S&T - Lovelace vision document: ARIA's proposed twin - a network of new research laboratories using different organisational principles
S&T - Lovelace vision document: ARIA's proposed twin - a network of new research laboratories using different organisational principles
‘You will not concede me Philosophical poetry. Invert the order! Will you give me poetical philosophy, poetical science?’ Ada Lovelace
The applied metascience movement is trying to reinvent how science and technology is done to make it radically more effective and address structural limitations, unlocking new areas of technological creation. As Michael Nielsen and Kanjun Qiu brilliantly explore in their ‘vision of metascience’, foundational aspects of the career path, incentives, and methods of sharing ideas in innovation may need to be reinvented. It is in this nitty gritty detail of the social processes of science that transformative change can come about. This search for a new kind of science is now gathering pace with philanthropic funders in the US. In my view there is an urgent global need to create new structures that complement the strengths of academia and diversify to address its weaknesses.
One of our guiding lights of our Number Ten science and technology team was that there are examples of transformative science and technology organisations in the past that hold lessons about how to do transformative research that have been forgotten. These organisations operated on fundamentally different principles to modern structures. Our teams push to create the UK’s new Advanced Research and Invention Agency was modelled in part on the early 1960s ARPA.
What else is there in this idea space?
ARIA was one of several creations we wanted to birth in the UK to make the UK a global science and tech hub. ARIA had a twin program, the Lovelace Program. Whilst the program manager is the core of the ARPA model, the Lovelace program looked to complement this with physical labs operating on different principles to almost all existing labs today, learning from major historical examples of outlier success. Whilst not seeking to clone them like for like, they provide ‘existence proofs’ of very different ways of incentivising and organising researchers.
Whilst the UK has quite a number of non-university research labs, they are typically structured in a way that is quite similar to academic departments. Yet this structure was created to teach and transmit theological knowledge hundreds of years ago, not to promote invention, suggesting other models might hold value. Yet today there is relatively little attention to how research itself is organised. For example, there is a strong excitement and attention around what DeepMind is doing, but very little attention to how DeepMind is organised differently that lets it achieve its results (Demis studied Bell Labs in setting up DeepMind).
I was fortunate to meet and discuss how to reinvent science and tech with a range of people who had been at what are probably the 3 most outstanding labs in the 20th century. These included Nobel laureates Sydney Brenner and Eric Betzig, as well as the Turing Award winner Alan Kay (me with them in photo below, slide from my metascience 2022 talk).
What was striking about these conversations and subsequent study was that, though the laboratories they worked at were radically different in scale and focus area, the features that they described as being important to their success were highly similar. More intriguingly still, these common elements were radically different to how research is organised today. This suggested a method to designing new research laboratories that looked at the intersection of the Venn diagram, asking what is common to them?
A related question is: if you were to create a 21st century version of the Max Planck Society, learning the lessons from 20th century science, what principles would you use? One part of the motivation for this was a conviction that there was a massive global resource for the UK in disenfranchised young scientists and inventors for whom the existing system was exploitative, and we should invent things to attract them.
Below is a vision document that we wrote based upon this study. To see some of the raw research notes underlying this, please see this blog post. The doc was oriented to researchers who might be interested in helping in such a program, not policy makers, and was shared widely in 2020.
Since we wrote this, pioneering organisations like Arcadia in San Francisco have begun pursuing new ways of doing science. In early 2021 we discussed and shared the Lovelace documents with the primary funder of what became Arcadia, and I’m extremely excited by the path they have gone down. They are by some distance the most unusual and distinctive lab at scale that I know of, seeking to reinvent quite fundamental aspects of how science is done.
Credit to the brilliant artist Evie Kitt for the logo design of Lovelace. Credit also to Adam Marblestone, Chiara Marletto, Alexey Guzey, Virginia Rutten, Gaurav Venkataraman, and my brother Matt who gave a significant time in 2019/2020 discussing and helping with the thinking around this and reviewing and improving the doc. It was also kindly reviewed by a number of people who had worked in the institutes we wanted to learn from, such as Eric Betzig and Alan Kay.
Note that this was written for a specific time and context, but 99% of it I think holds true.
Endnote #’s are in brackets [].
This is a length limited vision doc, not a plan or detailed policy doc. I presented some of this argument at the Metascience 2022 conference in California, and I will write up a less poetic and more practically focussed version of this along the lines of that talk. To receive that doc, please subscribe:
‘You will not concede me Philosophical poetry. Invert the order! Will you give me poetical philosophy, poetical science?’ Ada Lovelace
The Lovelace Program, V1.2[1], August 2020
“Impishly, whenever [Max Perutz] 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 [people] 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.”
Obituary[2] of Max Perutz, First director of the MRC Laboratory of Molecular Biology, [similar 3]
“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……Today [we] have developed a new culture in science based on the slavery of graduate students…….The most important thing today is for young people to take responsibility, to actually know how to formulate an idea and how to work on it. Not to buy into the so-called apprenticeship.”
Sydney Brenner, a pioneer of molecular biology, see also [4]
The Lovelace Program, V1.2
Summary - An Ada Lovelace Society
Motivations to diversify beyond academia
Inspirations: commonalities between Bell labs, Xerox PARC and the Cambridge LMB
Lessons from recent smaller independent institutes
Four core features of a new type of national lab structure: Lovelace Fellows, Campuses, Custodians, and Programs
Requirements, including budgets, governance and evaluation
Scale over time should match elite global competitors
A self-governing society
The society should be externally evaluated as a collective whole
Challenges & approximate 5-year milestones
Summary - An Ada Lovelace Society
Ada Lovelace dreamt of a ‘poetical science’[5] for a better world, a creative venture at odds with the cold and calculating image of today’s science. Lovelace’s most famous vision saw computers going beyond barren computation, writing that they ‘might act upon other things besides numbers... the Engine might compose elaborate and scientific pieces of music of any degree of complexity or extent'. She foresaw by a century the computer revolution and our technological society. Visions such as these extend to other fields, beyond computing. Lovelace herself also dreamt of a world of human flight, of influencing the mind-body relationship, of enhancing thought itself[6]. Yet we forget that those planes in the sky and medicines in ourselves were once but dreams in the minds of the Lovelaces of the past, and so we forget to nurture the Lovelaces of today.
The Lovelace program will make the United Kingdom the home of today’s Ada Lovelaces, birthing and nurturing their visions of a better world through technology and science. There is a key global need for junior researchers7 to pursue long-term research programs, to self-organise and collaboratively pursue visions of a better future through invention as they could in the postwar period that created our technological world. To achieve this, the Lovelace program will create a new scientific structure in the United Kingdom, complementing the university system. Its design features are inspired by the striking commonalities between the early Cambridge Laboratory for Molecular Biology, Bell labs, and Xerox PARC: three model institutes which birthed new fields of research and industries in the postwar period. These institutes recognised that great science cannot be planned or engineered, but comes from fostering an independent culture honed for collaboration, risk-taking, and creativity between individuals like Ada Lovelace.
The Ada Lovelace Society will have a small number of campuses, each centered on a broad topic encompassing many distinct specific visions and missions, with strong core technical support to avoid the need for individuals to have large laboratories. A key feature of the Ada Lovelace Society is giving long-term elite-level research resources to promising and highly-driven junior researchers (not necessarily young - see footnote definition), who are placed in a flat hierarchy and self-organise into collaborations. Rather than follow the traditional academic career path, these fellows will be subject to renewal at long-term timescales, more akin to a medical or civil service career, freeing them to take substantial risks which are necessary to enable bold discoveries and inventions. Fellows will be recruited based not on prestige, but rather on their promise to produce transformative research and invention over a 15 year time period: a Royal Society for active researchers. They will be mentored, but not controlled, by groups of senior custodians who similarly have strictly enforced team size limits. The senior custodians will also contribute to recruitment, renewal decisions and resource allocation, as in the model institutes. Despite great individual research freedom and long-term support, Lovelace Fellows will not have tenure, and they will be restricted to publishing in non-profit journals, and precluded from applying for external grants. Larger collaborative research programs will also be housed on these campuses, as will many visiting scientists and students. Each campus will begin as a Lovelace School, that will organise courses, conferences, and workshops offering training in the advanced topics that underlie their vision and providing access to top talent for recruitment. These outreach programmes will interact closely with applied research structures similar to the envisioned Faraday Society[8]. Together, Lovelace campuses will function as national incubators of the visions of today’s Ada Lovelaces.
Motivations to diversify beyond academia
The UK is strongly dependent upon academia for research. Yet we believe that solely investing more in universities is not the best path to making the UK a global hub for science and technology. The key is, instead, to achieve more structural diversity. We argue that fundamental features of academia are not well suited to many of the types of research and invention we seek to promote. Contributing factors include:
The strongly hierarchical and static career structure and power distributions within it are optimised for teaching, not for disruptive research. Industrial laboratories such as Lockheed Martin Skunkworks, DeepMind, or Bell Labs do not replicate the academic structure.
The incentives and credit assignment mechanisms in academia focus on short term incremental advances, favouring publication and projects with guaranteed outcomes in the short term, as opposed to long-term research and trial-and-error invention. These incentives are largely set by the global academic culture and both real and perceived academic recruitment policies and will be challenging to fix in the UK only.
It is challenging in academia to bring a critical mass of researchers to one place to work on a common problem or vision over an extended period of time. The pressures of academia and ways credit are allocated promote a soloist mentality; furthermore, the division of universities and funders into departments labelled by historical fields further impedes cross-talk and inter-disciplinary collaboration.
Many types of ‘medium scale’ research that require major technical support, such as whole-brain connection mapping, are not suited to the academic career path.
Increasingly, researchers across fields are now either ‘trainees’ on short-term low pay contracts working for a senior individual, or are senior individuals who are then often effectively forced to become managers, write grants and build empires to survive, rather than engage directly in creative research. For instance, it is not uncommon for one laboratory head to have twenty ‘trainees’ and to be the senior author on their papers, despite little direct involvement, in order to get more grant money[9].
The reliance on inherently risk-averse central bureaucracies for funding decisions produces a homogenising effect upon research endeavours, because researchers are strongly incentivised to write grants they think will appeal to their peers, rather than forging new paths that involve risk.
There is increasing concern about the present culture in academia and its corresponding impact on research integrity and long-term influence[10].
These endemic challenges make it doubtful that an Ada Lovelace, and the teams that together fulfilled her vision, would flourish within current academia. How should we address this?
The key is to diversify and to allow for more local autonomy.
We should first note that the other leading scientific nations have much more diverse research structures:
The US has well-funded national labs[11], and also has tech monopolies which are pursuing some of the world’s most cutting edge innovation freed from normal constraints.
Notably, our own prized research asset, DeepMind, is owned and funded by Google.
Germany has a series of relatively self-governing elite scientific societies. Of particular relevance is the Max Planck society, an elite basic science society, providing long-term career support and operating parallel to academia with core federal funding of approximately €2 billion, and a large amount of external funding:
Germany’s scientific funding societies: self-governing, elite, and lasting
We conclude that the UK should likewise prioritise the creation of parallel scientific structures, at similar scale to those in Germany and the US. Given the lessons of the 20th century’s successes, how should they be structured?
Inspirations: commonalities between Bell labs, Xerox PARC and the Cambridge LMB
Though recent policy attention in the UK has rightly been on the transformative Advanced Research Projects Agency (ARPA), physical institutes were another crucial element of the postwar scientific flourishing that built our world. Attempts to replicate their success should begin with a clear study of the internal structures, power distributions and incentives that produced the research. The three most impactful laboratories of the twentieth century shared striking structural commonalities that are at odds with modern academia, and we lack equivalents of them in our current research environment. These institutes (see figure below) each created whole fields of research. Though we focus on these, other institutes pioneering new fields, such as the Oxford Center for Quantum Computing and Google DeepMind, have similar features.
Three transformative laboratories with striking structural commonalities despite different sizes and topics: Bell labs, Xerox PARC, and the early Cambridge Laboratory for Molecular Biology
These institutes could operate as self-organising systems due to being communities united by a vision. The crucial features were:
‘Visions not goals’[12]. Rather than a collection of disparate departments[13], each of these institutes were physical labs collectively focused on broad visions of the future of technology and science[14]. Visions for each can be found in footnote15.
‘Physical proximity was everything’[16]. The institutes were tight communities which nurtured connections by bringing researchers to work in one place. Transformational advances emerged when a critical mass of driven researchers formed a bottom up ‘living organism’[17] of research and invention.
Talent magnets for elite and diverse talent. They attracted the worlds best researchers from disparate fields; the common vision succeeded in uniting people with different skills, and led the institutes to be known as the ‘place to be’ for a given topic[18]. Individual researchers were funded to pursue their best ideas, but overall headcount was tightly restricted.
They had start-up like power and control distributions, as opposed to hierarchies:
Full internal funding[19] and internal review. Researchers were evaluated internally at periodic intervals by those who knew their work well, not by distant committees[20]. This allowed assessment of them as people not paperwork. There were essentially no applications for external grants, allowing a culture of independence and free thinkers to flourish. Healthy competition between like-minded researchers provides a natural review process also[21].
Directors as custodians, not micromanagers - The leadership job was a ‘service’ to the community. ARPA-IPTO turned over directors every 2 years; Bell labs managers were strictly forbidden from interference in junior research. They were ‘managers who didn’t manage’ - they oversaw that the values of the community were maintained and removed the need for decision by committee[22].
Small labs, flat hierarchy, self-organisation. Groups were restricted to be very small[23]. Projects happened mostly via spontaneous self-organisation rather than being dictated by superiors. Principal investigators worked at the bench.
Strong technical staff scientist support. All institutes had cutting-edge core facilities, accessible to all labs. Labs did not have to grow simply to amass enough funding in order to gain individual access to technology. This required highly valued core facility members providing both technical knowhow and continuity.
Empowerment of the young. Both the LMB and Bell had a culture of letting the young pursue their ideas and take responsibility (‘ideas not reputations’). At PARC’s peak, the oldest researcher was 30.
Incentive and credit assignment structures differed starkly from academia:
Research aimed at the 20-30 year timescale[24]. They believed they were building the foundations of the future and worked accordingly. Each transformed their fields at much shorter timescales, as this perspective let them venture to different parts of the unknown.
Internal hiring was welcomed. Whilst today this is frowned on, Bell and ARPA-PARC were grown via internal promotion[25], encouraging a culture where mentoring was critical to collective success.
Limited ability to grow in size: Not all the best researchers chose to become lab managers/PIs (for example, Shannon never did). The flat hierarchy and small labs meant that focus was on research, not career progression: don’t convert first class researchers to 2nd class administrators.
Credit strictly going to the researcher who did the work: there were no ‘empires’ or postdocs doing the work of others.
Prioritised invention as much as scholarship - ‘Either do something very useful, or very beautiful’ [26]. Creating a new method was seen as equally important as a discovery.
Recognition that ‘failure’ is a necessary overhead for the big, unexpected wins.
Lessons from recent smaller independent institutes
Non-profit funding agencies have recently funded a number of institutes, each of which have been successful and are now international leaders in their field[27]. Excellent researchers are increasingly moving to these institutes rather than staying within universities. However, these institutes could not fully develop their own culture as they were wedded to the academy: ultimately people had to return to the existing academic system. This manifested in several ways:
These independent institutes imported the normal hierarchical laboratory structure - they have not abolished fixed labs, academic hierarchy, or fundamentally changed the incentive structures, though private enterprises have.
Individuals, especially postdocs, are dependent on the traditional academic career structure after leaving. As a result, individuals know themselves to be judged by the external measures of success, most obviously publication, which leads to the institutes, and the researchers within it, competing by metrics and standards set by the external culture.
Relatedly, with exceptions (notably the most successful), they are still dependent on external grant funding, which in turn forces setting of short term objectives and avoidance of high risk/high gain projects.
They rarely promote members from within, meaning a cohesive culture doesn’t grow over time.
They rarely have a cohesive vision of the kind seen at LMB, PARC and Bell Labs, perhaps because the visions were imposed externally from academia rather than grown from within.
Four core features of a new type of national lab structure: Lovelace Fellows, Campuses, Custodians, and Programs
“All we have to do is create opportunities for those who want to take risk. If we start funding this, there will be a long line of young people who are willing to participate, and will release a huge energy which has been so far suppressed. That’s why I’m trying to promote this message.” Garry Kasparov, 13th World Chess Champion, in an interview that helped initiate this proposal.
The Lovelace Program is motivated by a conviction that the hierarchical bureaucratic structures that usually come with human organisations are often antithetical to the self-organising, start-up style dynamics needed for world changing research, and that the disempowerment of youth is at odds with the history of the world-changing research that has come from the young - see comments by Brenner[28], Einstein[29], Newton[30], and Feynman[31]. Human organisations naturally drift in this prestige-based, top-down direction and so creative environments need continual, conscious correction and renewal. We further believe that the notion of what constitutes research in universities is overly constrained, and that blue skies invention without obvious immediate application has historically been a major driver of progress yet is not well funded by public R&D nor rewarded in academia. The name Ada Lovelace Society enshrines this need to look for creative vision and promise, not reputation alone.
Rather than specify elaborate bureaucratic processes to achieve this, those involved with the program should read about the institutes whose cultures and structures serve as the inspirations behind this program (see reading in the footnote[32]). As with an ARPA program vision, individuals within the Lovelace program can then align to those examples. To guide this, there are 4 simple conserved features of the Ada Lovelace Society: the Lovelace Fellows, Lovelace Campuses, Custodians, and Programs.
1) The core feature is the Lovelace Fellowship. This is a ‘semi-tenure’[33] for people of high creative promise in research, taking risky bets on their drive and talent and freeing them to pursue their ideas and inventions collaboratively outside the constraints of the academic career path.
Fellows will be recruited solely based on their promise and vision of doing world class research over the next 15 years. The Lovelace Fellow career path would involve renewal every 7 (+2.5 buffer) years. One can view this as replacing the hierarchical tenure-based academic career path with a renewable, longer-term ‘MacArthur fellowship for small teams’[34].
The Fellows’ mission would be simple:
“Either do something very beautiful, or very useful.”, following Bell labs[35].
Applicants/awardees are preferably to either come straight from graduate school, be refugees from academia, or come from industry, with a preference for those coming through Lovelace schools (below): we are not looking for typical polished CV’s with Nature papers. They could also be senior individuals who wish to return to closer involvement in frontline research. The award will come with support for 1-3 other staff scientist hires[36], world class technician, engineering, equipment and consumable support, and the ability to apply internally for limited additional support as needed. Fellows will therefore be able to pursue their 1-2 most promising ideas with world-leading support and resources. The limitations from being restricted to small group sizes will be compensated by this access to the world’s leading core resources and technical support, along with self-organised larger collaborative efforts between fellows, as in Bell labs. Fellows unsuccessful at renewal will be offered a well-paid staff scientist role within the society, removing the ‘cliff edges’ and zero-sum dynamics. It is expected that roughly 50% of fellows are renewed, with around 30% choosing to leave for start-ups, industry, or academia, and 1 in 5 becoming staff scientists (critically de-risking failure).
2) Each Fellow will be based in a Lovelace Campus[37], focussed upon a very broad topic and containing a large number of smaller visions. As at Bell labs, a given topic such as “systems physiology” could encompass research all the way from methods development to theoretical physics. An example topic could be the study of the interaction between the brain and body (see, for example, ‘The Music of Life by Denis Noble).
One can view each campus as a largely autonomous scientific community with full internal funding and decision making, collectively tasked with being the world’s pioneer in a given broad area of research and invention. The original 3-5 campuses should all be located in a restricted geographic area in the north of England, contributing to leveling up and facilitating interactions between them. Each should be near to a) a major metropolitan area, b) world-class universities, c) an international airport, as well as d) close to an area of natural beauty (Peak district). They should be university adjacent, not university dependent.
The campuses are each like ‘loose archipelagos’ of smaller buildings[38], organised around a large central Bell labs-like ‘hub’ structure including conference facilities, gym, catering, hotel facilities etc, surrounded by a connected network of smaller PARC-like vision-focussed laboratory buildings, some with 200 people, some may have as few as 10. So the campus is a kind of umbrella organisation incubating more specific visions, which can then be scaled up locally. These buildings need not be fancy designs, but rather built up in a Los Alamos/MIT Building 20 way to promote informality and a freewheeling, self-organising spirit - the early LMB was like a ‘tightly packed….warren[39]’.
Corporate-behaviour and emergence of bureaucracy on campus must be strictly avoided[40]: this will require difficult decisions but the hidden costs of such formal processes are evidenced by the current bureaucratised academy. A single director would run each individual sub-building.
These campuses should also be hubs for broader science. This includes the Lovelace schools, which provide courses and discussions in edge-of-the-art technologies, acting as both an idea incubator and as a recruiting stream for new Fellows. They will also provide international academic and industry workshops, conferences, and potentially UKRI-investigator grant renewal talks that would also serve as recruitment processes. For instance, a renewal requirement for UKRI grant holders could be to present at a Lovelace Campus conference every two years to encourage cross-fertilisation between UK researchers. This would also fill a crucial gap for the UK: we lack an equivalent of Cold Spring Harbor Laboratory or Woods hole courses and conferences.
Campuses should have an emphasis on strong core support, and will not rely on employing large numbers of postdocs for labour, therefore having a more sustainable career distribution. A lab of a fellow + 2 staff scientists could conceivably draw on 4-5 technicians, engineers and programmers, for instance. They will also have an unusually large fraction of visiting scientists including students[41].
A more sustainable and efficient distribution of career roles within a Lovelace Campus. Note the lack of postdoc ‘trainees’, whose requirements for academic advancement are at odds with core priorities and values of the society.
3) Custodians are individuals temporarily taking on managerial roles, as at Bell labs: Senior individuals are crucial to the mentoring, recruitment and resource allocation that is central to the scientific ecosystem. Even at PARC, Bob Taylor was a ‘manager who didn’t manage’, providing light touch guidance to the community of researchers. As in model institutes, campuses will have senior individuals in these roles, drawn from a group of more senior fellows to advise the individual Directors on recruitment and resource allocation. Being in a custodian position will not entitle one to greater research resources, rather it is to be considered a service to the community.
4) Lovelace Programs provide crucial scaling of research ventures: The small group structure is expected to be ideal for birthing disruptive ideas[42], but less so for scaling-up programs of research. Further, there is a critical global deficiency of funding for programs in the ‘medium-size’ range - those of size between that of a single lab and CERN. Examples of such programs include whole-brain connection reconstruction, development of cutting edge implantable devices, or large-scale protein screening for bio-engineering. These fall outside the remit of essentially all existing funders. Having these relatively unique medium-scale projects present in a Lovelace campus would not only attract talent and derive economic benefits, but would also help establish the campuses as global hubs in their chosen fields. Programs would be run by a fellow nominated as program manager, along with a group of advisory Fellows, and would be authorised by the director following an application either by Fellows (preferred) or external group. Varying by campus, together the collection of these medium-sized programs could conceivably amount to 25% of a campus budget, potentially in collaboration with UK ARPA. A useful model here would be the Janelia Project Teams or those at the Allen Institute.
Requirements, including budgets, governance and evaluation
“[Licklider] was a "nice guy" (he really was) but he also knew that enterprises like ARPA-IPTO would die if there were attempts at democracy, shared planning. etc. What he set up was "patronage" (as with the Medicis) or "MacArthur Fellowships for groups". He essentially found Principal Investigators who thought they could do something great with the vision, and Lick guessed at which ones to fund [and was willing to "take percentages" on the results]………..This worked well, and every two years there was another IPTO director -- drawn from the community -- who did the very same things.”
Alan Kay writing to us, echoing Perutz & Brenner
It must retain a strict ‘program manager/Senior Responsible Officer[43] (SRO) + advisory board’ decision making ethos throughout[44]. Per Perutz, strictly no committees. This is to be akin to an ‘ARPA for institutes’.
Turnover of positions of power at short timescales.
Internal promotion should be welcomed, especially for positions of power (with leaders and boards selecting their own successors for continuity). This is essential for maintaining culture and vision.
Lovelace must not be made to compromise with ‘stakeholders’. The only stakeholders are the researchers we seek to recruit and the funders/taxpayers who support them. See footnote for advice from a Nobel laureate from outside the university system to this effect[45].
A commitment to an absolute minimum bureaucracy aided by the ARPA director/program manager model: the Defence Agency of Advanced Research Projects Agency (DARPA) is currently a $3 billion organisation with around 125 staff, despite having existed for over 60 years. This comes from a strict culture of full individual autonomy and responsibility.
Scale over time should match elite global competitors
The Ada Lovelace Society must be large enough for the ecosystem and career-path within it to be relatively self-contained and form a geographic ‘talent attractor’. This allows researchers to leave academia for Lovelace without the worry of having to secure hasty, high-profile publications to guarantee them a later job in academia[46].
Lovelace should be of sufficient scale to be a flagship endeavour of UK science and innovation, placing it ‘on the map’ and achieving global recognition to attract talent.
A total budget per evaluation time period should be set, and then the responsibility of funds and resource allocation placed on, and left to, the individual campuses.
A substantial fraction of Lovelace campus budget should be kept unallocated - they must operate with money to spare in order to be flexible and allow for speculative investments when opportunities arise - they must have what Alan Kay calls ‘mad money’, or what Sydney Brenner calls a ‘casino fund’.
To attract the best researchers in the world, Fellows salaries should be tied as a fraction of the industrial market rate, with baseline and ceiling levels. Resource support should match the leading competitor. Priority on less but better throughout.
A self-governing society
It should be a self-governing scientific society funded over 10 year periods as a nonprofit organisation - scientists must be the final decision makers. This is also true to some extent in the Max Planck Society. Attempts to exert control by funders are a typical failure mode of these endeavours, yet government is notorious for seeking such control [47]. Per Alan Kay, “the goodness of the results correlates most strongly with the goodness of the funders”.
A 5 person Governing Board, and a Society Lead Custodian as executive. The campus and laboratory directors provide the input to this board.
A 10-person Fellows Board with a Lead Fellow + two Deputy Lead fellows (see below) to be the ‘voice of the Fellows’ in senior proceedings and ensure the proceedings align to the core values. Having a Lovelace Society governed solely by senior people would be very odd.
A Trustee board: this Trustee board must be composed of individuals who have worked in the environments we seek to recreate, and their task is to report to the funders whether or not the labs are following the spirit of the institutes they seek to replicate. It will also function as an advisory panel.
This structure is to be replicated similarly at the level of each individual campus, which will have a single director for each ’PARC’ on campus.
The society should be externally evaluated as a collective whole
Attempts to produce short-term (5 years) results from individual programs to please funders will curtail long-term success. Instead, the advisory board should provide oversight, along with 10 yearly external reviews of the society and its funding as a whole along with its individual campuses. A 5 year advisory review, available to the custodians, directors, and funders, should also be performed. For evaluation, it is crucial to enshrine the evaluation criteria ahead of time, to ensure that the usual standards of success — which the Lovelace society means to steer away from — are not unintentionally projected by the external assessors (for example criteria, see footnote[48]).
Challenges & approximate 5-year milestones
The model institutes grew organically out of existing processes and communities such as ARPA and the Cavendish, whereas we are setting off from a relative standing start. This poses a challenge: where to find the visions to pursue, and the people to pursue them?
The shift from formal process back to trusting individual judgement places great emphasis on finding the right people with the right motivations.
With this in mind, beginning with open bids[49] for initial visions and missions is intuitively appealing but perilous[50]. This will immediately become subsumed by the existing system, with bids being necessarily being judged by current external standards. Further, a Licklider would not win such an open bid, as evidenced by his difficulty finding patronage even in the 1950s.
The visionary mindset that was core to ARPA-PARC has been strongly selected against in recent times.
There is a trade off between legibility[51] and success - attempts to clearly plan such a program will detract from its ability to operate flexibly, self-organise, and ultimately find the right path for itself. If you can write a summary of a research program that a central bureaucracy can understand and approve, it is necessarily not an emerging or disruptive research proposal.
These points converge to argue that the Lovelace Program must be grown, not overly pre-planned, beginning with broad topics and only later focussing on more specific PARC-like visions which grow and are scaled internally.
One main strategy to find the best and brightest Lovelace Fellows will be to begin by hosting world-class courses in the most advanced topics. Training in the “edge-of-the-art”[52] is currently massively underserved by existing international programmes, which focus on introducing the core fundamentals. Therefore, each campus will run a Lovelace School for active researchers that will host and coordinate training for those at the bleeding-edge of knowledge. These schools will attract talent and serve as the engine of the Lovelace Society.
Growing a ‘loose archipelago’ of PARCs from a central Bell labs like structure and initial Lovelace school, all connected by walkways in a ‘non-raincoat needed’ rule, with food only served in the central hub.
Endnotes
1. Next version to include diagrams of things such as Fellow recruitment/career cycle, and an example campus plan.
2. https://www.theguardian.com/news/2002/feb/07/guardianobituaries.obituaries
3. Max Perutz: “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.”
4. ‘I think the job of research, with very few exceptions, is really the job for young people’ — Sydney Brenner [youtube link]
5. ‘You will not concede me Philosophical poetry. Invert the order! Will you give me poetical philosophy, poetical science?’ — Ada Lovelace. P96 Hollings et al
6. P59, Hollings et al. ‘Ada Lovelace: the making of a computer scientist’
7. Junior research definition: Though the preference is to hire straight from graduate school, junior does not necessarily correspond to age - it refers to researchers who remain actively involved in research and do not employ large numbers of researchers beneath them. Eric Betzig and David Deutsch, for example, could be considered Junior researchers by this standard despite having both created new fields of research and one of them having a Nobel Prize.
8. A scaling of the UK catapult network.
9. Nobel Prizes are now being given to senior individuals for work done almost entirely by their young ‘trainees’, including the ideas and theoretical underpinning. This is a worrying breakdown in credit assignment: it is not entirely new but the growth of thirty person labs has made it very much worse, as it means the best way to succeed is to build an empire.
11. The US Argonne national laboratory campus, for example, has a budget of $750 million
12. The quotation is from Licklider.
13. Currently, research programs are distributed across academics in different universities and departments, with each department focussing on myriad different topics. It is difficult to gather large groups of like-minded researchers together in university departments due to the career dynamics and laboratory structure, a problem JCR Licklider himself encountered at MIT in the 1950s working on ideas that led to Man:Computer symbiosis.
14. ‘[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.
15. Vision statements of the institutes included: Bell Labs: ‘One policy, one system, universal service’ for communications. LMB was centered on a view that the understanding of cellular life would come from understanding interactions between molecular structures, building the foundations of modern medicine. PARC: ‘Computers are destined to become interactive intellectual amplifiers for everyone in the world universally networked worldwide’
16. Gertner, ‘The Idea Factory’, page 151. See also ‘how buildings learn’ by Stewart Brand
17. Bell labs President Mervyn Kelly’s description of Bell labs.
18. “In the budding field of molecular biology, his [Perutz’s] operation became known as the place to be” — John Sulston, who shared the Nobel Prize in physiology or medicine with Brenner and H. Robert Horvitz. (Pennisi 2003)
19. We note that these were not funded by private competitive companies, but were either government funded or were funded effectively by monopoly companies.
20. ‘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.’ — Brenner
21. Unix Pioneer Brian Kernighan ‘Given an environment like that where everybody is better than you, you don’t slack, you try to keep up with them.’
22. Until Perutz retired in 1979, it [LMB] had no director. Perutz did not himself use the title of director at the LMB. “Perutz didn’t want to be one, and it meant he could retain his lab space after retirement. Instead, labs had a loose management committee, which met occasionally and saw its main job as attracting outstanding talent to the lab.” — Bynum 2012
23. At the LMB, formally a group did not exceed around 6 people. At Bell it was 1 person + a technical support member or two. PARC had no separate individual labs in the 1970s.
24. 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’.
25. “It is difficult to sum up ARPA-PARC, but one interesting perspective on this kind of funding was that it was both long range and stratospherically visionary, and part of the vision was that good results included “better problems” (i.e. “problem finding” was highly valued and funded well) and 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”” — Alan Kay
26. A sign at Bell. H/T to Alan Kay for pointing this out to us.
27. For example: the Champalimaud Center of the Unknown, the Sainsbury Wellcome Center, Janelia Research Campus, the Crick Institute, the CZI BioHub.
28. “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…...Today the Americans have developed a new culture in science based on the slavery of graduate students.” — Brenner
29. “A person who has not made his great contribution to science before the age of 30 will never do so.” — Einstein
30. “All this [discovery] was in the two plague years of 1665-1666 [age 23]. For in those days I was in the prime of my age for invention & minded Mathematics & Philosophy more than at any time since.” — Newton
31. “Science is the Belief in the Ignorance of the Experts…….science - a.k.a. research - is in the making, belongs to the (unknown, yet to be discovered) future, while expertise is based on the past, with in-built obsolescence.” — Feynman
32. The Dream Machine — Waldrop [best book on ARPA-PARC]
33. Rather than a binary tenure decision at age 40, proceeded by many short-term trainee contracts, you instead have 3-4 renewal cycles in the career, with ‘light touch’ monitoring in between for example via internal talks.
34. One could envisage that, rather than 1 fellow employing 2 other staff scientists, some people may wish to do joint applications, so 3 Fellows recruited as one. This would preferably be for those who have worked together in the past. There is a complete global lack of this type of team support internationally.
35. “A visit to the classic Bell Labs of its heyday would reveal many things. One of the simplest was a sign posted randomly around: “Either do something very useful, or very beautiful”. Funders today won’t fund the second at all, and are afraid to fund at the risk level needed for the first.” — Alan Kay
36. As desired and depending upon field and performance of the fellows in internal talks.
37. They can move freely between campuses, but have a residency requirement to be at a campus for a given proportion of their Fellowship.
38. There are some different ways to do this, this is illustrative of one approach. But whatever approach is gone for, putting a critical mass of them in a small space is crucial.
39. “In the LMB, we were….tightly packed….that may be just a romantic sentiment, but I feel that when you are living in more of a warren, you may be more productive” — Sydney Brenner
40. There are no ‘people and culture’ directors, ‘HR business associates’, or other symptoms of an emerging bureaucracy in which scientists are disempowered.
41. Note that students are real scientists.
42. Large teams develop and small teams disrupt science and technology, 2018, Wu et al Nature
43. In effect, there must be one single decision maker on all things, not consensus.
44. Of course this risks some ‘abuses of power’, but this is far better than the rampant system gaming, politics, dilution of vision and stagnation that occurs otherwise.
45. From one of the world’s top scientists and technologists who invented a whole new field of research: “The most important piece of advice I can give is to not compromise. Putting together an initiative that includes consensus with a bunch of existing UK supported "stakeholders" is bound to be watered down and ineffective….. The pipeline from academia to industry is well known, and is arguably the most important function of academia. The reverse pipeline is nearly non-existent, and consists of only a few oddballs like me……. It is my belief that industry is teeming with passionate, talented scientists who avoided academia due to its political correctness, low pay, and screwed up incentives favoring grants and papers over scientific output. I further believe that some of these scientists have developed ideas they feel passionately about, but are unable to execute within their current organizations. There is an opportunity to tap into this vast well of creativity, determination, and focus to revitalize both science and industry. I recommend you start by asking industry leaders or leading academics with close industry ties about how to identify and incentivize industry scientists to join an organization where they can pursue their best ideas….Do not expect effect change through any fraction of your “billions a year” allocated to current academics in their current institutions….. If you cannot find enough people to seed a new institute with a broad charter to pursue science as they see fit, you’ve misjudged the opportunity to transform UK research.”
46. This has been a common critical weakness of existing attempts to replicate these institutes, as it locks them into the external culture.
47. "As you know from looking at the past, most of the attempts to do something like ARPA-Parc ran afoul of a variety of human problems involving people trying to control instead of nurture, spreading the money around instead of concentrating on talent, trying to set problems rather than fund "problem finding", and many more ways people have found to mess up good ideas and processes. "Alvey" here in the UK and "Esprit" in Europe were just two examples of doing this badly -- and of course there have been a few in the US, including "Interval" which also messed up.” — Alan Kay writing to us.
48. Is Lovelace doing things differently, or has it reverted back to competing with the existing system? Is it aligned to the core vision of empowering junior individuals, or has it developed an entrenched ‘oligarchy’ of senior scientists hoarding opportunities based on past success? Is it a destination of choice internationally? Has it managed to successfully use ARPA-like internal promotion without succumbing to nepotism? Ultimately, has it produced research that would not occur in a UK University? And the crucial assessment: considering Lovelace has only been running X years, was investing in this diverse approach better overall than putting the additional money into more established university research? If not Lovelace could be converted into a series of technical universities.
49. This is not the same as not openly consulting - the point is that the leadership and the broad topic proposal must not be tied together. A ‘call for ideas’ would be fine, and will come as part of the call of Lovelace schools.
50. Open bids for institute research topic areas were recently cancelled by a leading medical charity.
51. https://www.ribbonfarm.com/2010/07/26/a-big-little-idea-called-legibility/
52. quote from Alan Kay
I really, really enjoyed reading this! It’s very inspiring.
What sort of OS will the Lovelace Campus run on?