One of the most important cognitive tradeoffs we make is how to process information, and perhaps more specifically, the deluge of information that bombards us every day. A study out of UCSD in 2009 estimated that Americans read or hear more than 100,000 words a day — an increase of nearly 350% over the prior three decades (and that was before Slack and Substack!) It would seem logical that more information is always better for decision-making, both for individuals and for societies. Yet, that’s precisely the tradeoff: humans and civilizations must balance greater and better information with the limits of their rationalities.
So what’s the value of learning more information, and can there even be a downside to knowing more?
Such was the emerging theme of two classic books I happened to serendipitously smash together this week. The first was psychologist Philip E. Tetlock’s original 2005 magnum opus, Expert Political Judgment, a book that preceded his hyper-popular bestseller Superforecasters. The other was The Collapse of Complex Societies, a monograph published in 1988 and written by anthropologist Joseph A. Tainter. These two books, drawing from divergent fields and investigating different problems, ultimately converge around a central thesis: there is a marginal return to information, and it often reaches zero — or even negative — far earlier than we might expect.
As with all economic analysis, the marginal return of information is the relationship between a “unit” increase in information and the value created by that new knowledge. If I have never studied biology and I read one textbook, the marginal value of information is high: I am learning a lot of new information and immensely expanding my knowledge of the subject. However, reading 20 biology textbooks doesn’t mean I am 20 times smarter on the subject. There are diminishing returns — the marginal value of each additional textbook I read will get smaller and smaller.
Tetlock’s central thesis, based on extensive laboratory data, is that specialization, education, and depth of knowledge does not lead to better political forecasting. In fact, a generalist who regularly reads a quality set of news sources is often much more likely to correctly predict the future than a specialist. The key predictor rather is not knowledge but cognitive style. The best forecasters are “foxes,” from Isaiah Berlin’s metaphor of the fox who knows many things and the hedgehog who knows one thing well. As Tetlock writes:
Several foxes commented that good judges cultivate a capacity “to go with the flow” by improvising dissonant combinations of ideas that capture the “dynamic tensions” propelling political processes. For these self-conscious balancers, the status quo is often in precarious equilibrium and “seeing the truth” is a fleeting achievement for even the most Socratic souls.
The best forecasters are able to draw from their extensive kens and selectively process and synthesize the information they have at hand.
Can flexibility of thinking or “improvising dissonant combinations of ideas” (a phrase I really love) ever have a negative marginal value? Well, actually, yes. In a series of experiments, Tetlock asks participants to weigh probabilities of outcomes across a range of scenarios that get increasingly complicated. What he finds is that “…foxes become more susceptible than hedgehogs to a serious bias: the tendency to assign so much likelihood to so many possibilities that they become entangled in self-contradictions.”
Since foxes use self-doubt to correct their predictions, feeding them more information on other perspectives or contradictory evidence can actually turn a relatively statistically accurate prediction into a much weaker forecast. The marginal return on information doesn’t just tend toward zero — it can actually become negative in certain contexts.
Tetlock approaches this marginal value of information from the perspective of a psychologist, and the core of his book covers his research on various human biases as well as subgroups of hedgehogs and foxes and when they perform particularly well and when they slump. Tainter, on the other hand, is interested in societies and how agglomerations of people can suddenly lose their productive sophistication. Or, as he puts it, “A society has collapsed when it displays a rapid, significant loss of an established level of sociopolitical complexity.”
The word “complexity” typically has a negative valence in our modern commentary, but complexity is generally good. As societies move from hunter-gatherer models of economic production to more specialization, surpluses of goods like food grow, and that allows for a greater number of roles across society as well as the development of culture.
Tainter’s goal is to find a theory that encompasses dozens of different societal collapses in history, while avoiding many of the logical contradictions in other hypotheses. For instance, if additional complexity adds more productivity to an economy, then shouldn’t it precisely be the most advanced and complex economies that are able to weather exogenous shocks and avoid collapse? What he finds and develops into a thesis is that there are limits of how much benefit complexity offers. In his words:
It is the thesis … that return on investment in complexity varies, and that this variation follows a characteristic curve. More specifically, it is proposed that, in many crucial spheres, continued investment in sociopolitical complexity reaches a point where the benefits for such investment begin to decline, at first gradually, then with accelerated force.
Societies can continue to grow in complexity, but then they start to stretch themselves thin.
Yet a society experiencing declining marginal returns is investing ever more heavily in a strategy that is yielding proportionately less. Excess productivity capacity will at some point be used up, and accumulated surpluses allocated to current operating needs. There is, then, little or no surplus with which to counter major adversities.
That major adversity can be a climate catastrophe or a barbarian invasion, but when it comes, there is no slack in the system or any unused fount of knowledge to draw upon. Upkeep of the vast complexity that manages society consumes immense resources, and that hinders people from adapting to their new and more challenging context. Without immediate economic growth or technological innovation to compensate, the system falls apart.
Tainter’s theory is that there are declining marginal returns to investment in complexity, but he also implies that such complexity can also turn negative. There is a point at which complexity begets further complexity with no increase in productivity, essentially leveraging a collective “tax” on everyone to maintain an ever more complicated bureaucracy for no value.
Thus, Tetlock and Tainter converge from disparate paths to what might be dubbed a “theory of marginal stupidity.” While education is extremely valuable, and more information and complexity is generally good for decision-making and societal productivity, there is a turning point where further information or complexity can befuddle us and simply raise costs without any concomitant value. Yes, the world is always changing, and much like the Red Queen in Alice in Wonderland, we always need to be learning just to avoid getting dumber (or as our scientist-in-residence Sam Arbesman titled his book, there is The Half-Life of Facts). But critically, there’s a limit to how much knowledge consumption is beneficial.
That theory has huge implications on fields like science, where nearly all indicators of productivity have been scorchingly negative for decades. Even Tainter, writing in 1988, notes in an extensive section that “… in any field, as each research question among the stock waiting to be answered is resolved, the costliness of deciphering the remainder increases.”
How do we square the declining marginal return of knowledge to the fact that it seems we need more knowledge than ever to make any impact in many fields? I’ve written before about the “dual PhD” problem for TechCrunch, that more and more deep tech startups essentially require founders with two PhDs in intersecting domains like machine learning and biology). Others analysts have noted the same trend such as Benjamin F. Jones, who wrote a quality economics analysis on the subject titled “The Burden of Knowledge and the ‘Death of the Renaissance Man’.”
The answer is to recognize that additional information is often not helpful, and instead, we should explore frontiers that have never been investigated. Tainter notes a widely observed fact that “It is no coincidence that the most famous practitioners historically in each field tend to be persons who were instrumental in developing the field, and in establishing its basic outline.” The most value comes from mapping uncharted territory, and while more of our world has certainly been mapped, that doesn’t mean there are no areas left to explore. The tradeoff is plunging into the unknown, but potentially finding something new. It certainly beats reading biology textbook number 21.
A tale of two fundings
From our Lux portfolio, two important new rounds were announced:
Anagenex, a platform that uses a combination of machine learning and biotech to screen for potential small molecule medicines, raised a $30 million Series A led by biotech veterans Catalio. Lux was the founding seed investor in 2020.
Remember when PC games and software came in packaged boxes? Many were boring squares, but Hock Wah Yeo designed a number of imaginative packages for games like Brøderbund’s Prince of Persia and Electronic Arts’s Ultrabots. Phil Salvadorcompiles these unique software boxes and narrates their history in a fun look at a completely different time in PC gaming.
Peter Hébertshares this powerful clip from Peter Fortenbaugh, Co-CEO and Chief Community Builder of Boys & Girls Clubs of the Peninsula, as he wrestles with advanced cancer and the meaning of a life of service.
For those who have read Emily St. John Mandel’s Station Eleven (I have, and loved it), well, her latest novel is out and our scientist-in-residence Sam Arbesman is a huge fan. Sea of Tranquility is a speculative fiction work about time travel, metaphysics, and the connections between all of us and across time.
Finally, “Securities” reader Andrew Thompson has been publishing a beautiful site, in-depth analyses, and a bounty of comprehensive datasets on subjects ranging from Product Hunt and consumer tech review media to Bandcamp. Lots of fun detail, and more data than anyone should ever have time to play with.
That’s it, folks. Have questions, comments, or ideas? This newsletter is sent from my email, so you can just click reply.
Forcing China’s AI researchers to strive for chip efficiency will ultimately shave America’s lead
In incididunt ad qui nostrud sint ullamco. Irure sint deserunt Lorem id officia dolore non. Anim dolor minim sit dolor et sint aliquip qui est. Ex in tempor laborum laboris dolor laboris ullamco quis. Enim est cupidatat consequat est culpa consequat. Fugiat officia in ea ea laborum sunt Lorem. Anim laborum labore duis ipsum mollit nisi do exercitation. Magna in pariatur anim aute.
In incididunt ad qui nostrud sint ullamco. Irure sint deserunt Lorem id officia dolore non. Anim dolor minim sit dolor et sint aliquip qui est. Ex in tempor laborum laboris dolor laboris ullamco quis. Enim est cupidatat consequat est culpa consequat. Fugiat officia in ea ea laborum sunt Lorem. Anim laborum labore duis ipsum mollit nisi do exercitation. Magna in pariatur anim aute.
Right now, pathbreaking AI foundation models follow an inverse Moore’s law (sometimes quipped “Eroom’s Law”). Each new generation is becoming more and more expensive to train as researchers exponentially increase the number of parameters used and overall model complexity. Sam Altman of OpenAI said that the cost of training GPT-4 was over $100 million, and some AI computational specialists believe that the first $1 billion model is currently or will shortly be developed.
As semiconductor chips rise in complexity, costs come down because transistors are packed more densely on silicon, cutting the cost per transistor during fabrication as well as lowering operational costs for energy and heat dissipation. That miracle of performance is the inverse with AI today. To increase the complexity (and therefore hopefully quality) of an AI model, researchers have attempted to pack in more and more parameters, each one of which demands more computation both for training and for usage. A 1 million parameter model can be trained for a few bucks and run on a $15 Raspberry Pi Zero 2 W, but Google’s PaLM with 540 billion parameters requires full-scale data centers to operate and is estimated to have cost millions of dollars to train.
Admittedly, simply having more parameters isn’t a magic recipe for better AI end performance. One recalls Steve Jobs’s marketing of the so-called “Megahertz Myth” to attempt to persuade the public that headline megahertz numbers weren't the right way to judge the performance of a personal computer. Performance in most fields is a complicated problem to judge, and just adding more inputs doesn't necessarily translate into a better output.
And indeed, there is an efficiency curve underway in AI outside of the leading-edge foundation models from OpenAI and Google. Researchers over the past two years have discovered better training techniques (as well as recipes to bundle these techniques together), developed best practices for spending on reinforcement learning from human feedback (RLHF), and curated better training data to improve model quality even while shaving parameter counts. Far from surpassing $1 billion, training new models that are equally performant might well cost only tens or hundreds of thousands of dollars.
This AI performance envelope between dollars invested and quality of model trained is a huge area of debate for the trajectory of the field (and was the most important theme to emanate from our AI Summit). And it’s absolutely vital to understand, since where the efficiency story ends up will determine the sustained market structure of the AI industry.
If foundation models cost billions of dollars to train, all the value and leverage of AI will accrue and centralize to the big tech companies like Microsoft (through OpenAI), Google and others who have the means and teams to lavish. But if the performance envelope reaches a significantly better dollar-to-quality ratio in the future, that means the whole field opens up to startups and novel experiments, while the leverage of the big tech companies would be much reduced.
The U.S. right now is parallelizing both approaches toward AI. Big tech is hurling billions of dollars on the field, while startups are exploring and developing more efficient models given their relatively meagre resources and limited access to Nvidia’s flagship chip, the H100. Talent — on balance — is heading as it typically does to big tech. Why work on efficiency when a big tech behemoth has money to burn on theoretical ideas emanating from university AI labs?
Without access to the highest-performance chips, China is limited in the work it can do on the cutting-edge frontiers of AI development. Without more chips (and in the future, the next generations of GPUs), it won’t have the competitive compute power to push the AI field to its limits like American companies. That leaves China with the only other path available, which is to follow the parallel course for improving AI through efficiency.
For those looking to prevent the decline of American economic power, this is an alarming development. Model efficiency is what will ultimately allow foundation models to be preloaded onto our devices and open up the consumer market to cheap and rapid AI interactions. Whoever builds an advantage in model efficiency will open up a range of applications that remain impractical or too expensive for the most complex AI models.
Given U.S. export controls, China is now (by assumption, and yes, it’s a big assumption) putting its entire weight behind building the AI models it can, which are focused on efficiency. Which means that its resources are arrayed for building the platforms to capture end-user applications — the exact opposite goal of American policymakers. It’s a classic result: restricting access to technology forces engineers to be more creative in building their products, the exact intensified creativity that typically leads to the next great startup or scientific breakthrough.
If America was serious about slowing the growth of China’s still-nascent semiconductor market, it really should have taken a page from the Chinese industrial policy handbook and just dumped chips on the market, just as China has done for years from solar panel manufacturing to electronics. Cheaper chips, faster chips, chips so competitive that no domestic manufacturer — even under Beijing direction — could have effectively competed. Instead we are attempting to decouple from the second largest chips market in the world, turning a competitive field where America is the clear leader into a bountiful green field of opportunity for domestic national champions to usurp market share and profits.
There were of course other goals outside of economic growth for restricting China’s access to chips. America is deeply concerned about the country’s AI integration into its military, and it wants to slow the evolution of its autonomous weaponry and intelligence gathering. Export controls do that, but they are likely to come at an extremely exorbitant long-term cost: the loss of leadership in the most important technological development so far this decade. It’s not a trade off I would have built trade policy on.
The life and death of air conditioning
Photo by Ashkan Forouzani on Unsplash
Across six years of working at TechCrunch, no article triggered an avalanche of readership or inbox vitriol quite like Air conditioning is one of the greatest inventions of the 20th Century. It’s also killing the 21st. It was an interview with Eric Dean Wilson, the author of After Cooling, about the complex feedback loops between global climate disruption and the increasing need for air conditioning to sustain life on Earth. The article was read by millions and millions of people, and hundreds of people wrote in with hot air about the importance of their cold air.
Demand for air conditioners is surging in markets where both incomes and temperatures are rising, populous places like India, China, Indonesia and the Philippines. By one estimate, the world will add 1 billion ACs before the end of the decade. The market is projected to before 2040. That’s good for measures of public health and economic productivity; it’s unquestionably bad for the climate, and a global agreement to phase out the most harmful coolants could keep the appliances out of reach of many of the people who need them most.
This is a classic feedback loop, where the increasing temperatures of the planet, particularly in South Asia, lead to increased demand for climate resilience tools like air conditioning and climate-adapted housing, leading to further climate change ad infinitum.
Josh Wolfe speaking at Nvidia Auditorium at Stanford. Photo by Chris Gates.
Josh Wolfe gave a talk at Stanford this week as part of the school’s long-running Entrepreneurial Thought Leaders series, talking all things Lux, defense tech and scientific innovation. The .
Lux Recommends
As Henry Kissinger turns 100, Grace Isford recommends “Henry Kissinger explains how to avoid world war three.” “In his view, the fate of humanity depends on whether America and China can get along. He believes the rapid progress of AI, in particular, leaves them only five-to-ten years to find a way.”
Our scientist-in-residence Sam Arbesman recommends Blindsight by Peter Watts, a first contact, hard science fiction novel that made quite a splash when it was published back in 2006.
Mohammed bin Rashid Al Maktoum, and just how far he has been willing to go to keep his daughter tranquilized and imprisoned. “When the yacht was located, off the Goa coast, Sheikh Mohammed spoke with the Indian Prime Minister, Narendra Modi, and agreed to extradite a Dubai-based arms dealer in exchange for his daughter’s capture. The Indian government deployed boats, helicopters, and a team of armed commandos to storm Nostromo and carry Latifa away.”
Sam recommends Ada Palmer’s article for Microsoft’s AI Anthology, “We are an information revolution species.” “If we pour a precious new elixir into a leaky cup and it leaks, we need to fix the cup, not fear the elixir.”
I love complex international security stories, and few areas are as complex or wild as the international trade in exotic animals. Tad Friend, who generally covers Silicon Valley for The New Yorker, has a great story about an NGO focused on infiltrating and exposing the networks that allow the trade to continue in “Earth League International Hunts the Hunters.” "At times, rhino horn has been worth more than gold—so South African rhinos are often killed with Czech-made rifles sold by Portuguese arms dealers to poachers from Mozambique, who send the horns by courier to Qatar or Vietnam, or have them bundled with elephant ivory in Maputo or Mombasa or Lagos or Luanda and delivered to China via Malaysia or Hong Kong.”