I’ve been putting together a progress studies reading list for myself, drawing upon reading lists found on the internet. I’ve also seen people repeatedly asking for reading lists and being pointed in different directions. So, I thought it may be helpful to some to collect the various progress studies (related) reading lists I’ve come across thatI’m using for my own reading.
Quick caveat, the progress studies field is quite diverse consisting of people working on different fields: growth, economic history, innovation etc. Thus, I’ve kept the scope fairly broad. Also, I imagine there’s substantial overlap between many of the lists.
Chad Jones (Professor of Economics at Stanford) has some of his teaching materials online, with a focus on economic growth. An intermediate class on growth, and for the more adventurous, a PhD class on growth, with a focus on analysing papers at the frontier of economic growth. *this link is temporarily down.
Before going further, briefly pause to ask yourself, how many people were killed by Smallpox?
The answer, is approximately 500 million.
On December 9th, 1979, Smallpox was eradicated. Celebrating Smallpox eradication day is a great reminder of human ingenuity (although, some celebrate it in May). It’s also a reminder that often people who do a tremendous amount of good in the world can go completely unrecognised.
“This might not seem like a great achievement now, but at the time it was. At the time no disease had ever been eradicated before, and people argued that doing so was therefore impossible. How could man eradicate disease? How could the bureaucratic problems of coordinating the actions of 100 or more countries in the world be overcome?”
The motion was passed by the World Health Assembly by only two votes.
We are beginning to recognise the achievements of many previously unacknowledged individuals who played a huge role, in one of humanities biggest successes. Today, The Future of Life Institute, is giving William Foege and Viktor Zhdanov the 2020 future of life award in honour of their achievements. At noon EST, Anthony Fauci, Bill Gates, and Jennifer Douda will present the award. Details available here.
Another new Progress Studies related book was recently published, “Open: The Story of Human Progress” by Johan Norberg. Quick caveat, the book has been released in the UK and other European countries. However, it’s only coming out in the US on the 15th November.
Do we need more ‘yellow-beret’s? To escape being drafted for the Vietnam War, “Between 1955 and 1973, almost 3,000 medical school graduates enrolled in the program”. From this pool of people, “Nine physicians who trained at the NIH during this period went on to win Nobel Prizes.”
“Does economic history point towards a singularity? “Over the next several centuries, is the economic growth rate likely to remain steady, radically increase, or decline back toward zero? This question has some bearing on almost every long-run challenge facing the world … Ultimately, I found very little empirical support for the Hyperbolic Growth Hypothesis.”
John Myers on the hard question in Progress Studies. Myers concludes “how do we in practice actually engineer that a government fixes these policies and laws? There is astonishingly little focus on that hard question in developed countries, when it may give a higher return on effort than almost any other human endeavor. I believe it should be a core focus of Progress Studies and would like to work with others to summarize the state of the art and identify areas for future research.”
They begin the book with an interesting poll conducted by YouGov in 2016 spanning 17 countries. I’ll pose the question asked in the poll to readers here:
Only a meagre 11% of people responded with “things are getting better”. In the US, it was even worse at only 6%.
Bailey and Tupy posit a few reasons for why many individuals feel things are getting worse:
There’s an asymmetry between positive and negative experiences. Negative events impact us more than positive events. The authors suggest that the media often think along the lines of, “News is bad news; steady progress is not news.” Because many of us follow the news – and the news tends to dwell on negative events – we often think that the world is far worse than what it actually is. In 1973, Kahneman and Tversky identified a cognitive bias they called the “availability bias”. Therefore, we have a tendency to think that the examples that come readily to mind are much more representative than what is actually the case. Because of this, the authors suggest that focusing on the news creates a bias towards being overly pessimistic about progress.
Bailey and Tupy suggest that humans’ over-emphasis on negative trends may be due to evolutionary psychology, “A Stone Age man hears a rustle in the grass. Is it the wind or a lion? If he assumes it’s the wind and the rustling turns out to be a lion, then he’s not an ancestor. We are the descendants of the worried folks who tended to assume that all rustles in the grass were dangerous predators and not the wind.” Humans developed to be cautious, instinctively focusing on potential negative events. Despite this, “the upshot is that we are again often misled into thinking that the world is worse than it is.”
Thirdly, we underestimate the progress (of humanity) because as we make progress, our attention is captured by newer problems, rather than the progress we have made so far. Daniel Gilbert and his colleagues suggest, “When problems become rare, we count more things as problems. Our studies suggest that when the world gets better, we become harsher critics of it, and this can cause us to mistakenly conclude that it hasn’t actually gotten better at all. Progress, it seems, tends to mask itself.” [emphasis mine].
“I see all this progress, and it fills me with conviction and hope that further progress is possible. This is not optimistic. It is having a clear and reasonable idea about how things are. It is having a worldview that is constructive and useful.”
With this Rosling spirit in mind, the authors add to the burgeoning literature documenting the extraordinary progress humanity has seen across multiple domains over the last few hundred years.
The book covers 78 progress trends. Each trend is only a few pages long, with a figure at the end. The book covers progress in multiple domains, split up into the sections covering, “Top 10 trends, people trends, health trends, violence trends, work trends, natural resource trends, farm trends, tech trends, and US trends.” Thus, it really paints a holistic picture of progress.
I won’t spoil the book but a few of the trends I particularly enjoyed (for different reasons) were:
Trend 28 “Vaccines are saving lives”: “In the 20th century alone, the disease [smallpox] is thought to have killed between 300 million and 500 million people.” It has now been eradicated.
Trend 61 “Lighting costs near nothing now”: This is a famous paper I really like by William Nordhaus. The price of lighting has dramatically plummeted: “our Paleolithic ancestors labored 58 hours, mostly gathering wood, to “buy” 1,000 lumen-hours of light… In 1992, 1,000 lumen-hours required 0.00012 hours of human labor.”
To summarise, I think this book may be worth a quick skim for readers of this blog (it can be read in a single sitting). However, readers of this blog probably don’t need much convincing about the dramatic progress humanity has experienced over the last few hundred years. It may be better as a gift to pessimistic friends, who would be hard-pressed not to accept the vast amounts of progress humanity has made in the last couple centuries. I would guess that there are many books like this to come in the near future.
I think there are issues with the poll. I would guess that we would see more positive responses if the question was reframed to something along the lines of, “All things considered, do you think the world is getting better or worse, or neither getting better nor worse, over the last 200 years?“.
I think it’s important to note that it’s possible to simultaneously espouse both that humanity has made tremendous progress in the last few hundred years, and also be worried about the progress made in recent years. Reminding people of the tremendous amount of progress made in the past might make people more optimistic about progress in the near future.
“Since 1990, it is estimated that 420 million hectares of forest have been lost through conversion to other land uses, although the rate of deforestation has decreased over the past three decades.”
Technically this statement is true. A lot of forests have reduced in size. However, a letter published in Nature (ungated link here), “Global land change from 1982 to 2016” by Song et al. (2018), shows that the number of trees have increased since 1982. Song and coauthors use global satellite imaging data to investigate this question alongside others.
The authors explain:
“A global net gain in tree canopy contradicts current understanding of long-term forest area change; the Food and Agriculture Organization of the United Nations (FAO) reported a net forest loss between 1990 and 2015. However, our gross tree canopy loss estimate (−1.33 million square kilometres, −4.2%) agrees in magnitude with the FAO’s estimate of net forest area change (−1.29 million square kilometres, −3%), despite differences in the time period covered and definition of forest.”
Therefore, there was an overall net gain. There was a net loss in the tropics, but there was a larger net gain in the subtropical, temperate, and boreal climate zones.
Observant readers will notice that I have only used the word trees so far, I haven’t mentioned the word forests. This is because tree coverage does not necessarily correspond to forest coverage. For example:
‘Cutting down a 100-hectare tract of primary forest and replacing it with a 100-hectare palm plantation will show up in the data as no net change in forest cover: the 100-hectare loss is perfectly offset by the 100-hectare gain in tree cover. Yet, that activity would be counted as “deforestation” by FAO. Therefore tree cover loss does not directly translate to “deforestation” in all cases.’
I should stress here, in the example above, despite the effects ‘cancelling each other out’, it doesn’t take into consideration the effects on biodiversity or other negative ramifications.
Nevertheless, I found this paper interesting. My original prior when confronted with this question was massively wrong. I thought that the number of trees in the world was decreasing overall. It seems likely that following the media coverage of deforestation shaped my opinion on this particular fact. Thus, it was refreshing to have a positive picture painted in this (very specific!) domain. We’re making some progress on this front, an area often thought to be plagued by a complete lack of progress. Overall, the paper paints a negative picture (I didn’t cover all of the findings), however, sometimes it’s good to dwell on the positives, even if in the grand scheme of things, it’s pretty small.
Patents were originally created to incentivise innovation. However, they also have a lot of downside implications that may deter innovation:
– They don’t always provide the best incentives for original research because inventors cannot fully capture the consumer surplus available in the market. – Inventors don’t receive the benefits from spillovers to other new ideas. – Patents lead to distortions in the areas in which companies innovate. This is because it may not make economic sense to research areas where a lot of patents exist already. – Firms may engage in wasteful spending on things like the reverse-engineering of competitors’ patents. Thus, the problems associated with patents may hinder innovation and progress (section 1 describes these drawbacks in more detail).
Kremer suggests a new mechanism, the ‘patent buyout’. The government steps in and buys patents, which are then freely distributed to the public, who can enjoy the benefits associated with the patent. Additionally, companies are now free to make improvements upon the original patent because they are no longer constrained by having to negotiate with the original patent holder to get rights/access to it (go straight to section 2 if interested in a description of the patent buyout mechanism).
Section 3 discusses the patent buyout in action, briefly describing Daguerreotype photography process.
What are the problems with patents?
In the absence of a patent system, the incentives for research would be substantially lower because anything a firm creates could be appropriated by its competitors. Fear of this happening, and the lower financial remuneration associated with this, reduces the incentives for original research. Thus, to prevent this from occurring, patents bestow a temporary monopoly (approximately 20 years) to a firm that has a new idea. This enables the patent-holder to profit from the new idea, increasing the incentive to innovate.
However, patents can also stifle innovation. In the last five or so years, we have seen a proliferation of 3D printers. They now have more commercial applications but also 3D printing is now affordable to a lay-person who finds 3D printing interesting.
However, this could have happened much earlier. Innovation was stifled due to the patents that restricted other companies from entering the market. Once these patents expired, the bottleneck was removed. Prices of 3D printers plummeted because of new entrants into the market. Additionally, these new companies improved upon the existing technology, making the products better. The patent buyout mechanism (described in the next section) could have brought forward this this surge of innovation.
In an 1851 editorial, The Economist wrote that granting patents, “inflames cupidity, excites fraud, stimulates men to run after schemes that may enable them to levy a tax on the public, begets disputes and quarrels betwixt inventors, provokes endless lawsuits, bestows rewards on the wrong persons, makes men ruin themselves for the sake of getting the privilege of a patent.” This is perhaps an exaggeration but there are indeed a number of problems with patents:
Some consumers cannot access the good/service because the product that is patented is charged at the monopoly price. This means that some consumers can’t benefit from the product, despite willing to pay above the cost price of production (but not at the monopoly price). For example, AZT (azidothymidine) is a drug used to prevent mother-to-child spread of HIV/AIDS during birth. There are potentially millions of cases of mother-to-child infection in developing countries, where individuals/governments/NGOs may have been prepared to pay above the cost price of the drug but not the monopoly price.
Patents don’t allow the patent-holder to capture a large amount of the consumer surplus that their idea generates, which leads to lower incentives for original research. For example, Michael Milken (founder of Prostate Cancer Foundation) would presumably pay hundreds of millions of dollars for a drug that was effective in tackling prostate cancer but pharmaceutical companies don’t take this into account since they would not be able to extract this value from Milken.
The empirical evidence suggests that new research usually creates positive externalities for other research. However, patents don’t reward innovators for these positive spillovers. Without taking these externalities into account, patents lead to lower incentives for original research.
Patents may distort the direction of research from firms because firms are incentivised to work on areas where there are less patents restricting their innovation, rather than areas where patents already exist. Kremer explains that this has happened in the past:
“For example, the development of the high pressure steam engine was blocked by Watt’s patent covering all steam engines; Watt’s steam engine was blocked by a previous patent until he found a way to invent around it; and Edison’s improved version of the telegraph was blocked by Bell’s prior patent for many years [Mokyr, 1990].” (I may write a post on this!).
Finally, patents also lead to wasteful spending because firms waste resources reverse engineering patents.
2. Patent buyouts
Kremer’s idea is simple. The government steps in, buys the patent, and destroys it. Now anyone (consumers or firms) can access it and build upon it.
However, we need to know how much the government has to pay for the price of that patent. Kremer suggests a mechanism that is used to determine the price the government pays.
First, patents are submitted by entrepreneurs to an auction. Then firms bid on this auction, revealing their valuation of the patent. Once the bidding is complete, the government offers to buy the patents at the winning price plus a markup. I won’t talk about how exactly to determine this markup (it should be the difference between the social and private value of inventions), but for now let’s say the mark-up is 10%. Thus, in our example, the government would pay the price determined by the auction plus a 10% premium. Once this process is complete, entrepreneurs owning the patent get to choose whether to accept or reject the governments offer. If the offer is too low, entrepreneurs maintain the right to reject the deal.
Couldn’t firms simply bid extremely high prices? How do we incentivise firms to give truthful valuations of their bids? To avoid this, the government randomly selects some bids that would be sold to the next highest bidder. So, let’s say 20% of the patents are sold to the next highest bidder. The other 80% of patents would be bought by the government and made available to the public. Thus, if a firm goes wild and bids extremely high prices, they would have to pay above market price for their poor bidding strategy. So, firms are incentivised to reveal their true valuations of the patents, otherwise they’ll be punished financially.
This mechanism isn’t calling for a complete abolition of the patent system. Indeed within this system, the patents not bought by governments are sold to firms. Thus, patent buyouts act in parallel with the existing system, rather than completely overhauling it.
The biggest difficulty with the mechanism (as Kremer acknowledges in the paper) is that it can be plagued by incompetent or corrupt government officials. Kremer suggests ways of overcoming these problems. For example, rather than the government matching the highest price (plus a premium) in the auction, they would instead select the third highest bid. This lowers the chances of the government having to pay for overzealous bids (also known as the winner’s curse). Additionally, if firms tried to collude to get higher prices, it would have to be three firms that collude, making collusion more difficult.
Finally, another problem lies with the fact that the government has to select the right patents. The government could pick patents to buyout that do not benefit society much.
Similarly, imagine there are two products, A and B. Let’s say product A is superior to product B, but the government chooses to buy the patent of product B. This means that the government could flood the market with an inferior product.
3. The Daguerreotype process of photography: A historical example
In 1837, Louis Daguerre invented the daguerreotype process of photography. The video below shows how it works.
Daguerre was struggling to sell his new invention. Fortunately for him, Francois Arago, a politician and member of the Academie des Sciences argued “that the government should compensate M. Daguerre direct, and that France should then nobly give to the whole world this discovery which could contribute so much to the progress of art and science.” (quoted in Kremer).
In 1839, the French government bought the invention from Daguerre and put the process into the public domain. After this patent buyout, Daguerreotype photography spread across other countries and was subject to a number of improvements. Furthermore, the technique had spillover effects into improving innovation in chemistry and the production of lenses.
This is one of my favourite papers. Consumers gain through access to new innovation, innovators gain because they are paid a premium to their patents, and governments gain by improving the welfare of their citizens. Kremer suggests it could be experimented with on a smaller scale at first, and if successful, gradually expand its application. I hope it is experimented with in the future.
In a later post, I will address another paper in a later post, ‘Advance Market Commitments‘, which is something Kremer also pioneered.
Finally, some may find this interesting. The word patent originates from the Latin word ‘patere‘, which ironically means, ‘to lay open’.
When can fiction change the world? by Timothy Underwood. “As part of this I did some thinking about when fiction seemed to exert an influence on public policy, and then I looked for academic research on the subject, and I think there are people… who will find this write up about the subject interesting and useful.”
There have also been a few podcasts over the last couple of weeks.
Village Global’s Venture Stories podcast with Jason Crawford. They discuss: “the key aspects of human progress; the history of progress over time; whether we’ve traded off progress for safety; why the idea of progress is relatively new; what the nature of science fiction writing tells us about our vision for progress; why progress happens differently in different domains; how to think about safety with respect to new technologies; the impacts of slowing population growth.”
“He joined Tyler for a conversation about which areas of science are making progress, the factors that have made research more expensive, why government should invest more in R&D, how lean management transformed manufacturing, how India’s congested legal system inhibits economic development, the effects of technology on Scottish football hooliganism, why firms thrive in China, how weak legal systems incentivize nepotism, why he’s not worried about the effects of remote work on American productivity (in the short-term), the drawbacks of elite graduate programs, how his first “academic love” shapes his work today, the benefits of working with co-authors, why he prefers periodicals and podcasts to reading books, and more.”
Sam Hughes (Centre for Global Development) and I are working on a few posts together. One of these is a post on taxonomising criticisms of Progress and Progress Studies. This forum post falls under one of the categories we’re writing about, that being critiques of Progress/growth due to existential risk/unintended outcomes.
Vollrath explains “the underlying idea in growth economics that it is innovations and invention that drive growth in the long run. But the similarity of the Silicon Valley experience to how Mokyr describes the Industrial Revolution experience does suggest that this innovation and invention is less a function of economy-wide aggregate features (e.g. demographics, trade) and more on niche groups of innovators embedded in specific places and cultures.”
Then Vollrath suggests, “You could also file this as an example of the idea that it is better to concentrate your investment and R&D (and education?) rather than making it broad-based.” hat tip Matt Clancy
He suggests that Sebastian Cabot (1474-1557), could be a “possible candidate for [the] most influential person in England in mid-16th century”. Cabot could be responsible for introducing patents, using these to bring in investment to form joint stock companies, and also led a number of foreign expeditions. Anton said he would be writing this up and I’m looking forward to reading that.
At the moment it’s hard to get a time that suits everyone because people are from different parts of the world. If we fail to converge on a time zone that enables a significant number of people participating, I was thinking of resorting to do it via a Forum.
There will be a Progress Studies Study Group (with an amazing line-up!), hosted by Jason Crawford. I should say it’s $2,400, and $1,200 for students (although even with the discount, it’s out of my budget sadly).
I’ve recently began to organise a virtual Progress Studies reading group. We’re still in the early stages of launching it (organising the reading agenda and a time that suits everyone). Everyone is welcome to attend. A few details:
The format we’re going for is that one person every week is selected to summarise the designated reading and present that summary to the rest of the group. We then proceed to discuss that topic.
If you would like to join, we’re organising/communicating via the Progress Studies Slack channel. We’re still in the process of setting the agenda for what to read, so feel free to join and suggest things you find interesting (this can be journals, blogs, books, podcasts, etc.). So far we have approximately 5-10 people interested, including Jason Crawford (Roots of Progress) and Sam Hughes (Centre for Global Development).
I’m beginning my journey into Progress Studies by summarising and synthesising some of the literature scattered around.
Here, I show an example of the potential power of Progress Studies. The policy implications of the paper I summarise in this post demonstrate the low-hanging fruit available to the field. If we were to implement the policy prescriptions of the paper below, we could improve the rate of scientific knowledge being created.
The paper attempts to better understand the determinants of idea/knowledge production.
There are two key findings:
Firstly, individuals who are ‘talented’ as teenagers (proxied by International Mathematical Olympiad [IMO] score, a popular international Maths competition) are very capable of advancing the knowledge frontier later in life. On average, the higher the IMO score, the more likely a participant is to: obtain a Maths PhD, obtain a Maths PhD from a top 10 research school, publish more in academic journals, receive more citations, receive notable accolades such as the Fields Medal, or be a speaker at the International Congress of Mathematicians (ICM).
Secondly, if these capable individuals were born in poorer countries, they are much less likely to contribute to the knowledge frontier than individuals born in wealthier countries. For example, on average IMO participants from low-income countries produce 34% fewer publications and receive 56% fewer citations than IMO participants from richer countries with the same IMO score.
The findings suggest that one way of developing the knowledge frontier faster is by creating policies to target these low-income students in order to support them in their academic careers. By doing so, we could improve the rate of scientific progress.
I’ll now proceed to write a slightly longer summary of the paper. The structure will stay largely similar to the original paper i.e. follows the same order, but with less detail and excludes the regressions.
Key questions the paper investigates
How much does talent displayed in teenage years affect the amount of knowledge produced later in life?
Conditional on a given level of talent in teenage years (IMO score), how much does the country of birth influence the quantity of knowledge produced later in life?
The International Mathematical Olympiad (IMO) is a prestigious mathematical competition held annually since 1959. Individuals from across the world represent their countries and compete to win bronze, silver, and gold medals. Participants confront six questions with different levels of difficulty. Each problem is worth a maximum of seven points and the highest possible score is 42 points. The authors collect data on 4710 IMO participants from 1981-2000. This data contained the year of participant, country of origin, points scored, and type of medal achieved.
The authors also collected data on the long-term outcomes in mathematics for these individuals. To collect data on PhD theses, they used the Mathematics Genealogy Project. This project aims “to compile information on all mathematicians in the world.” This source had information on: the name of the student, the university they attended, the name of the advisor, graduation year, and dissertation topic.
The authors used MathSciNet to collect bibliometric data. This source has information on the total publications and citations broken down by author.
Agarwal and Gaule also collected data on particularly notable/prestigious awards in Mathematics. These were IMO participants who later became speakers at the International Congress of Mathematicians (ICM) or won Fields medals.
Finally, data was collected on the employment of the IMO medalists (2272 of the 4710 participants won medals), to see if they were employed in mathematics academia, outside of mathematics but in academia, in industry, or didn’t have an online profile.
1. How does talent (proxied by IMO score) as a teenager relate to later research success?
Figure 1 (below) plots various mathematical achievements (y-axis) against the number of points scored on the IMO (x-axis).
The first four graphs (three on the top row and one on the bottom-left) show a clear positive relationship between points scored at the IMO and subsequent mathematical success. Thus, on average the more points a participant scores at the IMO, the more likely they are to get a Math PhD, a Math PhD from a top 10 school, more publications (in logs), and more citations (in logs). The other two graphs show the relationship between points scored and exceptional Mathematical achievements later during their research careers (Fields Medal and ICM speakers). These show a positive but weaker relationship than the other graphs.
The authors then regress points scored at the IMO (dependent variable) and subsequent achievement (the six outcomes in Figure 1), whilst controlling for cohort, and country of origin. The regression (see paper) suggests that for each additional point scored at the IMO, there is on average:
1 percentage point increase in the likelihood of obtaining a PhD
2.6% increase in publications
4.3% increase in citations
0.1 percentage point increase in the likelihood of becoming an ICM speaker
0.03 percentage point increase in the likelihood of becoming a Fields medalist
Additionally, they find that if individuals score more points in the more difficult problems, then this is more predictive of future mathematical achievements than scoring points in the ‘easier’ problems.
Agarwal and Gaule then compare IMO medalists to mathematicians who didn’t participate in the IMO. They constructed a sample of all PhD students obtaining a PhD in Maths, and a subsample of PhD students from top 10 schools. They compared these samples against the accomplishments of bronze, silver, and gold IMO medalists who have a PhD.
Figure 2 (below) shows that for each outcome, the medalists (particularly the gold medalists) outperform PhD students and PhD students from top 10 schools, who did not participate in the IMO. Medalists consistently get more publications, citations, go on to become speakers at the ICM, and receive Fields medals during their careers.
In fact, the probability of receiving a Fields medal is fifty times larger for IMO gold medalists than the corresponding probability for a PhD graduate from a top 10 mathematics programme.
2. Conditional on a given level of talent in teenage years (IMO score), how much does the country of birth influence the quantity of knowledge produced later in life?
For the purpose of statistical analysis, the participants are grouped in terms of their countries income level (high income, upper middle-income, lower middle-income, and low-income). This grouping is a proxy for differences in opportunities and environment. The authors emphasise, “while our regressions explicitly control for IMO scores, it is worth noting that participants from developing countries do not score lower at the IMO than participants from developed countries.”
Figure 3 shows points scored at the IMO in five-point intervals (x-axis) against the share of PhD students receiving a PhD in Maths (y-axis).
Largely we can see that for each number of points scored at the IMO, the high-income countries (black) obtain the most Maths PhDs, followed by upper-middle (green), then lower-middle (grey), and finally low-income countries (blue), who received the least amount of Maths PhDs.
The authors then conduct another regression. This time the dependent variables are successful outcomes: PhD in Maths, PhD in Maths from a top school, publications (log), and citations (log). The main independent variable here is the income group of participants alongside control variables, and importantly the number of IMO points scored. Thus, we’re controlling for ‘talent’ and investigating the effect of the country’s income on future successes in Mathematics.
For the sake of brevity I have left the regression table out. The results of the regression suggest that participants from low and middle-income countries have less research accomplishments in their later careers than those individuals from richer countries with the same IMO score:
“IMO participants from low-income countries are 16 percentage points less likely to do a PhD and 3.2 percentage points less likely to do a PhD in a top school; they produce 34% fewer publications and 56% fewer cites.”
In fact, IMO participants from low-income countries are approximately half as likely to obtain a PhD from a top-tier school than their rich country participants with the same IMO score. This effect holds true of participants from middle-income countries, albeit with a smaller effect size. The IMO participants from low and lower-middle income countries are not more likely to be employed in non-mathematics academic positions, or in industry jobs.
The authors suggest that if IMO participants from low-income countries were producing knowledge at the same rate as those from high-income countries, then they would produce approximately 10% more publications, and 17% more citations.
Conclusion and policy recommendations
Firstly, the paper finds a clear link between ‘talent’ as measured by IMO score, and future research success. Secondly, the paper finds that conditional on a given level of talent, IMO participants from lower-income countries contribute to the research frontier substantially less than richer country participants. That is, despite achieving the same IMO scores, they contribute significantly less research later on in their careers.
Although this paper focused on Maths (the dataset was a really innovative way to overcome many empirical challenges), the same arguments could be applied to other fields. A particularly poignant point is that in Maths, people who have the highest levels of talent as teenagers (gold medalists) are for more likely to do significantly better across a variety of metrics designed to measure successful research compared to other individuals. This talent is rare. Losing this talent seems to suggest a significant waste in terms of scientific progress that could have been made otherwise. The authors suggest that “many geniuses from poor countries are never discovered or given the chance to excel as teenagers in the first place.”
There are a number of policies that could be administered in order to support the individuals from lower-income countries. Firstly, we could give scholarships/fellowships to these students in order to alleviate their financial concerns and support their development. Secondly, elite schools could encourage applications from talented individuals in developing countries. For example, MIT reaches out to international communities of talented individuals and provides them financial support if they need it. Finally, it could be possible to improve the training standards at universities in developing countries to nurture the talented individuals who don’t want to leave their country of birth.
I left out a number of caveats the authors raise. For example, there are potential competing explanations for the results.
A couple ideas for future research, perhaps:
Firstly, we have the results for the Math Olympiad. Can we generalise the results to other subjects? One way of doing so would be to look at the other (less well known) international competitions, such as the Physics Olympiad etc.
Secondly, this makes a clear case for international comparisons. I wonder how things operate at a within-country level? I.e. are there any similar set-ups, where we could exploit the use of domestic national competitions for example.
Thirdly, let’s say that it’s hard to obtain (expensive) financial scholarships for students from developing countries to top tier universities such as MIT/Harvard. One alternative solution may be to match students from developing countries with famous Professors from developed countries. Maths research is often an individual pursuit. So it may be possible to emulate the academic environment that leads to later Maths success, by providing access to mentors or access to online classes at top-tier universities.
It may be able to ‘sell’ this policy to companies/rich individuals. If we have mechanisms enabling us to identify talented individuals from an early age, a company could profit from a smartly administered income-sharing agreement.
Sam Altman recently had an interesting twitter thread that relates to theme of tapping into under-utilised potential: