises a related point: ‘the important breakthroughs come from loonshots, widely dismissed ideas whose champions are often dismissed as crazy.’ (Bahcall, page 2). When truly important things arise, they look different from what we expect (or we would already have found them), and thus alien, suffering rejection. Twitter, Instagram, Google, Facebook etc were recognised as great ideas only after their success.

Nobel Laureate William Kaelin warned in Nature in 2017 that we must ‘adopt more humility about predicting impact, which can truly be known only in retrospect’ highlighting that ‘transformative discoveries such as restriction enzymes, yeast cell-cycle mutants and CRISPR–Cas9 were once considered simply oddities of nature’, and said that the bureaucratic systems emphasizing impact over quality in both funding and publishing were leading to a literature full of ‘mansions of straw’ and ever longer training periods.

Consider this discovery of CRISPR-Cas9, perhaps the single most important practical and impactful scientific advance in thirty years. Jennifer Doudna, a Professor who was involved in the research team that did critical work, remembered thinking at the time of starting that project that ‘this is probably the most obscure thing I ever worked on’ and that she ‘certainly never anticipated would lead to something like this’93.

Even when such an innovation actually appears, their impact is usually not anticipated broadly. Who predicted the impact of electricity in the 1800s? Or take Bell Labs: An internal AT&T market research piece from 1971 which somewhat stalled mobile phone development at Bell Labs - ‘There was no market for mobile phones at any price’ (Gertner, 289)94. Or the development of the transistor, perhaps the most influential innovation of the 20th century, which was relatively unnoticed outside Bell Labs at the time (and its full use was not appreciated even there)(p104 Gertner). Fiber optics, ARPANet and the internet, all fundamental steps to the 21st century, were met with muted reception regarding importance or were thought impossible by the majority opinion (p258, Gertner, p417, Waldrop).

Brenner summarises:

In order to do science you have to have it supported. The supporters now, the bureaucrats of science, do not wish to take any risks. So in order to get it supported, they want to know from the start that it will work. This means you have to have preliminary information, which means that you are bound to follow the straight and narrow……...There’s no exploration any more except in a very few places….. he would need to do something that’s important to advance the aims of the people who fund science.

All of this should cause serious alarm bells about the entire UK government funding system now being predicated on predictable ‘impact’.

These points show that a research system in which every researcher must send their ideas for centralised approval by a committee is a very bad system for genuine innovation. As Thiel writes, ‘the best place to look for secrets is where no-one else is looking’.

There are ‘trillion dollar notes on the floor’95 - system-level dysfunction has led to large amounts of the profitable scientific unknown being not just unexplored, but unexplorable. The UK must have eyes for them.

Use internal evaluation to allow assessment of an individual’s ecosystem contribution, with external evaluation of the institute

So the institutes provided almost entirely internal funding, and focussed on selecting top talent who could pursue ‘crazy’ ideas (rather micromanaging what researchers could do according to prevailing trends of the time).

How did LMB/PARC/Bell hire and evaluate researchers96?

An important difference between University departments and these institutes was that they used internal assessment as opposed to external assessment of researchers/funding decisions; they did not even bring in external opinion to assess individual researchers. Today University money comes from outside grants, and these outside grants are externally assessed, so researchers are beholden to the trends of the time.

Assessments about renewal at the institutes were done by those familiar with the individual, their whole contribution to the ecosystem: they evaluated people, not paperwork. As Brenner said in an interview:

‘I think one of the big things we had in the old LMB, which I don’t think is the case now, was that we never let the [external] committee assess individuals. We never let them; the individuals were our responsibility. We asked them to review the work of the group as a whole. Because if they went down to individuals, they would say, this man is unproductive. He hasn’t published anything for the last five years. So you’ve got to have institutions that can not only allow this, but also protect the people that are engaged in very long term, and to the funders, extremely risky work.’ Dzeng interview in King’s review

Brenner’s view was that you have to review institutes as a whole, simply because evaluating the contribution of the parts from the outside is impossible to do effectively.

The same was known at Bell Labs. Unix Pioneer Brian Kernighan said in an interview with the University of Nottingham on the topic of evaluation/incentivisation of individual research, and how people stayed motivated despite not being told what to do:

‘The way it worked was, once a year you had to write down on a side of one sheet of paper what you had done in the year, and they used that to determine how much they would pay you next year…. In some sense It [managing individuals research] didn’t matter so long as this collection of people did things that were useful. And they had produced things over the years that were useful….. Given an environment like that where everybody is better than you, you don’t slack, you try to keep up with them.’

Both were systems where good emerged from collections of excellent people motivated by internal motivation not external pressure.

All of this comes down to the obvious but almost totally ignored point that individuals contribute in ways other than their own research papers (itself very difficult to assess). Current centralised individual project assessment can never capture this. Gertner (2012) said of Bell Labs:

‘... some lawyers in the patent department at Bell Labs decided to study whether there was an organizing principle that could explain why certain individuals at Bell Labs were more productive than others. They discerned only one common thread: Workers with the most patents often shared lunch or breakfast with a Bell Labs electrical engineer named Harry Nyquist. It wasn’t the case that Nyquist gave them specific ideas. Rather, as one scientist recalled, “he drew people out, got them thinking.” More than anything, Nyquist asked good questions.” Gertner p135.

Nyquist would himself make a great contribution to research (the Nyquist frequency), but even had he not, he was making a critical influence on the research ecosystem. Contributing to the health of the scientific society in such a way as Nyquist rather than publishing as many papers as you can is a very, very bad, even fatal, career move nowadays. An example from Oxbridge in the footnotes97. A recent (2019) paper in the new statesman highlights that this state of affairs would have been unthinkable 50 years ago98.

At the early LMB, this tolerance and use of local knowledge allowed ‘misfit’ and atypical characters like the 35-year old Francis Crick to remain in science despite no productivity99.

Evaluate the collective, not the parts. Let the collective self-organise.

Balance internal and external hiring to maintain culture and reward contributions (do not ban internal hires)

An important corollary to collective evaluation is internal promotion. In modern Universities internal hiring is frowned on - you are meant to hire people you know only as paperwork and a brief talk. This is a mistake: ecosystems need a balance of internal/external promotion/recruitment to a) maintain culture and reward internal contributions but also b) get outside ideas and prevent stagnation of a club-like atmosphere.

Without internal promotion, it is always in one’s interest to have an eye on the door and be beholden to the criteria imposed externally, not internally. If you cannot grow in an ecosystem, one has little ‘stake’ in the community and ultimately no reason to care about it beyond ethics. If merit does not let you grow in a research environment, you have no incentive in the collective. The institute is nothing but a vehicle to your own personal success.

PARC/Bell internally promoted extensively100. Every Bell Labs president came from within Bell Labs, as did most/perhaps all of their managers, and they hired straight from graduate school and promoted through-out scientists careers101. Brenner, Crick etc came from the early MRC unit at Cavendish that grew into the LMB proper. PARC famously formed from the graduate alumni of Licklider’s ARPA IPTO. Trinity College Cambridge is similarly renowned for nurturing of talent and internal promotion (famously, Newton)102.

A lack of internal hiring also strongly reduces the motivation to create the next generation of researchers: as Kay wrote of ARPA/PARC in a reply to a Telegraph article co-written by the author: “good results included ‘good people’ (i.e. long range funding should also create the next generations of researchers). In fact, virtually all of the researchers at Xerox PARC had their degrees funded by ARPA, they were “research results” who were able to get better research results.”

Though it's good to get a range of experience in different places, lack of internal promotion or makes it hard to build self-organising teams, and prevents maintenance of culture. The current culture of endlessly moving between institutes in one’s youth prevents a stable, collaborative atmosphere arising amongst scientists and promotes selfishness.

Facilitate research via strong core support, technical assistance and resources

A simple point to bring out that could go anywhere in this paper - these institutes achieved economies of scale by sharing resources between labs, and funded people at a high level. This allowed small labs to act as big labs, getting around a major problem with science in Universities in that small labs cannot afford top equipment. We expand on this in a footnote103.

The relative focus of the labs made this sharing of resources possible, as they were using similar tools and techniques across groups.

Consequently and relatedly, the institutes were also all involved in methods development, recognising the multiplicative benefits of new technologies. We cannot be certain but suspect this is due to them having a focus on one area, having many groups focussed on a topic allowed them to see the ‘arc’ of the subject and overview where new techniques were needed and cost-share/collaborate to make it happen. Methods development in University departments, especially with government funding, is very difficult to organise, due to lack of core support and due to funding agencies not funding it (the NIH will essentially not fund pure methods research). This suggests a niche as methods are highly influential.

A point picked up in paper 2 is that as science has become more method intensive, it has responded by lengthening career paths with scientists expected to be experts in all areas. This is an area where basic science should learn from industry and allow technical support. See, for example, ‘the case for scientific consulting in neuroscience’ by MIT-based researcher Jakob Voigts.

A brief summary

In the first two sections of this part of the paper, it has been argued that Bell/PARC/LMB labs differed from Universities in two critical ways:

They were fully internally funded ecosystems focussed upon a research vision, which allowed critical mass to build.
They had minimal bureaucracy and management, instead simply focussing on selecting the best researchers available and letting them pursue their own interests (people not projects). This is in contrast to the University system, where researchers must apply individually for each project funded, and must appeal to collective approval for funding by peers, driving conformity, risk-aversion and herding.

To close out this section, imagine if Bell Labs researchers had to apply to corporate HQ for each crazy yet revolutionary idea they had? It wouldn’t have happened, and that's modern academia. Instead, we should allow researchers to be exceptional autonomous decision makers.

3. Organise as self-organising start-up style small teams, not hierarchical bureaucracies

The final thematic difference: the relative weighting of power in the hands of junior researchers vs senior supervisors differed strongly at Bell/LMB/PARC (pro-junior researchers) compared to the University system (pro-senior supervisors). Rather than existing in large, fixed, tenured lab structures, the organisations were instead self-organising, composed of flexible small groups without entrenched top-down leadership.

Age balance in research societies: balancing youthful ignorance and senior wisdom

The great disruptive discoveries in history come predominantly either from people under the age of 40 or from outsiders to the field104. In contrast to its lack of acknowledgement in science funding establishments, the view that scientist’s peak in the first half of their career has been clearly expressed by many of history’s greatest scientists including Einstein, Shannon, Watson, Newton, Brenner, Cajal etc105. One does not need to agree completely with this statement to accept that we need to consider the balance of power across age in research. We at least need balance between stages.

An interesting mini-study from the start-up world by Brian Timar 106. He took 33 major companies (Microsoft, Apple, Google, Walmart, Tesla etc) and plotted the age of the founders:

The median and mean age of founders of the 33 most important companies(arbitrarily chosen) were both around 30.

In contrast to a cult of youth and start-ups in silicon valley, the power structure in academia is against newcomers/outsiders/disruptors and in favour of existing ‘expertise’107. The Professorial system is inherently hierarchical and based on people being ‘experts’ in the field, which has benefits for teaching108. Yet for innovative research this knowledge, and also age itself, may not be a clear advantage.

To have an innovative society, you need a strong bottom up force of a) new ideas and b) empowered eccentric thinkers who question the mainstream view and can act on this questioning.

Even putting aside other changes with age109, there is tremendous value in ignorance in research110:

“I strongly believe that the only way to encourage innovation is to give it to the young. The young have a great advantage in that they are ignorant. Because I think ignorance in science is very important. If you’re like me and you know too much you can’t try new things. I always work in fields of which I’m totally ignorant.”
Sydney Brenner, pioneer of molecular biology

The eyes of children can see things that adults cannot.

Those who have built the current view of science are also unlikely to tear down their own work, as Santiago Ramon Y Cajal wrote:

‘Two phases may often be noted in the careers of learned investigators. First there is the productive time devoted to the elimination of past errors and the illumination of new data, and it is followed by the mature or intellectual phase (which does not necessarily coincide with old age) when scientific productivity declines and the hypotheses incubated during youth are defended with paternal affection from the attacks of newcomers. Throughout history, no great man has shunned titles or failed to extol his right to glory before the new generation. Rousseau’s bitter quote is sad but true: “There has never been a wise man who hasn’t failed to prefer the lie invented by himself to the truth discovered by someone else.’”111
Ramon Y Cajal, Nobel Laureate, viewed as founder of modern neuroscience. (see footnote for a similar quote from James Watson - ‘‘more bricks into an edifice that already has enough rooms’112).

This system blocks the Newtons and Einsteins of the world from breaking through against the prior paradigm113. A recent example from a talk by Geoffrey Hinton “‘a CVPR reviewer reject..[ed] Yann LeCun’s paper even though it beat… the state of the art. The reviewer said it…[told] us nothing114 about computer vision because everything … [was] learned.”115 (the subsequent AI advances came in learning).

This suggests that we should be mindful of the ages and seniority of people who hold power in a scientific system, ensuring that the young maintain an ability to be independent and disrupt. Given it would run contrary to established power, it may be an area where ‘the Haldane principle’ cannot apply (that is, government must intervene over senior scientist’s wishes). Rather, this requires deliberate system intervention against the interests of the powerful, like antitrust regulation.

Relatedly, recent increases in the cost of techniques and methods have meant that the powerful can prop up their pet hypothesis with technical power whilst the young cannot afford to compete, relegating new PIs to third tier journals or to engaging in unethical practices.

Flatten the hierarchy: positive feedback in academic power and “The slavery of graduate students”

Rather than promoting dissenting views, quite the opposite is actually occurring. There is consistent long-term drift in the direction of power toward the more senior in bureaucracies, as rules are by definition made by those in power and the young face pressure to be sycophantic to the old. There has been essentially no push back against this in academia.

The ratio of senior/junior power is now very imbalanced in favour of seniority, and worsening, which alone justifies creating a parallel system. Risk-averse funding agencies generally fund prestige not promise which favours seniors, creating a strong positive feedback loop on prior success over time (see Yudkowsky 116). Another related issue is that even if you find a brilliant young lab head to work for, they will not be able to give you the glitzy letter of recommendation you need or the oil to get the paper past the custodians of Nature as needed for promotion.

Young scientists now exist as the bottom layer in a virtual pyramid scheme, ‘training’ for permanent positions that barely exist (Ben Barres, Stanford, write that perhaps 10% of postdocs get lab head jobs in the US, which matches similar estimates117). Sydney Brenner highlights this:

'Today the Americans have developed a new culture in science based on the slavery of graduate students. Now graduate students of American institutions are afraid. He just performs. He’s got to perform. The post-doc is an indentured labourer. We now have labs that don’t work in the same way as the early labs where people were independent, where they could have their own ideas and could pursue them.’ Sydney Brenner, Dzeng interview in Kings Review 2014

This culture has spread to the UK. Young researchers spend their twenties and thirties in ‘training’, working on necessarily short-term projects, which must yield publications to get career advancement, and then the credit for this work is largely taken by senior individuals on the paper (as Dawkins notes in the introduction to the selfish gene, many senior reputations were built on the work of their ‘trainees’ even in the 1970s/1980s118. Dawkins was describing the current situation more closely than he could have known, which has severely worsened since he wrote these words119.) Recent Nobel prizes could be used to illustrate this….

Credit has shifted from lead researchers to the supervisory professors. These professors take the large share of the credit despite in many cases having almost no substantial impact on the research, and this is now reflected in the everyday language of science120. In Wellcome’s new (2020) survey of research conditions, fewer than half of researchers reported having received ‘expert advice’ from their supervisor in the past year (Figure 9). A nobel prize was recently given to authors who had done little but supervise a project, with the critical insight coming from a theorist not even on the author list121. Today, surely we would credit the DNA double helical structure to the Perutz/Kendrew group at the Cavendish, rather than Watson & Crick 1953122? Credit assignment, a crucial component of adaptation in a complex system, is broken in (at least biomedical) academia123. This strongly favours the established powers who can take credit for others work (see footnote for examples of academic culture reported in Nature, where postdocs cannot even take their own work with them when they found their own independent lab124). A suggestion from a reader is that much of the culture problems in academia arise from this125.

This graduate student to postdoc ‘training’ period, in which people work at depressed salaries under control of seniority, has progressively lengthened126. In this time the trainees are totally dependent on these senior figures for references, and quitting a project midway would sabotage their own research and career. As Brenner notes, this leads to a generally demoralised system of young scientists127:

‘I now spend quite a lot of my time trying to help the younger people in science to enjoy it and not to feel that they are part of some gigantic machine, which a lot of people feel today.’ Brenner, Dzeng interview

A reader commented: lab heads “...can [also] prevent you from graduating. It creates the perverse incentive of preventing the best students graduating, while helping the worst students graduate faster” It is in the direct interest of a mentor to be a poor mentor, keeping the students as cheap labour.

These issues are entirely absent from government reports (see the 2015 UK Government report128 for an example), and the continuing patronising culture of viewing almost anyone under the age of 40 as ‘early career’ or a ‘trainee’ continue.

At least in biology, we have an increasingly disembodied academy in which the real work, ideas and innovation increasingly come from a relatively subjugated group, whilst generally the more senior people take the credit for it - a vicarious science. This is not universally the case but it is a disturbing trend. The very idea that the young might be the real drivers of science is heresy and almost culturally forbidden.

True research independence now typically comes in the late thirties at the earliest. It is challenging to perform truly independent work in another’s lab, as you will lack technical support129, and modern ‘independent fellowships’ in practice do not provide independence due to a) low research funds and b) the need to exist in within the available resources of a parent lab130. Once one has the professorship, one’s expected task is to begin building an empire.... But even once a professorship is obtained one is not independent. Since 1980 the mean age at which your first independent RO1 grant has been achieved has increased by 10 years from 35 to 45, see figures below. Tenure itself further contributes to lower disruption by fixing the employment structure for decades and more131.

Figure 1 of Levit and Levit 2017, PNAS, showing age of first RO1 (the main US NIH grant), the most common marker of achieving an independent, funded lab, averaging at almost 45 years old:

Figure from Maher and Anfres 2016, Nature, highlight a lack of success for younger researchers (with less track record) in the UK funding system as well:

These academic ‘empires’ consist of large numbers of ‘trainees’ working under a single supervisor. Brenner described these large group meetings, ‘which seemed to be the bane of American life’ as ‘the head of the lab trying to find out what’s going on in his lab.’ (Dzeng interview). As a reader commented - should we question the very notion of a lab?

The critical issue here is that the very people who benefit from it are those in a position of power, yet there is no mechanism for the young scientists to rebel against the system. (ultimately, they themselves must want to become one of the senior people as it is the only way to proceed in a pyramid scheme…). To use an analogy with silicon valley, there is no way for them to ‘quit college, get venture capital and start a basic science lab’. Further, funding ‘established’ scientists is far less ‘risky’ which plays into the preferences of bureaucracy..

Power in the research system thus progressively becomes more and more top-down, more and more concentrated, and working more and more in favour of the past not the future. The ‘let's give power to the young’ voice is powerless.

Remaining dynamic: do not give tenure to researchers

The career dynamics at Bell/PARC/LMB were very different to Universities. They almost entirely lacked the foundation of the academic career path: Tenure132. Tenure provides a ‘job for life’ in academia and sharply divides the scientific ecosystem (and power) into tenured and non-tenured (effectively, trainees). Nowadays tenure is seen as fundamental to the research career: it may be good for teaching but it is hard to see a justification for it in research133. Though not technically present in the UK system anymore it largely effectively is.

Rather than making an all-or-nothing decision on a researcher in the short hiring/tenure period, appointments at Bell/PARC/LMB were made on a renewable basis. This obviously allows much greater risks to be taken in the hiring process, and ensures that people stay positive contributors and not turn into dead wood.

The lack of tenure in turn allowed the full internal funding, because you were assessing at regular intervals. Being at the institute was equivalent to being fully funded, greatly simplifying assessment procedures.

As Andrew Gelman (Professor, Columbia, Ex-Bell Labs) wrote in Physics World 2012:‘….. At Bell Labs it was harder to be deadwood.’134 At Bell, researchers who were not renewed had no problem finding jobs in the University system teaching, and older researchers such as Shannon and John Pierce took up University positions later in their careers135. Brilliant young researchers were able to focus solely on maximising research during their most productive age, and then would use their experience to teach later on at Universities. Nowadays the opposite system exists: In universities, it is often the junior professors who do the bulk of the teaching, with more senior people with large grants being able to buy their way out of teaching entirely (by using their large grants to cover their teaching salary).

As opposed to Tenure, combined contract renewal with elite research funding on a long (15 year?) timescale would provide a form of career security to explore the unknown, and at the same time prevent complacency and deadwood. Accountability, scholarship and innovative disruption could co-exist.

Credit assignment & ‘pianists without fingers’: we need research supervisors

‘Nowadays, most of the people who say they’re in science aren’t really in science. They’re in something else. They’re in the management of science.’ Sydney Brenner, at the ‘Grand Challenges for science in the 21st century’ meeting in 2016.

‘These very people go about inflated and pompous, clothed and adorned not with their own labour but with those of others.’ Leonardo Da Vinci, Codex Atlanticus 117rb/323r

Managers, or ‘supervisors’, played a critical role in these institutes. At the LMB supervisors did their own research, even after winning Nobel Prizes136.

Rather than build their own empires, managers in the Xerox/Bell systems literally managed. They were coordinators, messengers linking hubs, getting the big picture across the institute and interfacing with the outside. A useful piece from Gertner (p102, Gernter): at Bell Labs ‘The supervisor was authorized to guide, not interfere with, the people he (or she) managed. “The management style was, and remained for many years, to use the lightest touch and absolutely never to compete with underlings137,”. This was strict, when Shockley [a Nobel laureate] violated this, he was ‘never forgiven.’138

The emphasis on researchers doing research was such that managerial roles were sometimes not sought after at Bell, as people preferred to remain in research. Bell labs president Meryvn Kelly declined the job of chief science adviser to multiple US presidents, and only took on the Bell Labs presidency once being assured he could also keep an involvement in research.

The role of people ‘high’ in the hierarchy in the Bell system was very different to University professors. These individuals served crucial roles in sharing and distributing the discoveries and technologies invented by Bell. When CCDs, now a fundamental technology in digital photography, were invented, Bill Baker took the discovery directly to the US NRO as he realised their potential use in satellite technology (p261, Gertner). This was facilitated by the staff all being technically competent - Fisk’s deputy, Julius Molnar, was famous for knowing more about the phone network and relevant engineering than’ anyone else alive’ (p261 Gertner).

They also provided the credit to those who did the research, not those who managed it139. Credit assignment mattered. It is now common for research reports from major funders not to even mention the names of those who lead the research, instead giving all named credit to the PI140.

There was also a recognition that not all top scientists should be managers or lab heads141 ((Shockley, Shannon etc were researchers who could not manage), and vice versa (Bob Taylor at PARC had no PhD - a ‘concert pianist without fingers’))

[Edit Feb 2020 - Larry Testler of xeroc PARC fame, who invented ‘cut, copy and paste’, said in 2012 of silicon valley: "There's almost a rite of passage - after you've made some money, you don't just retire, you spend your time funding other companies…..There's a very strong element of excitement, of being able to share what you've learned with the next generation." From the BBC.

Be people number limited: prevent empire building and promote cross-disciplinary collaboration via enforcement of small team sizes

Alan Kay, quoted in Wired: "When a really difficult thing is being worked on and you get synergy from the small team in just the right way, you can't describe it. It's like love." says Alan Kay, a visionary computer scientist who worked with Taylor at Xerox PARC before moving on to Apple and Atari. "You're trying to nurture this thing that is not alive into being alive."

There is increasing evidence to support the long held notion that science does not scale linearly. Taking a group of three people and tripling its size can lead to sharply sub or supralinear results. There are types of science that need very large teams, but this tends to be large scale data collection (such as the Allen institute) more posing engineering challenges as opposed to scientific challenges. Both are needed, but arguably this large scale science is something larger economies are better placed to provide.

One of the difficulties of academic structure is it prevents assembly of lasting teams, as you are either a permanent lab bead or a temporary trainee.. It is also very difficult to build a culture if the majority of your research workforce is only there for 3-4 years and overlaps minimally with others.

Small, ‘start-up’ style groups appear to be more effective for disruption (new ideas outside the norm) whilst larger groups are better for developing existing work. See Wu, Wang & Evans 2019 Nature. Wu et al find that large ‘development’ groups are relatively overfunded and small ‘disruptive’ are underfunded (see Wu et al paper), yet there is little to no consideration of such sociology of science in the world of research policy (unlike in the business world). All that counts currently is total paper number and citations. The only way to accumulate the resources to do edge-of-the-art science now is to grow a big empire - we don’t fund small groups to an elite level. This is a critical niche that the UK should be trying to fill - size restricted groups of elite people.

It is now extremely difficult to compete as a small group in technology-intensive academic research. Labs instead grow larger and larger, accumulating ‘empires’, with lab heads often having little involvement in the research itself142. It is not uncommon for labs to have forty people in them in the US, with trainee numbers far exceeding the number of available lab head positions (part of the ‘ponzi/pyramid scheme’ accusation in modern academia). All incentives in the current University system align with the individual accumulating power and against forming teams across expertise/disciplines/seniorities.

Instead of this pyramid structure, at Bell/PARC/LMB enforcement of small team sizes ensured these institutes were heavily modular, not monolithic with empires and emperors. Groups at Bell Labs were a lab head plus 1 or at most 2 people, who were support technicians (Gertner, 2012). At the LMB they maxxed at 6 (Rubin 2006). Xerox PARC did not have labs (Waldrop, ‘the dream machine’). The presence of shared core resources (above) meant this did not limit researchers. Instead, it let spontaneous collaborations across labs with different expertise occur, facilitating new ideas by mixing of ideas143. Large groups formed by self-organizing of small ones.

Chapter seven of Gertner’s book highlights that Bell Labs appreciated that 20th century innovation was complex, with Bell Labs Nobel Laureate Shockley saying “Things are much more complex than they were probably when Mendel [early geneticist] was breeding peas” As Gertner adds, ‘An effective solid-state group, for example, required researchers with material processing skills, chemical skills, electrical measurement skills, theoretical physics skills, and so forth. It was exceedingly unlikely to find all those talents in a single person.” (p134). This is very difficult to achieve in a University laboratory, where almost everybody is a ‘trainee’ of the single lab head - this was recognised in the mid twentieth century but is even more difficult now and this is not recognised in science policy.

To date academia has dealt with the increasing research complexity by a) greatly lengthening postdoc training periods and the number of trainees (whilst also culturally banning discussions of careers outside the academy), b) concentrating power in the hands of a few big labs which internally function as pyramids and c) turning graduate students into technicians (and increasing their number).

Instead, having small, collaborative labs, focussed on a vision, with strong shared resources and technical support is likely a much more sustainable, creative and effective model for elite research than ‘training’ ever larger numbers of people on depressed wages. (This dynamic small groups structure intriguingly resembles the chinese Haier company case study featured in the Harvard Business Review (Hamel and Zanini 2018, HBR) that is worth reading with respect to organisational structure144).

In general the lack of credit for teams as opposed to individuals is a cultural shift we need to make. Could a philanthropist fund ARPA/PARC computing pioneer Bob Taylor’s wish: “Taylor had long pushed for an award to honor group creativity, explaining that he did not think most innovations could be traced to a single individual.”145?

Empower promising young researchers

We put this point last so that it sticks in the mind.

In contrast to university hierarchies, there was both great trust and power put in the hands of the young, who were given responsibility decades ahead of what they could do today (we prioritise age and ‘experience’ over merit of ideas, energy and talent). Today those that pioneered the 20th century’s revolutions would be working in a senior person’s lab as ‘trainees’ gaining ever more ‘experience’ and doing the work their bosses win prizes for146.

When Bell Labs wanted to set up a new research area to create the amplifiers that would underlie early 20th century technology, rather than bring in a seasoned researcher they went to leading US universities:

[first, a quote from Bell Labs president Jewett]: “‘Let us have one or two, or even three, of the best of the young men who are taking their doctorates with you and are intimately familiar with your field. Let us take them into our laboratory in New York and assign them to the sole task of developing a telephone repeater.’ [now Gertner] and so came about the Vacuum tube, through two years of trial and error by Arnold, improving on an earlier design bought by Bell from De Forest (from 1907).” (Gertner 22). [as shown above the researchers would be given essential free reign on their problem - but note the top-down assignment of broad topic - the presence of top-down direction at Bell is interesting and is in paper 2]..

When projects composed of groups of people were needed, they let young people take the lead. Jim Fisk was charged during War time, at age thirty, with ‘perhaps the most important scientific project in the United states’ (Gertner, 69).

This imbued an attitude of greater risk taking. One of the pioneers of cellular (mobile) phone technology at Bell, Joe Engel, said of the mobile effort: ‘you have to understand, we were all very young, we were unscarred by failure. So we always knew it was going to work”. (p289).

At Xerox PARC this was more dramatic: there were no senior researchers. At a talk at HHMI Janelia in September 2018, in response to a question from this author Alan Kay said “I was the oldest researcher at Xerox PARC. I was thirty.” Instead of senior researchers, a brilliant manager, Taylor, kept the group harmoniously productive.

There was a striking lack of hierarchy/seniority at these institutes relative to other institutes (the Bell Labs president would deliberately dine with support staff to keep a broad view of the laboratory).

ARPA program director hiring practices - junior, internal hires

An interesting example of this from ARPA and thus PARC: In contrast to tenure-style slow turnover at the top, ARPA had fast turnover in the senior positions and hired relatively young.

ARPA IPTO (its revolutionary computer tech office in 60s) famously supported outside the box hiring. Most conventional was Licklider. Licklider, ‘the father of it all’, was the first director of the information processing techniques office at ARPA, appointed by the 3rd ARPA director Jack Ruina (1962, Ruina was then aged 39). As Encylopedia Britannia said, Ruina had ‘recognized that the problem of command, control, and communication of the nation’s military forces was one that computer technology might affect……..As head of IPTO from 1962 to 1964, Licklider initiated three of the most important developments in information technology: the creation of computer science departments at several major universities, time-sharing, and networking.”

Licklider did more than almost any manager in history, yet Licklider’s term was brief. “The tradition was for ARPA program directors to step after two to three years.. to make way for people with new ideas” (p254, TDM147) So Licklider joined the research division of IBM. How to replace him? “Fortunately, one of ARPA’s traditions was that outgoing office directors were allowed to recruit their own successors….. Because of the great freedom and power ARPA program directors had…. “it was critical to get the right person in place, or his whole movement could collapse.” (p255, TDM).

Licklider’s next move was very different to modern science: Waldrop says “all lick had to do with was to come up with an experienced, senior person with good credentials and everybody would be happy. So Lick characteristically nominated a totally inexperienced and very junior candidate whose credentials were about the most awkward and inconvenient conceivable in the Pentagon.”. Curiously and tellingly, at that time senior people often did not actually want such managerial jobs148.

That successor was 26 year old Ivan Sutherland, who created the pioneering Sketchpad graphical user interface for his graduate thesis work, which Alan Kay described as ‘probably the most significant thesis work ever done’149. Relatively few saw the significance of Sutherland’s Sketchpad at the time, and so had Licklider not been able to to pick to pick his successor, Sutherland would not have got into position to continue the vision: as Brenner frequently highlighted, large (>3 people) committees cannot make such exceptional decisions because of averaging between the committee members some of whom will not be bold.

Promoted from being Sutherland’s deputy, Bob Taylor was the next ARPA IPTO director (aged around thirty five) - and he didn’t even have a PhD. Having overseen the birth of the early-internet ARPANET at ARPA, he went on to be a kind of research concierge as associate manager at Xerox PARC150 (ironically not officially as director, because he didn’t have a PhD). When he passed away in 2017, Wired wrote ‘Last week the world lost the most important tech pioneer whom hardly anyone has heard of: Bob Taylor.” Taylor’s role was so influential it was described as ‘magic’ by former colleagues, yet he ‘never meddled’ (Alan Kay).

The early ARPA’s combination of fast turnover of senior leadership with allowing of appointing own succession is a very interesting way to have both absolute power in directorships and prevent empire building with term limits (a presidential style of science), whilst keeping a stable/coherent vision between the directors. Today’s trend of appointing directors via committee decision of course obviates any succession of vision.

The overall heads of ARPA were themselves young (not just programs, this is the entire DARPA.) Ages of the first eight ARPA directors at appointment (later ones DOBs not all online) averaged 42 years at age of appointment with a 3.25 year average term151:

1958-9: Roy Johnson, born 1906, appointed aged 52, 2 year term
1960-61: Austin W Betts, born 1912152, appointed aged 48, 2 year term. See footnote for an excellent interview153
1961-63: Jack Ruina, born 1923 ,appointed aged 38, 3 year term
1963-1965: Robert Sproull,born 1918, appointed aged 45, 3 year term
1965-1967: Charles M. Herzfeld, born 1925, appointed aged 40, 3 year term
1967-70: Eberhardt “Ed” Rechtin, born 1926, appointed aged 41, 4 year term
1970-75, Steve Lukasik, born 1931, aged 39 at appointment, 5 year term.
1975-77, George H. Heilmeier, born 1936, aged 39 at appointment, 3 year term

(Time of ARPA directors tenure can be found here154):

Compare this with the original 2017 UKRI leadership shortlist155.

A comment on this piece was “... there is a serious problem with incumbents hoarding opportunity. Just make them compete with their grad students and postdocs. Make them get back into the lab. Make them do the analysis. That would sort things out pretty quickly.”

Rather, at ARPA/PARC, occupying and moving between multiple leadership positions was consciously frowned on (instead, ‘Train your successor quickly then get back to work’ as Alan Kay describes it). At LMB/Bell leadership was either deliberately hands off or reluctant.

In the next paper we explore how to put this into action and where the pitfalls are..

Some further resources

Here are a subset of resources potentially hard to find with a simple google search or not in footnotes/main text links:

Jim Austin 2013, Science, ‘Give Science Some Slack’

Sydney Brenner 2014, Science - Frederick Sanger (1918-2013)

Nicolas Colin 2016 - Medium - A brief history of venture capital

https://salon.thefamily.co/a-brief-history-of-the-world-of-venture-capital-65a8610e7dc2

Athene Donald, 2017 - The Guardian - All eyes are on Sir Mark Walport, the new supremo of UK science

Elizabeth Dzeng, 2014, Kings review - How Academia and Publishing are Destroying Scientific Innovation: A Conversation with Sydney Brenner. Available at:

https://orion-society.com/2020/03/26/resource-2014-kings-review-e-dzengs-interview-with-sydney-brenner-how-academia-and-publishing-are-destroying-scientific-innovation-a-conversation-with-sydney-brenner/

An interview which captures Sydney Brenner’s views on the state of innovation. Brenner was a rare voice who, having been perhaps the most important biologist of the 20th century, dedicated the second half of his life to fighting for and defending real scientists from the rise of the ‘managers of science’.

Nesta - Sept 2019 - Greg Falconer - Fuelling the future of UK innovation

Friedberg, 2010 ‘Sydney Brenner: a biography - The only biography of Sydney Brenner.

Richard A.L. Jones 2019 - - A Resurgence of the Regions: rebuilding innovation capacity across the whole UK.

Nielsen and Collison 2018 - The Atlantic - Is Science Stagnant?

Elizabeth Pennisi - 2003 - Science - ‘A hothouse of molecular biology’

A historical overview of the cambridge laboratory for molecular biology

James W. Phillips - April 2013 - Garry Kasparov: Cherwell Profile

https://cherwell.org/2013/06/07/the-cherwell-profile-garry-kasparov/

James W Phillips & Matthew Philips - September 2017 - Simplifying and decentralising UK science funding. A paper outlining critical problems with bureacratisation, risk aversion and pyramid-scheme dynamics in the UK science system.

James W Phillips & Matthew Philips - The Telegraph - July 2018 - Science holds the key to economic success……. Original title ‘Invest in high risk science to create the future’. This is a distillation of the longer-form piece, ‘simplifying and decentralising UK science funding’ 2017 by the same authors.

Available without paywall at https://orion-society.com/2020/02/20/science-holds-the-key-invest-in-high-risk-science-to-create-the-future/

Matt Ridley - Wall Street Journal - The Myth of Basic Science - Oct 23, 2015

Available here: http://www.rationaloptimist.com/blog/the-evolution-of-technology/

Chris Skidmore - 2019 - Speech on research talent, given at the LSE

David Willetts 2019 - The road to 2.4 percent: Transforming Britain’s R&D performance https://www.kcl.ac.uk/policy-institute/assets/the-road-to-2.4-per-cent.pdf

David Willetts 2019 - The road to 2.4 percent lies beyond the University - https://wonkhe.com/blogs/the-road-to-2-4-per-cent-lies-beyond-the-university/

Appendix - Bureaucracy vs start-up

The approach of entrepreneurial start-ups as opposed to institutional, corporate bureaucracies is difficult to reduce to simple points. But we can make a basic sketch of the start-up bureaucracy differences (all broad generalities with many exceptions etc):

Start-ups are small, forming from a self-organising group of like minded people (entrepreneurs). These people are united by some shared view of how the world might be improved by a new idea that runs counter to those being pursued by the established forces, and who would not support it (more cynically, they want to make a profit but these things are relatively aligned hence capitalism broadly works)156. Startups are usually but not necessarily composed of young people (see below but also footnote for critique)157, without extensive prior experience in the field. They have minimal formality and hierarchy, with members all actively engaged in contributing directly to the product with substantial individual autonomy158. Their work is high risk, having a high chance of complete failure, but a large payoff if successful. Individuals have a stake in the outcome. They often require significant long-term investment prior to achieving results. They disrupt existing fields and promote new thinking; they find secrets.
Institutions are larger, more corporatised groups of people. They have steeper and more elaborate hierarchies/seniority159. Their incentive structures support doing more of the same in a more stable, low-risk, short-term, incremental manner. Being in an institution is less personally risky and more immediately financially rewarding than membership of a start-up, favouring different types of individuals. However, incentives favour conformity, so they favour the consensus/compromise view of committees rather than individual vision. As we see in paper 2, start-ups will become institutions over time as they grow unless active measures are taken to prevent this happening (if this is desirable - both end of the start-up/corporate spectrum have benefits). Corporations and government bureaucracies are examples of institutions. In technology, Institutions develop established ‘secrets’ and scale them.

In the research realm:

The university research system is now almost entirely in the second, bureaucratic phase, and this is a necessity of the way that universities are structured to provide excellent teaching and cannot be changed or reformed. People within them do not realise this because they have never worked in alternative systems. See footnote for an outline of this.160
The greatest research laboratories in history a) resembled start-up incubators in many key respects and b) did not look like universities (Paper 2). In ⅔ cases they were funded by tech monopolies161 which the UK lacks.

Venture capitalists still only fund things at a relatively low risk level and short-term horizon, creating a need for a government correction of market failure. We need government funded equivalents of venture capital for the more discovery/invention-centered, longer-term research. We need to promote national research systems, analogous to start-ups, that allow small groups of self-organising people to ‘take the hidden paths’, go against the currently held views, and take bold risks without obvious short term payoff.

Endnotes

1 See a) The authors interview in 2013 with Garry Kasparov , b) Peter Thiel’s ‘Zero to One’, c) Cowen and Southwood 2019, d) Nielsen/Collison 2019 (The Atlantic), e) Bhattacharya and Packalen, 2020 exploring the evidence and drives for innovation stagnation, arguing the current scientific system incentivises exploiting other fundamental discoveries rather than attempting to develop new ones

2 Even during the war it was realised that this spirit of purpose and rise in government funding had yielded much faster technological progress than was the norm. Mervyn Kelly, who would shortly become Bell Labs president, wrote in 1943 that ‘Progress has been made in some fields of technology in a four-year interval that, under normal conditions of peace, would have required from ten to twenty years’ (p60 Gertner).

3 The only two breakthroughs of genuinely paradigm shifting, world changing significance of the last 30 years are CRISPR, which is still quite some way from being a usable technology in the real world, and genetic sequencing, which is partly a scaling of existing paradigms from the early 20th century. Both the 1940s and 1950s contained many such changes (transistors, nuclear fission, DNA structure, genetic code, information theory, neural networks to name a few). For the ‘top 10 discoveries’ of the 2010’s decade, see this new scientist piece.

4 Author of reinventing discovery

5 Nielsen/Collison’s methodology is worth looking at more closely—it is surprisingly hard to find a good metric to capture or disprove the widely held view that we are making fewer important innovations.

6 “The results are slightly more encouraging than physics, with perhaps a small improvement in the second half of the 20th century. But it is small. As in physics, the 1990s and 2000s are omitted, because the Nobel Committee has strongly preferred earlier work: Fewer prizes were awarded for work done in the 1990s and 2000s than over any similar window in earlier decades…. The picture this survey paints is bleak: Over the past century, we’ve vastly increased the time and money invested in science, but in scientists’ own judgement, we’re producing the most important breakthroughs at a near-constant rate. On a per-dollar or per-person basis, this suggests that science is becoming far less efficient."

7 “Back in 1966, Time magazine published an essay called The Futurists in which it was suggested that, by 2020, “machines will be producing so much that everyone in the US will, in effect, be independently wealthy.” Apparently the average family would take home roughly $300,000 in today’s money for doing pretty much sod all. Which would be lovely, were it not so far from the truth.” From Alix Norman in the cyprus mail, 2019 “What did past experts predict for the 2020s?”

8 Even if science has gotten more complex, academic science has failed to make the structural adjustments needed to deal with increasing complexity, such as reforming career structures to handle the need for greater technical expertise and support (instead turning graduate students and postdocs into effective slaves).

9 ‘In addition, many measured researchers may increasingly not be doing full-time research, but rather they are engaged in various bureaucratic enterprises specific to the modern university, or teaching.’

10 Even without Einstein, early 20th century physics is an excellent example of the unexpected nature of scientific advances; In 1924 Max Planck told of advice he received from his teacher back at the beginnings of his physics education some 50 years earlier: “he portrayed to me physics as a highly developed, almost fully matured science...Possibly in one or another nook there would perhaps be a dust particle or a small bubble to be examined and classified, but the system as a whole stood there fairly secured, and theoretical physics approached visibly that degree of perfection which, for example, geometry has had already for centuries.” Physics was utterly revolutionised in Max Planck’s life time, with Quantum mechanics fundamentally and unexpectedly altering our conception of reality.

11 Yet discovery of secrets is not random: they can be searched for. As Feynman said in his Lectures on Physics, ‘a thorough reading of the literature of the time shows they [late 19th century physicists] were all worrying about something’. The seeds of great advances were there, the clues, waiting to be picked up on by curious minds willing and able to challenge the established view. As Kuhn highlighted in his landmark Structure of Scientific Revolutions, there are major institutional impediments to such curious minds having an effect, with consensus paradigms crowding out the new. Kuhn highlighted Universities as a particular engine of this stagnation due to their career structure.

12The first is an increase in incrementalism - ‘... which is why academics usually chase large numbers of trivial publications instead of new frontiers.’.
The second is risk aversion - ‘by definition, a secret hasn’t been vetted by the mainstream’.
The third is complacency - what Kasparov describes as a desire for a ‘safe, comfortable life’ - tenure! (J. Phillips, 2013, Cherwell).
The fourth is flatness - the ability to simply globalise something that already works disincentivizes creating something new.

13 “DARPA: Into the Future” By Fernando Fernandez, 2001

14 In a nutshell, it is increasingly potemkin because what is rewarded is the short-term appearance of success (citations, number of papers, impact factors).

15 Even a local control environment HHMI Janelia, much better than the US NIH in this regard, suffered from this effect - people formed collaborations, prized by the director, simply to look like good ‘janelians’. Our suspicion is that the solution to this is longer timescales of evaluation. The longer the timescale, the more ‘doing’ matters over ‘talking’. You can’t talk your way out of 10 years of bullshitting as easily as you can 1 year of bullshit.

16 The 2020 Wellcome report (see below) concurs with this view, with only 14% of respondents thinking ‘current’ metrics have helped research. Our contention is it is not merely current metrics that are the issue, but the basic idea of distant micromanagement of a complex system.

17 See the second part of this paper, for example the Nyquist example. Essentially, science papers are not all ‘made the same’, some are made by using hordes of postdocs as cheap slave labour, some are made by a great collaboration between professor and student (risking rewarding politics/empire building as opposed to excellence). This kind of thing is hard to assess centrally.

18 Campbell’s law: ‘The more any quantitative social indicator is used for social decision-making, the more subject it will be to corruption pressures and the more apt it will be to distort and corrupt the social processes it is intended to monitor.’

19 Paper by these authors titled the ‘natural selection of bad science’, via Royal Society Publishing, can be found here: https://royalsocietypublishing.org/doi/full/10.1098/rsos.160384

20 Success cannot be ‘ensured’ in a risky enterprise. The attempt to do so, and the language it brings, creates a culture of risk-aversion, planning, and bureaucracy. Even a minor tweak would have helped: “Promoting a successful UK research culture”. That is the mindset we need.

21 “We’re Incentivizing Bad Science: Current research trends resemble the early 21st century’s financial bubble” - James Zimring 2019, Scientific American Blog

22 Yong 2019 on how an entire research field can turn out to be studying noise: ““Beyond a few cases of outright misconduct, these practices are rarely done to deceive. They’re an almost inevitable product of an academic world that rewards scientists, above all else, for publishing papers in high-profile journals—journals that prefer flashy studies that make new discoveries over duller ones that check existing work. People are rewarded for being productive rather than being right, for building ever upward instead of checking the foundations. These incentives allow weak studies to be published. And once enough have amassed, they create a collective perception of strength that can be hard to pierce.””

23 Somehow the conflict of interest statement at the end of papers must include commercial ties but the strong dependency of career progress on paper publication introduces a conflict of interest in academia that is not disclosed…. Arguably if you are personally invested in research being true (industrial research) as opposed to it simply getting published in a top journal (modern academic research) your incentives are better.

24 Another cause is highlighted by David Colquhoun in 2013 (here). He also suggests the university course focus on teaching elementary statistics classes could be driving reproducibility problems (the opposite of their intention), by favouring the simple metric of P<0.05 (which Colquhoun persuasively argues leads to false positives at a >25% rate) rather than an actual understanding of good experimental design and analysis.

25https://slatestarcodex.com/2013/02/17/90-of-all-claims-about-the-problems-with-medical-studies-are-wrong

26 Miyakawa 2020, Molecular brain, “No raw data, no science: another possible source of the reproducibility crisis”

27 The deep academic reliance on a broken commercial publishing system needs serious attention at the national level as it strongly distorts incentives, but is beyond the scope of this paper (we will not comment further on this except to highlight suggestions from AI researcher Lecun) http://yann.lecun.com/ex/pamphlets/publishing-models.html and HHMI’s similar https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3000116 suggestions on resolving this.

28 In a nutshell, technology allows that papers could become more modular. Senior lab heads could, for example, write reviews every 5 years describing the state of the labs work (effectively opinion/reviews/monographs of their own lab’s work), but would not be on the papers themselves, which would describe experiments more directly (this was more the case before people were so incentivised to publish papers). Contemporary papers have often become effectively reviews with single figures being what once would be papers (many are ‘mansions of straw’ as Nobel laureate Kaelin calls it).

29 An excellent comment on this piece from a reader: “A reason for this might be skin in the game.

Managers get credit for causing institutional change. So, if one launches a moon shot then moves onto a promotion at some other organisation, another manager may come in and recalls the rocket before it reaches the moon.

Or more likely, a manager will launch something that they know deep down will not work, but it sounds like it will or it sounds bold and revolutionary and then they leave and move onto other efforts.

In both of these cases the managers do not have the skin in the game to see their projects through to completion and suffer the consequences of failure. There is a rumor such dynamics exist in pharma. I do not know if that is the case, but I could try to figure out if it is.”

30 For example, let's say I want to game the system: if a central bureaucracy decides that the number of external collaborators on papers is a sign of innovativeness, it incentivises people to write papers with lots of largely spurious authors to appear more collaborate (conference opinion pieces do this very well).

31 A couple of excellent blog posts explain the topic: https://www.ribbonfarm.com/2010/07/26/a-big-little-idea-called-legibility/
https://slatestarcodex.com/2017/03/16/book-review-seeing-like-a-state/

32 The blog post contains excellent short description of the problem of metrics, and is found here.

33 Untrustworthiness is, in our opinion, less of a risk than sloppiness in order to get top papers.

34 Excerpts from Edwards/Roy 2017:
‘The modern academic research enterprise, dubbed a ‘‘Ponzi Scheme’’ by The Economist, created the existing perverse incentive system, which would have been almost inconceivable to academics of 30–50 years ago (The Economist, 2010). We believe that this creation is a threat to the future of science, and unless immediate action is taken, we run the risk of ‘‘normalization of corruption’’ (Ashforth and Anand, 2003), creating a corrupt professional culture akin to that recently revealed in professional cycling or in the Atlanta school cheating scandal.’

‘While quantitative metrics provide an objective means of evaluating research productivity relative to subjective measures, now that they have become a target, they cease to be useful and may even be counterproductive. A continued overemphasis on quantitative metrics will pressure all but the most ethical scientists, to overemphasize quantity at the expense of quality, create pressures to ‘‘cut corners’’ throughout the system, and select for scientists attracted to perverse incentives.’

‘....the combination of perverse incentives and decreased funding increases pressures that can lead to unethical behavior. If a critical mass of scientists become untrustworthy, a tipping point is possible in which the scientific enterprise itself becomes inherently corrupt and public trust is lost….’

35 A simple ‘silver bullet’ to some of these problems would be to directly replicate how nonprofits like HHMI and Wellcome allocate funding. They care about research outcomes to a greater extent than do governments, which care more about appearance of accountability.

36 “Several people I talked to describe a similar dynamic and tell me that there are niches dominated by a particular research group that guards that niche almost as its fiefdom. It takes a lot of courage and determination to move in and try to upend the dominant methodologies and research directions of the field. Peer review exacerbates this dynamics. Peer reviewers in your field are your competitors, who have not themselves solved the problem you claim to be able to solve. They have both personal and professional interest (especially so if funding is limited) in giving low scores to grant applications of competing teams and to recommend rejection of their journal submissions. Further, since they’re experts in the grant application topic, while rejecting your paper or grant application, they can lift your research ideas and then pursue them themselves. This happens more frequently than you would expect (a). Further still, committees reviewing grants have all sorts of weird interpersonal dynamics that make funding anything unconventional even more difficult than it would be in their absence, because, while being on a committee people are usually averse to be publicly approving towards anything that seems weird.
An anonymous redditor writes (a):

In my field virtually 100% of the papers in “top” journals come from the same 5-10 senior authors, and they can just about get away with murder.“

37 From Phillips/Phillips 2017 - “A recent experiment by the Conference on Neural Information Processing Systems (NIPS) tested how consistent their ratings of publication quality are. They took 10% of that year's submissions and asked two committees to accept 22.5% of them, then compared the outcome of the two committees. 57% of accepted papers were disagreed upon between the two committees – a clearly unacceptable level of inconsistency that makes acceptance to this prestigious conference in large part a lottery. To put the level of inconsistency into context, if the committees chose totally at random, on average they would disagree on 77.5% of accepted papers – not too far off the actual figure of 57%. Based on these statistics, if the review process were rerun, most of the previously accepted papers would now be rejected (95% CI of 40-75% - for a breakdown of the stats and what they mean for conference acceptance, see: http://blog.mrtz.org/2014/12/15/the-nips- experiment.html). ”

38 From Wellcome’s website, “Does the way Wellcome makes funding decisions support the right research?” By Jonathan Best

39 Nature (not the journal) is full of stochasticity and thrives on noise. Evolution relies on it.

40 Fund people not projects

41 A reader comment: “The valuation of some start-ups in Silicon Value (Wework) are a fascinating symptom of this problem. One suggestion has been to transition from VC funding that purchases equity to debt (a more traditional means of distributing capital). I do not know if there is a way to fund science using debt... (perhaps prizes).”

42 “This planet has - or rather had - a problem, which was this: most of the people living on it were unhappy for pretty much of the time. Many solutions were suggested for this problem, but most of these were largely concerned with the movement of small green pieces of paper, which was odd because on the whole it wasn't the small green pieces of paper that were unhappy.”

― Douglas Adams, The Hitchhiker's Guide to the Galaxy [Notably, researchers were not unhappy prior to the green paperwork frenzy! The desire for control has induced unhappiness]

43 The survey was a) international and b) self-selecting, but the broad accuracy of its conclusions for UK science was confirmed by smaller focus groups.

44 The words ‘young’ and ‘junior’ collectively appear once in the entire 2015 report titled ‘ensuring a successful UK research endeavour’. The word ‘postdoctoral’ in all its forms does not appear once, and ‘postgraduate’ appears once (ironically, it is in the report only because it was quoted in the definition of the research council mission that he is meant to be reviewing). There was scarce mention of similar themes. Contrastingly: ‘Fund’ appears 150 times, ‘grant’ appears 26 times, ‘universit-’’ variants 35 times, total word count 18,939.

45 The report’s total comments on scientific culture (contrast this with the Wellcome report): ‘Good research of all types requires a high quality research culture with proper regard to good practice and ethical behaviour. Science is a high calling in the pursuit of truth which needs to be pursued in a proper and ethical manner. The Research Councils acting together should ensure that the research they fund is pursued in the appropriate ways, and given the scale and breadth of the research funding they provide they should take a leadership role within the UK for research ethics and culture. This is important for society more generally to maintain trust in the research endeavour. A related point for earning and keeping trust, is good engagement about science with the public at large. Effective communication, dialogue and engagement with the public are essential functions of the Research Councils as reflected in their Charters. This should happen in part at the level of individual Councils, and in part at the level of the Research Councils acting together.’

46 Nurse report, 2015: ‘Diversity should be protected in researchers, approaches and locations – recognising that novel approaches and solutions to problems sometimes emerge more readily outside the mainstream. The best research should be funded wherever it is found.’ - in context it is saying: not just universities

47 Such as ‘Policies need to be in place to bring about high quality, cost-effective research carried out to the highest standards, and to ensure that the knowledge produced benefits society and is supported by society, recognising the differences and similarities between discovery, translational and applied research.’ What these policies should be is not discussed.

48 (see Rachael Pells, 2019, for a summary of other problems with Haldane as practiced).

49 Thiel 2020: “We have determined the value of the fine-structure constant, which describes interaction between elementary particles, to a dozen significant digits. But if you ask the authorities for a measurement of the health of science as an enterprise, the reply will be vehement and very vague: “Never has the pace of discovery been so rapid, the range of achievements so broad, and the changing nature of our understanding so revolutionary,” asserted the former president of MIT in Science in 2018.

Such statements may seem like so much hand-waving, but they are hard to challenge given the extreme specialization and implicit invincibility of thousands of scientific subfields. We are forced to wonder whether things are as rosy as the grant writers insist. As for the world outside of MIT’s PR materials, it appears much the same as it was in 1969—just with faster computers and uglier cars.”

50 More generally, Schumpeter’s ‘creative destruction’ is effectively impossible in the system. Creative destruction is the "process of industrial mutation that incessantly revolutionizes the economic structure from within, incessantly destroying the old one, incessantly creating a new one" Schumpeter 1942

51 Biochemist Bruce Alberts (a distinguished scientist and editor-in-chief at Science) and colleagues wrote a 2014 article in PNAS titled ‘Rescuing US biomedical research from its systemic flaws’ that said:

‘The long-held but erroneous assumption of never-ending rapid growth in biomedical science has created an unsustainable hypercompetitive system that is discouraging even the most outstanding prospective students from entering our profession--and making it difficult for seasoned investigators to produce their best work. This is a recipe for long-term decline, and the problems cannot be solved with simplistic approaches. Instead, it is time to confront the dangers at hand and rethink some fundamental features of the US biomedical research ecosystem.’

This is of course of particular relevance in our heading to 2.4% R&D.

52For example, innovation hubs having fully internal funding seems ‘obvious’ - but the fundamental importance of this might not be clear on the surface. Without it, empire building kicks in as people try to get the most resources they can from an external bureaucracy. It is not just a time saver - it modifies behaviour.

53 Alan Kay wrote with Adele Goldberg later (1970s) of this vision: “Imagine having your own self-contained knowledge manipulator…..Suppose it had enough power to outrace your senses of sight and hearing, enough capacity to store for later retrieval thousands of page-equivalents of reference materials, poems, letters, recipes, records, drawings, animations, musical scores, waveforms, dynamic simulations, and anything else you would like to remember and change.”

54 https://www.wired.com/2017/04/youve-never-heard-tech-legend-bob-taylor-invented-almost-everything/

55 Gates could endow a new Bell labs with a relatively modest fraction of his income.

56 ‘.... the workday there [at Bell] was 7.5 hours long. That length was chosen because the leadership considered it optimal: If the day were any shorter it would leave productivity untapped, they surmised, but if it were any longer it would reduce the creativity and effectiveness of the labs' scientists and engineers. (Jim Austin, 2013, Science) (though Gertner has examples saying that they were encouraged to work outside these hours).

57 A critique of the Bell labs example is that Bell labs was much bigger than the others, with on the order of thousands of researchers in total. Some have argued that its returns are proportional to the investment - was it just money or did the structure matter? A counter point to this critique is that only a subset of those were in departments that were actually doing what we would think of as research: many were technicians working on problems for the phone line network (try finding/defining the actual number). A former bell employee estimated that around 200 people were working in the solid state physics department. A counter point to this coutner point is that the work of the ‘real’ researchers built on the more-technician type researchers, who provided problems to them, and so the technical side of Bell was part of the research. What can be said is that, for these authors, those who we have spoken to who worked at Bell speak of it with reverence as a magical place and credit it with the work they did in a way that University people seldom do [though, Universities are the ‘normal’ so maybe they don’t warrant highlighting]. You will see examples of this throughout this paper.

58 As we will see, total internal funding is incompatible with tenure, as tenure involves assessment in thirties/forties and lasts indefinitely, but researchers can decline. Providing total funding requires at least some assessment of an individual’s quality over time, and at these institutes membership of the institute was that assessment.

59 The Cambridge LMB was funded by the UK medical research council. Xerox PARC was funded by Xerox’s monopoly in photocopying. Bell was an industrial lab funded by the AT&T communications monopoly (as with Xerox’s domination of photocopying. The great scientific advances that came from it were the justification for the monopoly (its eventual decline came due to antitrust legal problems). Its monopoly was summarised ‘The [Bell] system constitutes the largest aggregation of capital that has ever been controlled by a single private company at any time in the history of business….. Its gross revenues… are surpassed by the incomes of few governments of the world….. [it owns] 98 percent of the long-distance telephone wires of the United States’.(Danielian, in p45, Gertner). Even these antitrust regulators admitted its enormous impact: ‘[Bell Labs] have not only made things better, but have created new services and industries,’ he wrote of the scientists and engineers. ‘They have also made significant contributions to pure science. For these, no one would wish to deny just praise.’ (Danielian, in p45, Gertner). Even relatively early in its history, Bell Labs contributions were seen as justification for its monopoly status, which made its contributions possible.

60 Full quotation: ‘The fact that annual deaths in this country from one or two diseases alone are far in excess of the total number of lives lost by us in the battle during this war should make us conscious of the duty we owe future generations…. New frontiers of the mind are before us, and if they are pioneered with the same vision, boldness, and drive with which we have waged this war we can create a fuller and more fruitful employment and a fuller and more fruitful life.’(p 34, Bahcall).

61 The origins of modern University research funding structures lie almost 80 years ago. When World War 2 occurred, the US needed to play catch up with the Nazis technologically, leading Vannevar Bush to propose creating a new funding mechanism. He went to the president directly, writing later: ‘I knew that you couldn’t get anything done in that damned town [DC bureaucracy] unless you organised under the wing of the president’. So in 1940 Vannevar Bush wrote a 4 paragraph proposal for President FDR to create an office of scientific research and development (OSRD), which FDR approved immediately. Bush’s OSRD proposal got pushback, with the military establishment saying the new group ‘was an end run, a grab by which a small company of scientists, engineers, acting outside established channels, got hold of the of the authority and money for the program of developing new weapons.’ Bush said: ‘That, in fact, is exactly what it was.’ This crazy, ‘end run…. grab’ system funded a large amount of the technological revolutions that helped win the war and was the predecessor of all major modern US science funding. (see Loonshots by Bahcall)

62 A list of ambitious things being accomplished quickly, to read to generate some anger at the enormous waste that stagnant bureaucracies impose: patrick collison’s Fast list

63 When leading US charity HHMI created a new research facility called Janelia (see part 3), it deliberately declined taking any form of government money in part due to the accompanying bureaucracy that would have stifled its mission.

64In 1954 Bell Labs essentially provided the first Solar Cell without aiming for it (60 Years Ago Today, Bell Labs Unveiled the Solar Cell)

65 A large part of modern medicine

66 Who in the MRC took these bold steps? We need to reward excellent funding allocation, crediting the individuals in the footnote of the paper - obviously peer review committees can’t be credited in this way. We need something like the impresarios in the Dance world (this suggestion was made to us by Alan Kay, who suggests ‘Dance to the piper’ as a book exploring how this works in the dance world).

67 There is a total absence of aiming for this in many bureaucratic circles.

68 As Buckley wrote to Jewett about Bell Labs (1938, AT&T archives & p77 gertner): “No attempt has been made to achieve the character of a university campus with its separate buildings… On the contrary, all buildings have been connected so as to avoid fixed geographical delineation between departments and to encourage free interchange and close contact among them.” The building had long corridors stretching 700 feet, with it being impossible to travel ‘without encountering a number of acquaintances, problems, diversions, and ideas’....It was like being ‘a magnet rolling past iron filings…. By intention, everyone would be in one another’s way’ (Gertner p77).

69 A reader from UCL commented ‘[this is] extraordinarily different to [University] departments, where people often don't even tell others what experiments they are doing or [what] data they have’

70 Brenner himself said:

‘I believe that this type of on-going conversation is critical in science. I’m not the sort of person who likes to think in isolation. When an idea germinates in my mind I know its at least fifty percent wrong initially. And it's only in playing with the idea by bouncing it off others than one can refine it and retain what’s essential’ (p118, Friedberg)

71 This was how Brenner was acting even when already clearly recognised as one of the greatest scientists of the 20th century.

72 Malcolm Geffer, now MIT Professor Emeritus (Friedberg, 2010, pg 117), remembered the LMB at this time:

“I remember the old days at the MRC more for the long lunch hours, coffee breaks and Saturday mornings than I do for the time spent at the lab bench…. One of the greatest lessons I learned was not how to do an experiment but how not to do one…. [Experiments] were the embodiment of an intellectual exercise and not a vehicle for filling a notebook for a publication, nor were they a vehicle to achieve fame or to beat the competition. Thus before doing an experiment at the MRC, it had to pass the rigorous intellectual challenges conducted over coffee. What survived these challenges were only the best experiments you could think of , amended by the jurors (your colleagues). If you survived coffee, how could you not do an outstanding piece of research?”

73 Richard Feynman, quoted in chapter 2 of Cal Newport’s Deep Work.: ‘To do real physics work, you do need absolute solid lengths of time…. It needs a lot of concentration… if you have a job administrating anything, you don’t have the time. So I have invented another myth for myself: that I’m irresponsible. I’m actively irresponsible. I tell everyone I don’t do anything. If anyone asks me to be on a committee for admissions, “no,” I tell them: I’m irresponsible”

74 Isaac Newton: “Truth is the offspring of silence and meditation. I keep the subject constantly before me and wait 'til the first dawnings open slowly, by little and little, into a full and clear light.”

75 Santiago Ramon Y Cajal, Advice to a young investigator, Chapter 2: ‘During the gestation period of our work, sentence ourselves to ignorance of everything else that is going on— politics, literature, music, and idle gossip…..The harm in certain things that are too distracting lies not so much in the time they steal from us as in the enervation they bring to the creative tension of the mind, and in the loss they cause to that quality of tone that nerve cells acquire when adapted to a particular subject.’

76 Einstein in his patent office is a rare and interesting exception to this need for absolute focus.

77 Diversions not only fragment the long uninterrupted periods of time needed for productive work, but they limit free time for collegial interactions and perhaps most importantly they clutter the mind and prevent the almost monastic discipline needed by many to do great work.

78 A reader commented on the implicit ‘pilot data’ requirement: “ the pilot data thing is absurd. you have to show the experiment will work before doing it, it means it's not an experiment..”

79 The successful multi billion dollar science funding experiment, HHMI Janelia, declined to be directly affiliated with any University to protect itself from their bureaucracy.

80 This is typical. I spent the first 2 months (of my 10 month 1st year placement) of my Cambridge PhD waiting for approval of a piece of paperwork we had submitted months ahead of time, and which I had already had approval for at Oxford.

81A quote from the new scientist article: ‘“The ambitions [of UK ARPA] are all sensible and laudable,... “The thing I find bizarre is the focus on creating a new institution. There is nothing in there [on the UK ARPA proposals] that cannot be achieved in the existing situation with UKRI.” If ARPA is within UKRI its autonomy, flexibility, speed and culture - the very things it is designed to be unique in - will be altered by the larger bureaucracy.

82When managing research, a critical mindset to have is that of Kelly Johnson in the 1940s Lockheed Skunk works: ‘Everything possible will be done to save time’ (p124, Rich, Skunk Works). I’ve not found detailed studies of Lockheed Skunk Works. However the basic model is: ‘All skunkworks are modelled on the Lockheed aircraft company's secret research-cum-production facility where, in the 1940s, staff were removed from the corporate bureaucracy and encouraged to ignore standard procedures in the hope that they would come up, in the first instance, with a high-speed fighter plane that could compete with those produced in Germany by Messerschmitt. So successful was the concept that the company continued with it, and its skunkworks came up with a number of other innovative products, including the notorious U2 spy plane.’ - From the economist 2008

83 Some big US labs work in this way, but a) they are much smaller than institutes and b) the senior lab takes the lion’s share of the credit for the work. Also young people overlap in the lab only for quite limited amounts of time making sustained collaboration challenging.

84 This came from a paper “UK Science Funding Observations from a Particle Physicist” that we were sent.

85 Additional quotations of the LMB’s focus on picking top talent: As a Guardian profile of the first MRC Laboratory of Molecular Biology Director Max Perutz put it:

“Impishly, whenever he was asked whether there are simple guidelines along which to organise research so that it will be highly creative, he would say: no politics, no committees, no reports, no referees, no interviews; just gifted, highly motivated people picked by a few men of good judgment. Certainly not the way research is usually run in our fuzzy democracy but, from a man of great gifts and of extremely good judgment, such a reply is not elitist. It is simply to be expected, for Max had practised it and shown that this recipe is right for those who, in science, want to beat the world by getting the best in the world to beat a path to their door.”

Indeed, Perutz did not actually call himself director, and there was little to no attempt to manage: instead, the management committee saw its role primarily as being to attract the finest minds to the LMB:

Until Perutz retired in 1979, it [LMB] had no director. Perutz didn’t want to be one, and it meant he could retain his lab space after retirement. Instead, the lab had a loose management committee, which met occasionally and saw its main job as attracting outstanding talent to the lab. (Bynum 2012).

86 As the first Bell Labs president, Frank Jewett, said:

an industrial lab is “is merely an organization of intelligent men [sic], presumably of creative capacity, specially trained in a knowledge of the things and methods of science, and provided with the facilities and wherewithal to study and develop the particular industry with which they are associated…. It is likewise an instrument which can bring to bear an aggregate of creative force on any particular problem which is infinitely greater than any force which can be conceived of as residing in the intellectual capacity of an individual.” (p32, Gertner)

87 From a reader: “An interesting hypothesis about why the selected individuals were unorthodox but turned out to be good is that past performance is not always a predictor of future success:

https://science.sciencemag.org/content/sci/354/6312/aaf5239/F1.large.jpg

The above figure shows how it is difficult to predict when a profound discovery will occur in a scientist's career, so much so that it is fairly random relative to the publication record.
Individuals who are hired when their publication record doesn't already contain a clear seminal discovery over ones who have such achievements might be considered unorthodox. Yet, those hired individuals would nonetheless be the correct choice.”

88 Kompfner happily made offers to people who still hadn’t even finished a PhD yet (this is an example of a top-down control in Bell, which we’ll come to later). This also highlights that in contrast to the modern approach of requiring extensive experience in a field before funding people, all of these institutes were willing, perhaps even preferring to hire outsiders to the field.

89 “The pressures of such models are clear. Working at Bell “was an incredibly highly competitive atmosphere”, says Cherry Murray, a physicist who spent 26 years at the lab in research and management positions, including research vice president, and is now dean of the Harvard School of Engineering and Applied Sciences in Cambridge, Massachusetts. “You were given some leeway, say for a few years after your arrival, to build up your research programme,” …. Evelyn Hu, an electrical engineer at Harvard who spent nine years as a Bell researcher, recalls a chilling prophecy from company management early on. “I remember attending an orientation for new hires and being told, ‘Look to your right, look to your left — in five years, only one of you will be here’,” she says. (Kaplan, 2011, Nature)

90 Idle curiosity could be seen as the opposite of (predictable) impact. Idle curiosity in this context is simply very smart people finding some part of the scientific unknown interesting and deciding to ask questions of it.

91Claude Shannon, the person who wrote what Scientific American called ‘ ‘the magna carta of the information age’ (p127 Gertner) underlying modern technology, said: ‘I am very seldom interested in applications….. I am more interested in the elegance of a problem. Is it a good problem, an interesting problem?’. He was not interested in the direct impact of his work but in the problem itself. To him, it was blue skies curiosity research. Even in the relatively bureaucracy free 1950s, ‘“[Shannon] couldn’t have been in any other department successfully.. But then, there weren’t many departments where people just sat and thought” (Brock McMillian, quoted in Gertner 132).

92 Even in its early days the LMB let researchers stumble around looking for gold, something unheard of today: “The major credit I think Jim and I deserve... is for selecting the right problem and sticking to it. It’s true that by blundering about we stumbled on gold, but the fact remains that we were looking for gold” Francis Crick.

93From John Fleischman at american society for cell biology news: ‘Doudna says that CRISPR demonstrates both the value and the unpredictability of basic research. CRISPR began for Doudna with a request from a Berkeley colleague, geomicrobiologist Jill Banfield, to sequence a bacterium sampled from the highly acidic wastewater of a mine in northern California. Recently Doudna told Andrew Pollack of the New York Times, “I remember thinking this is probably the most obscure thing I ever worked on.”

Confronted with her “obscure” quote, Doudna explains, “I guess my point there is that for all of us in science who are doing basic research, we don’t set out with a practical goal in mind beyond the goal of understanding how things work. That was certainly true here. For me and for my lab, the idea of working on a bacterial immune system and studying how bacteria deal with viral infection was a fun, fascinating project that I certainly never anticipated would lead to something like this.”
Doudna continues, “But that’s such an important thing for people, particularly for non-scientists to appreciate. That’s the nature of science. It’s about looking under the proverbial rocks and every now and then you find something. By just paying attention, you’ve found something that has broader implications.”’

94 Fortunately due to the culture of Bell Labs which was not impact obsessed, research continued and the expertise surrounding cellular technology stayed in house, it just didn’t get ‘picked up’ as quickly as it could have.

95 See page 32, Yudkowsky’s ‘Inadequate equilibria’

96 Simple answer: they grew talent by giving them independence then promoted them.

97 The example is of one of the world’s leading surgeons, who created a new way of doing surgery that is now used worldwide. Rather than take credit for others in later career, he gave up senior authorship on his papers, letting those driving the research take the credit. He himself took up a mentoring role, guiding these younger researchers. His reward? He did not get ‘Ref returned’ by the institute, an institutional humiliation. The system forces you to be a fiend and take credit for the work of the young.

98 Harry Lambert writes of Robbins’s report on Universities from the 1960s: “He went on: “We should deplore any artificial stimulus to research.” Published work, he added, “counts for too much”. Erudite teachers who never published were nevertheless “priceless assets” to any university.”

99 As Watson wrote of Crick in his 1968 book the Double Helix “ Already he is much talked about, usually with reverence, and someday he may be considered in the category of Rutherford or Bohr. But this was not true when, in the fall of 1951, I came to the Cavendish Laboratory of Cambridge University to join a small group of physicists and chemists working on the three-dimensional structures of proteins. At that time he [Crick] was thirty-five, yet almost totally unknown. Although some of his closest colleagues realized the value of his quick, penetrating mind and frequently sought his advice, he was often not appreciated…” Thanks to Steve Hsu for the quotation from his blog.

100 (LMB perhaps less than PARC/Bell but I lack evidence for LMB)

101 (I find no exception in Gertner’s book. This may also be true for LMB(all directors internal))

102 (Newton being history’s greatest internal hire - with his Professor resigning his position for him (Russell offered to do likewise for Wittgenstein at Trinity))

103 In Universities, equipment resources are largely owned by individual labs. As science gets more expensive due to increasing technologies, this strongly favours established forces, imposing a large start up cost and thus penalising disruptive/displacement forces. Further, even when core resources are available, most junior lab heads lack the resources to use them. There has been no pushback against this trend. In the LMB and Bell Labs resources were shared and there was excellent technical support through technicians (p154 Gertner for the technician levels at Bell Labs). Or at the LMB:

When LMB researchers needed a new instrument, Perutz made sure technicians and engineers were there to build it, a model he learned at the Cavendish. “We were interested in topics that stretched the techniques,” says Walker, explaining how the lab developed technologies such as x-ray crys- tallography, DNA sequencing, and confocal microscopes. (pennisi 2003).

As Gerry Rubin said in an internal HHMI Janelia talk ‘I never had resources turned down for an experiment at the LMB’.

104 A brief look at the career stage at which the most important discoveries in history were made will show that young minds do the most disruptive work. It is a simple exercise: Newton, Darwin, Curie, Einstein, Crick, Watson, Maxwell etc were young people (usually in their twenties) when they did their famous work. Counterexamples are relatively lacking. This is an inconvenient truth for the scientific establishment. (Tracking research output over a career, as many studies do, is not equivalent to showing that age is not a factor. One needs to simply look at the major discoveries, the genuinely disruptive ones that promote new ways of thinking.)

105 The view that young scientists are the drivers behind great science has been expressed clearly and often by the greatest scientists: Claude Shannon wrote that ‘I believe that scientists get their best work done before they are fifty, or even earlier than that… I did most of my best work when I was young.’ (Gertner 317) (Fellow computing pioneer Alan Turing of course was very young also and died before decline). Newton described his early 20s as ‘The prime of my age for invention’. Einstein wrote ‘"A person who has not made his great contribution to science before the age of 30 will never do so.”. James Watson [on how to produce great science]: ‘Take young researchers, put them together in virtual seclusion, give them an unprecedented degree of freedom and turn up the pressure by fostering competitiveness’

Indeed, there are warnings of the need to be naive and stay away from the past. ‘“You should take all elderly scientists with a grain of salt -- including me.” Roger Tsien, Nobel Laureate and pioneer of protein engineering, at age 61, quoted in The San Diego Union-Tribune, March 21st 2013. Cajal wrote: “It is idle to dispute with old men. Their opinions, like their cranial sutures, are ossified.” (Santiago Ramon Y Cajal, Nobel Laureate (1905) Widely viewed as being the father of modern neuroscience.)

106 Methodology in the link, thanks to Alexey Guzey for suggesting this.

107 This is true even when that expertise is founded on little of substance in immature fields - people fill the professor positions by default even without having made important advances. This is a major problem in neuroscience today. A quotation from Alexey Guzey relating to this: ‘One interesting anecdote from a friend of mine from Stanford: she believes that basically the entire field she's working on is bullshit because people didn't have good enough methods in the past, so everything they know is wrong. She wants to develop new methods and to make the field actually find something real. When writing NIH grants she has to somehow fundraise for the research that will invalidate all of the senior scientists' research BUT that they would approve to fund!’

108 A way in which the tenure-based teaching harms teaching may come from the slow turnover of subjects, such as the lack of programming courses in many leading science programs.

109 Look at what happens to professional chess players in the age range 20-60. Almost all top players are under 40. See 2700chess.com

110 We heard a similar quote directly from another Nobel laureate who invented a new field (in discussion 2019), who also knew brenner:‘I’m an ageist, the only way great things happen is to give power to the young, or to more senior researchers like me who deliberately keep things very small.’

111 We have been unable to source this Rousseau quotation

112 James Watson: ‘The older the scientist you choose to do your Ph.D. thesis with, the more likely you will find yourself working in a field that saw its better days a long time ago, possibly before you were born. Even when a mature scientist still has all his marbles, he often wants to put more bricks into an edifice that already has enough rooms……. Young professors in contrast are generally hired not for grandeur but because they represent a new intellectual thrust not present in a department, one with hopes of remaining lively over at least the next decade. Moreover, they are likely to have smaller research groups than more senior professors, around whom funds as well as stodgier minds tend to aggregate.’ from page 50, James Watson’s 2007 Memoir ‘Avoid Boring People, and other lessons from a life in science.’

113 Pages 100-200 of Einstein’s biography by Isaacson are replete with examples of how, even after publishing 4 of the most important papers in scientific history in 1995, he struggled to break through in large part due to the hierarchical university system, even modifying his thesis topic to one less likely to cause offence to seniors.

114 Edited with [] to change tense only.

115 Source: Geoff Hinton at ACM (youtube), 25 minutes in.

116 See page 30/31 of Yudkowsky, Inadequate Equilibria, most grantmaking in academia is about prestige not promise. -- “From the perspective of most grantmakers, the ideal grant is one that gets their individual name, or their boss’s name, or their organization’s name, in newspapers around the world in close vicinity to phrases like ‘Stephen Hawking’ or ‘Harvard Professor’.”

117 Ben Barres in Nature 2017, “Stop blocking postdoc’s paths to success.’’.

118 In the preface to the selfish gene, Richard Dawkins noted:

"I recently learned a disagreeable fact: there are influential scientists in the habit of putting their names to publications in whose composition they have played no part. Apparently some senior scientists claim joint authorship of a paper when all that they have contributed is bench space, grant money and an editorial read-through of the manuscript. For all I know, entire scientific reputations may have been built on the work of students and colleagues! I don't know what can be done to combat this dishonesty. Perhaps journal editors should require signed testimony of what each author contributed. But that is by the way.”

119 A pernicious feature is that papers are referred to as coming from X lab, as opposed to having been done by the researcher who did the work. Here is an example of this kind of language, in the NYT (Andrew Pollack, May 11th 2015): “Dr. Doudna, whose expertise was in working with molecules, not cells, reported such a demonstration in human cells in January 2013. But her report came four weeks after two papers were published simultaneously, one by George Church at Harvard and the other by the Broad Institute’s Dr. Zhang.” George Church and Feng Zhang are the professors, not the lead researcher.

Just this morning one of us read a paper with three very senior individuals (including a Nobel laureate and an institute director) and two junior researchers. The biggest contribution of any of the three senior authors (according to the papers author contributions list) was ‘conceptualization’ of the study, which is a very wishy washy thing. One of the senior authors didn’t even have a contribution listed. Yet the ‘correspondence’ email address was of course the two senior people despite their apparently somewhat limited familiarity with the work. In another, an author was listed simply for allocating financial resources.

120 This was not always the case: A senior scientist relayed a story to one of us how Max Perutz, nobel laureate in chemistry, giving a talk at HHMI in 2000 aged 86. He was describing detailed experiments his lab had done. After a while it became clear to the apparently obvious amazement of his audience: Perutz was describing experiments that he was doing himself, with his own hands. Likewise, Sanger, Brenner and the other founders of molecular biology were known to do experiments themselves, even when they were already extremely famous scientists (Sanger was a double nobel laureate).

121 We have been recommended ‘nobel dreams’ by Gary Taubes as an example of a similar but worse story of a physics nobel. Though we haven’t read it yet it may be a useful resource for others looking for a physics angle.

122 By this, we mean that Crick and Watson would be working in the labs of the senior scientists, and the credit would go to them, as would the nobel prize. A joint author paper between two junior researchers is almost unheard of today - what has changed?

123 Credit assignment is always complicated without the accompany distortant of a power asymmetry: A reply from Geoffrey Hinton on reddit.com is interesting re-credit assignment on scientific papers and how people misweigh their own contribution:

“Here is a really valuable fact of life: If some people collaborate on a paper and you get each of them to estimate honestly what fraction the credit they deserve it usually adds up to a lot more than 1. That's just how people are. They notice the bits they did much more than they notice the bits other people did. Once you accept this, you realize that the only way to avoid credit squabbles is to act in a way that you think is generous and encourage your co-authors to do the same. If everyone insists on getting the credit that they think is rightfully theirs you are likely to get a nasty squabble.”

Our contention is that lab heads/senior people can often unintentionally delude themselves about how important/irreplaceable their contribution to a project was.

124 In 2017 Nature published a featured comment/article by Ben Barres arguing “Stop blocking postdoc’s paths to success.’, criticising the common practice of not letting a junior scientist take their project/line of work with them when they start their own lab! “...... why do so many PIs do it? Highly competitive lab leaders wishing to become the best in their field can feel that they are working on a half-eaten pie if they focus on a research question to which oth- ers are contributing. They may imagine that their chances of winning a Nobel prize or other prestigious award are lessened if there are too many contributors to a field. Or they can under- standably feel they have invested their entire careers in developing a project area, whereas the post- doc has invested only a few years and relies on building on the PI’s previous work.”

125 “There really is a culture problem in academy... I wonder if people worked together more, side-by-side if that culture problem would still exist…. The Sanger example is supposed to demonstrate, I think, that if you keep a scientist in the lab or doing the analysis and not doing paper work, they are better people.”

126 A very senior HHMI scientist highlighted to us that postdocs used to just be a couple of years extra training to learn a new method from a new lab. Now 10 years of postdocs, consuming one’s thirties, are not unusual. There is a strong case for government examination of this as a form of wage dumping and an exploitative employment practice.

127 A further interesting point Brenner makes is that the creeping bureaucratisation has affected the ability to train exceptional creative scientists:

‘Cambridge [UK] is still unique in that you can get a PhD in a field in which you have no undergraduate training. So I think that structure in Cambridge really needs to be retained, although I see so often that rules are being invented all the time. In America you’ve got to have credits from a large number of courses before you can do a PhD. That’s very good for training a very good average scientific work professional. But that training doesn’t allow people the kind of room to expand their own creativity. But expanding your own creativity doesn’t suit everybody. For the exceptional students, the ones who can and probably will make a mark, they will still need institutions free from regulation.’ Brenner, Dzeng interview

128 A useful general practice for government across areas would be: whenever you commission a report from a ‘senior type’ (FRS, Lords etc), also commission one from a promising young person under 35. Our bet is the younger report will contain a great deal of value absent from the perspective of the elders. Would need to consider confidentiality of report as ‘speaking truth to power’ is a career suicide move.

129 There is a big difference between working in somebody’s lab, with their priorities, their equipment selection, their lab culture, their spending decisions, versus being given control as the principal investigator.

130 An irony: as a young researcher you can usually actually get more independence in a large lab than a small lab. But the credit for the work will often not go to you.

131 This tenure system and general career structure also strongly selects against risk taking (due to consequences of failure to publish) and disruptive/risky candidates (giving tenure to a risky bet is a big step), creates a large amount of dead wood, and places power almost entirely in the hands of the established forces and not the disruptors (the young).

132 Only the LMB had any semblance of a tenure system and that was very limited (Rubin 2006).

133 The idea, as we understand it, is it ‘frees’ people to do outside the box research. But given that outside the box research is so difficult to get funded for, in practice this doesn’t work as intended. Additionally, outside the box research seems more the province of outsiders/the young than the tenured.

134 Full quotation: “‘….. there was nothing to do at the [Bell] labs all day but work. I have known lots of middle-aged professors who don’t spend much time teaching but don’t do any research either. At Bell Labs it was harder to be deadwood.”

135 A tweet exchange on this point: Three time UK science minister special adviser Stian Westlake tweeted ‘An important rule of education is that 18-year-olds should be taught by full-time teachers, but 19-year-olds should be taught by R&D practitioners. The more I think about this, the curiouser it seems.’ One of our authors replied ‘What happened a lot more in the past in US (Bell Labs) was that people worked in industrial labs in their youth doing research, then semi-retired to a teaching University. Shockley, Shannon + others from Bell Labs did this. Research first, teaching later.’

136 Of Bell/LMB, ‘Group leaders were active bench scientists, and this was true even for Nobel Prize winners and department chairs.’ (Rubin 2006). Or as Pennisi writes of the LMB:

Even senior scientists worked at the bench—a tradition that continues today [edit: people I know at LMB seriously dispute this] and goes far to explain the lab’s vitality, says Gerald [Gerry] Rubin of the Howard Hughes Medical Institute in Chevy Chase, Maryland, who did his graduate work at LMB. Perutz spent 90% of his working time at the bench until his death last year, focusing most recently on neurodegenerative disease. Klug still maintains a lab, although he’s officially retired. (Pennisi 2003)

137 A comment from a reader about this sentence: “I think we have lost most of the virtues that were once intuitive. It would be interesting to know why.”

138In Universities many Nobel Laureates/senior people with lots of funding could get away with murder (not literally, but for their victims career), due to the enormous financial and political power they hold (though #MeToo has shifted this).

139Examples from the ARPA/Parc lineage are many. Ivan Sutherland, a PhD student, invented Sketchpad in 1963 as part of ARPA. It was a kind of touchscreen computer when all that existed was essentially mainframe computers. People could draw interactively on a computer screen for the first time, inventing both computer graphics and interactive computer graphics.. Sutherland (deservedly) got sole credit. Today Sutherland’s boss would get the credit (at least in the biology fields), as well as any resulting major prizes (Sutherland himself got the Turing award). At the LMB, examples are Watson/Crick’s paper, also papers form Brenner’s lab in the 1970s often did not even have Brenner on the paper, the common practice at the time. At Bell, a strict notebook system ensured credit was assigned correctly, working out ‘who did what’ as Gertner puts it (though the motivation was patents). Manager’s at Bell were therefore not taking credit for the work of the researchers, but rather had a sincere desire to manage well (p56-57 Gertner). Many modern researchers decry what some claim (we are no experts on it) happened to Rosalind Franklin (absolution of credit) whilst doing the same to their own students.

140 See, for example, https://www.hhmi.org/news/uncovering-the-neural-basis-for-hypothetical-thinking-in-rats

141 Bill Shockley, for example, is cited as being utterly unable to manage people despite being a nobel laureate (p77 Gertner). The Bell system recognised the differences in personalities that arose and were built spontaneous through interactions (p88 Gertner ), something not manageable centrally, and the fact that instigators, and that project instigators, managers and bench scientists were (p196). The strict linear career progression and power accumulation of University academia was not present in the Bell system and it does not seem to have harmed it.

142 In biomedicine professors sit in the position of managers but actually operate as a strange blend of top-down controller, grant application machine, and lab figurehead.

143 ‘Its the interaction between fundamental science and applied science, and the interface between many disciplines, that creates new ideas.’ Bell Labs president Mervyn Kelly, p345 Gertner

144 A deliberate strategy (called rendanheyi) at chinese company Haier aimed to have thousands of microenterprises within the company, connected in a flexible way (like the internet, as the article describes it), with around 10-15 people each. The company structure was designed to create the ‘the world’s first company for the internet age’ (Hamel and Zanini 201 HBR), a decentralised organization at scale without monstrous bureaucracy or micromanagement. I strongly recommend reading the article (as Hamel/Zanini say in their conclusion - ‘Who would have imagined that it’s possible to run a large global business with just two layers of management between frontline teams and the CEO?)’. Information technology should only assist the move from bureaucracy to self-organisation, as the usual information processing performed by bureaucracies can become automated.

145 From https://www.wired.com/2017/04/youve-never-heard-tech-legend-bob-taylor-invented-almost-everything/

146 It was very difficult to pull the same trick in LMB/Bell/PARC because you couldn’t grow your lab. The small lab sizes made the institutes non-zero sum - collaboration was intrinsic to the system and didn’t need special central incentivisation, and the system prevented gross power distortions.

147 The Dream Machine (TDM) page numbers refers to the 1st edition.

148 Do senior people want managerial positions? A janelia senior group leader commented to me when HHMI seemed to be struggling to find applications for team leader positions (a $250 million position over 15 years): “People like money to spend themselves, but they don’t seem to want to be director of allocation of money to others who then work independently with it”. These people might be as rare as nobel laureates.

149

150 Leslie Berlin in Wired: “Heroic lone-wolf entrepreneurs may be the preferred heroes of narratives spun by the media, but history has shown us that teams—and the networks that come from them—are the true engines behind innovation in Silicon Valley and far beyond. No one understood this better than Bob Taylor.”

151 As of the next individuals I struggled to find many of their dates of birth, but it looked like directors have gotten older at time of appointment, and that the terms have recently gotten longer also.

152 He was 94 when he died in 2006 according to this obituary.

153 An interview with Betts

154 Dates of terms taken from DARPA’s website: https://www.darpa.mil/attachments/DARPA_Directors_Sheet-web.pdf

155 See https://wonkhe.com/blogs/people-one-chief-to-rule-them-all-ukri/

156 Thiel (p105) quotes Faust in relation to this:
‘The few who knew what might be learned,

Foolish enough to put their whole heart on show,

And reveal their feelings to the crowd below,

Mankind has always crucified and burned.’

157 A reader critiqued this with a good point - many start-ups come from somewhat older people (Why Great Entrepreneurs Are Older Than You Think). Perhaps it is highly unusual/disruptive start-ups that come from the young. Perhaps age is not the crucial ingredient for outside the box thinking, but rather something it correlates with - ignorance. This is part of Sydney Brenner’s view of why youth matters most in science. An error we made in our discussions for this paper is to conflate age and ‘outsider’ status too strongly - they correlate strongly but are not the same.

158 This is almost tautological - start-ups can’t have bureaucracy because they are new. Our response: The only way to prevent bureaucracy is to make it impossible. Young companies, like young people, suffer (have) cancer less.

159 Bahcall in Loonshots emphasis how, even though they are ‘part’ of large companies, entities like Skunk Works, PARC, Bell labs were deliberately relatively siloed from the main corporation.

160 Three reasons: firstly, universities are large, bureaucratic institutions, and are correspondingly inherently slow, political and risk averse both in organisation and career structure. Secondly, career paths within Universities and related structures preclude self-organisation of young people and strongly favour established forces, with increasing concentration of power and responsibility in the hands of senior people over time. It takes around two decades for young researchers to achieve meaningful independence, by which point they are no longer young. Thirdly, attempts to centrally manage the university research ecosystem has introduced severe distorted incentives into the research process leading to an increasingly Potemkin academy in which productivity is rewarded over being correct.

161 As a reader commented, it is no accident these were tech monopolies.

Chance McAllister

Thinking & Learning in Public

Oct 7Liked by James W. Phillips

This is by far the best article (maybe not the right word but whatever it is) I've come across this year by far! Thanks for writing it.

Expand full comment

Ally

Nov 2, 2022

Maybe one for a separate blog post but would be really interested to hear about how and why ARIA made it through the 'Whitehall machine' and why the Lovelace twin did not.

James W. Phillips' Newsletter

Discussion about this post