#6. Science as Cities or Science as Corporations
Why are research teams less innovative the bigger they become, whereas the opposite is true of cities?
Each city has its own unique feel, you can sense it when you walk around one.
“In a hundred subtle ways, the city sends you a message…New York tells you, above all: you should make more money…You should be hipper…You should be better looking…You should be richer. What I like about Boston (or rather Cambridge) is that the message there is: you should be smarter. You really should get around to reading all those books you've been meaning to.”1
“Paris is a woman but London is an independent man puffing his pipe in a pub.”2
Cities feel unique because they are made up of people and people are themselves unique. When these individuals are scaled up into clusters, communities and networks, it is these networks that gives cities their unique feel. In the words of the historian Fernand Braudel, they are “like electric transformers. They increase tension, accelerate the rhythm of exchange and constantly recharge human life.”3
In some cases, people travel far and wide to experience these, or to meet and collaborate with like-minded people.4 In Coriolanus, when Sicinius asked “What is the city but the people?”, the Plebeians replied, “True, the people are the city.”5
In cities, each individual needn’t possess extraordinary capabilities, because when agglomerated with others, they can combine into something much more advanced.
Think of ants. Each ant is stupid, yet together they can build complex colonies far more advanced than any individual ant could possibly imagine.6 It is by bringing people together that cities develop; physically manifesting our social interactions.
On the surface, cities and animals don’t have much in common, but the physicist and allometrist, Geoffrey West argues they do: they both metabolise energy and resources, produce waste, process information, grow, adapt and evolve, and even develop what could be characterised as tumours and growths.
Regardless of the city, most people also want to travel from A to B in the shortest possible time at the cheapest cost; as do biological organisms. We can see this most clearly in research carried out by Hokkaido University researchers who compared the innate urban planning ability of a single-celled slime mold (Physarum polycephalum) with Tokyo’s rail planners, and noticed how similarly they performed.7
Kleiber’s Law tells us that irrespective of the animal, its metabolic rate scales 3/4 to the power of its mass. This means that as animals get bigger, they need less energy per pound of flesh. What about cities, though? How do they scale?
In the spirit of Kant, Geoffrey West set out to discover whether cities can be mathematically governed.8 That is, whether there are any fundamental laws that can help us to understand cities and how they scale: to understand whether cities are scaled-up organisms.
West’s research shows that cities scale superlinearly- meaning that if a city doubles in size, every measure of its socioeconomic activity more than doubles as a result. Interestingly, this was true regardless of whether it was construction spending, violent crimes, the amount of bank deposits, traffic, wages, or even AIDS cases, that were being measured.9
Just as Kleiber’s Law applies to animals regardless of their evolution or habitat location, West’s laws allow us to understand faraway places irrespective of their location or population size.10
Superlinear scaling is rare in the animal kingdom; other animals tend to get slower as they get bigger, but in cities the opposite happens. As cities get bigger, everything starts accelerating and they start innovating more. There is no equivalent to this in nature.
In 1926, J. B. S. Haldane published an essay in which he claimed that:
“Just as there is a best size for every animal, so the same is true for every human institution.”11
Whilst the optimal size for Greek democracies was that of a small city, recent research has shown governments’ cabinet size negatively correlates with effectiveness, political stability, accountability, life expectancy, and standard of living.12
In a similar vein, Geoffrey West and Luis Bettencourt set out to see if the same was true of corporations- that is, whether they scale the same way as cities—getting more productive as they grow—or, whether they’re more like Greek democracies and governments, and become less productive.
After examining more than 23,000 publicly traded companies, they found that corporate productivity—unlike urban productivity—scales sublinearly, meaning that as the number of employees in a corporation grows, the amount of profit per employee shrinks.13
Why is it then that when cities grow in size, they become more productive, but the opposite is true of corporations, governments and democracies?14
Cities are unique. They can be decimated by (nuclear) bombs, hurricanes, and earthquakes, yet they rarely ever die. Corporations on the other hand die every day irrespective of their size, location, or market share.15 Cities can innovate faster than the problems created indefinitely, whereas corporations can’t.
This can partly be explained by their structure. Companies are managed in a top-down fashion by a team of executives, whereas cities are unruly places, largely immune to the desires of politicians or planners. Cities benefit from spontaneous order.
“Think about how powerless a mayor is. […] They can’t tell people where to live or what to do or who to talk to. Cities can’t be managed, and that’s what keeps them so vibrant. They’re just these insane masses of people, bumping into each other and maybe sharing an idea or two. It’s the freedom of the city that keeps it alive.”
It is this that makes them hubs of innovation.16 This then poses the question: are cities unique in their ability to harness the benefits of agglomeration, or can we apply these lessons to other settings as well?
In a time when scientific innovation is seeing diminishing returns, is it time to have a look at how research institutes operate and rethink their hierarchical structure in favour of more spontaneous order?17 Is it time to learn how cities operate and apply it to research?
In 2019, three researchers, (Lingfei Wu, Dashun Wang and James Evans), analysed more than 65 million papers, patents and software products from the years 1954 to 2014 in order to discover how research output corresponds with team size: to determine whether research teams scale.
Their conclusion: whilst, “large teams develop science and technology, small teams disrupt it. [There is] little evidence…that larger teams are optimized for knowledge discovery and technological invention.” It is important to stress here that ‘disruption’ is a good thing, it is analogous to paradigm-shifting.
Their research found that large teams were more risk-averse. Whereas, small teams—because they had more to gain— were more inclined to seize untested and riskier opportunities that had a higher chance of either failure or high growth. This was true across all eras and 90% of disciplines.18
Interestingly, they found that solo authors were just as likely to produce high-impact papers (in the top 5% of citations) as teams with five members.19 However, solo-authored papers were 72% more likely to be highly disruptive.20
In contrast, ten-person teams were more likely to develop existing ideas already prominent in the system, rather than opening up new frontiers.
Wu, Wang and Evans’ research concluded that disruption and impact consistently diverge as teams grow in size. Which begs the question: why?
Just as West acknowledged that bureaucracy leads to the downfall of companies, the same is also true in research environments.21 The bigger they grow, the more bureaucratic they tend to become.22
The 1962 Nobel Prize winner, and founder of the Laboratory of Molecular Biology, Max Perutz, famously said:23
“Creativity in science, as in the arts, cannot be organised. It arises spontaneously from individual talent. Well-run laboratories can foster it, but hierarchical organisation, inflexible, bureaucratic rules, and mounds of futile paperwork can kill it. Discoveries cannot be planned; they pop up, like Puck, in unexpected corners.”
This is remarkably similar to both the director of research at Bell Labs, Harold Arnold, who said that “invention is not to be scheduled nor coerced,” and J. C. R. Licklider—one of the most important figures in the development of modern computers— who noted that “It [i.e. PARC] was more than just a collection of bright people. It was a thing that organized itself into a community so that there was some competition and some cooperation.” At Bell Labs “the supervisor was authorised 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 underlings.”24
This perhaps helps to explain why some of the best research institutes (such as Bell Labs, the Laboratory of Molecular Biology, and PARC) were so productive: they were organised non-hierarchically and promoted serendipitous interactions. They were more akin to cities than corporations.
“To understand why early cities thrived, look not to the temples of kings but to their subjects’ bustling neighbourhoods.”25
A Few Further Questions:
At the moment, the UK university infrastructure is programmed to commercialise spinouts and tech transfers. What might the impact of this be when they scale?
In the past, corporations often had in-house R&D departments. Today, less so- innovation has been outsourced. What is the impact of this in relation to scaling?
As the burden of knowledge is increasing, we are seeing an increase in the size of research teams. In the coming years though, it is possible team size might diminish due to the development of Large Language Models. Could LLMs help us to open up new frontiers, or help researchers expand or deepen their knowledge by reducing team size?
Further Reading
Eureka! On the clustering of geniuses by Rohit Krishnan
Death and Life of Great American Cities by Jane Jacobs
Democracy is the solution to vetocracy by Sam Bowman
A Review of Geoffrey West's 'Scale' by Blair Fix
This lecture by Geoffrey West, followed by a brief interview with Stewart Brand.
Emergent Urbanism by Mathieu Helie
To learn more about emergent cities and how spontaneity can benefit societies through bottom-up organisation, I recommend reading this article by Noah Smith: Secrets of Japanese urbanism.
Source: Paul Graham, Cities and Ambition.
Source: Jack Kerouac, Lonesome Traveler.
“The great metropolises of the word facilitate human interaction, creating that indefinable buzz and soul that is the well-spring of its innovation and excitement and a major contributor to its resilience and successes, economically and socially.” [Source: Geoffrey West, Scale: The Universal Laws of Life and Death in Organisms, Cities and Companies, p268]
Cities provide a pull factor for migration, attracting the best and the brightest. Suppose you’re smart and ambitious and want to go into a particular sector, you know to move to a particular city. You know that there you’ll have the best chance of being part of a group of a few select smart, ambitious people who want to do something together. You then influence these other ambitious people with your actions, just as they influence you.
In some instances, this gives certain places a comparative advantage as the areas specialise. This is something that Alfred Marshall noted:
“When an industry has thus chosen a locality for itself, it is likely to stay there long: so great are the advantages which people following the same skilled trade get from near neighbourhood to one another. The mysteries of the trade become no mysteries; but are as it were in the air, and children learn many of them unconsciously.”
This Works in Progress Article gives an indication on why we see certain areas specialising in particular fields:
“In the past, clusters would frequently populate around a specific natural resource that was necessary for the operation of the tech in question, i.e. natural harbors, coal and iron deposits, fast-flowing streams. But the ultimate resources for technology development today are the human minds that can come up with and then implement new ideas. So tech clusters today are mostly about attracting and organizing talented humans in cities they would actually want to live in.”
Recommended: The Next Einstein Could Be From Anywhere: Why Developing Country Growth Matters for Progress
NB: Italics added. Source: Act 3, scene 1
See: Emergence
Recommended: Emerging the City.
Learning and applying lessons from natural organisms is known as biomimicry. There are many interesting examples of this, particularly in vehicle design that I would recommend reading about.
“In any special doctrine of nature there can be only as much proper science as there is mathematics therein”- Kant
On top of this, cities also become more efficient as they scale, as they save on infrastructure costs. This is known as economies of scale and means that large cities have a smaller (per capita) carbon footprint.
Interestingly, archaeologists have also applied scaling laws to better understand ancient settlements.
See: Settlement scaling and increasing returns in an ancient society, Settlement scaling theory: Bridging the study of ancient and contemporary urban systems, and The Social Reactors Project
Here is an excerpt from this article:
“Using archaeological data, our group has identified superlinear scaling of wealth (house size) with population for many cases. That is, as the size of a settlement increases, the amount of wealth per capita (using house size as a measure of household wealth) goes up. Just as big cities today have higher GDP per capita than smaller cities, large settlements in the past had higher household wealth per capita than smaller settlements. Population density and energised crowding lead to greater wealth creation in larger cities, compared with smaller settlements, in the past and in the present.”
Source: On Being the Right Size.
One slight criticism of this research is that it looks at publicly-traded companies only. It would be interesting to see how/if scaling laws of private companies differ.
Is there something socially ingrained into us into thinking that as something grows, the apparent need to control it grows, perhaps out of fear that it might metastasize?
See: Parkinson’s Law
The modern corporation has an average life span of 40 to 50 years.
Source: World Economic Forum
Here, West used number of patents as a proxy for innovation. This is controversial as the correlation between patenting and innovation is much attested. Matt Clancy has written a very good summary of the literature on this, here.
James W. Philipps has written a lot about the problems of top-down planning in science and research. Here are some of his notes on the topic, from this working paper:
“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).”
“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). 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 law). 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) 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’ - unis are incentivised to purport ‘world-leading research’ metrics.”
The only consistent exceptions were observed for engineering and computer science, in which conference proceedings rather than journal articles are the publishing norm.
“Experimental and observational research on groups reveals that individuals in large groups think and act differently—they generate fewer ideas, recall less learned information, reject external perspectives more often and tend to neutralize each other’s viewpoints.” [Source: Wu, Wang and Evans, 2019]
“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
Measuring research impact by citation count is flawed. Bhattacharya and Packalen (2020) suggest the incentives created by citation in academia have increasingly led scientists to focus on incremental science, rather than potential (risky) breakthroughs.
Geoffrey West:
“When a company starts out, it’s all about the new idea […] And then, if the company gets lucky, the idea takes off. Everybody is happy and rich. But then management starts worrying about the bottom line, and so all these people are hired to keep track of the paper clips. This is the beginning of the end.”
On top of this, incentives can also change:
“As a group grows, the balance of incentives shift from encouraging individuals to focus on collective goals to encouraging a focus on careers and promotion. When the size of the group exceeds a critical threshold, career interests triumph.”- Safi Bahcall, Loonshots, (p186)
NB: The LMB was a highly impactful research lab- with fourteen researchers receiving Nobel Prizes.
Source: The Idea Factory cited by James W. Phillips
Source: Energised crowding