David Oks
3 followers · 8 following · 976 views
on the atlas — 49
- Evo 2 Can Design Entire Genomes3 savers
- How To Understand Things4 savers
- The Technium: 50 Years of Travel Tips9 savers
- Bill Teplow's Voyage to Hawaii in a Potter 191 savers
- An Action Plan for American Leadership in AI1 savers
- Childhoods of exceptional people - by Henrik Karlsson1 savers
- Deep Learning is Not So Mysterious or Different2 savers
- benkuhn.net19 savers
- AI Wars1 savers
- The End of Credentialism?1 savers
- Arctic Instincts? The Late Pleistocene Arctic Origins of East Asian Psychology1 savers
- Network Arbitrage1 savers
- Elicitation, the simplest way to understand post-training1 savers
- From Investment to Savings: When Finance Feeds on Itself1 savers
- From Prompt to Adventures : Creating games with LLMs and Restate's durable functions1 savers
- Vilhelm Hammershøi: the eminence in greys — Richard Morris1 savers
- Roy Prosterman, who fought poverty through land rights, dies at 891 savers
- Roy Prosterman's contribution to China's land tenure1 savers
- The "everyone is biased" bias1 savers
- Reality has a surprising amount of detail49 savers
- People's deeply held beliefs are surprisingly surface-level1 savers
- If You're So Smart, Why Can't You Die?2 savers
- ProtoSnap: Prototype Alignment for Cuneiform Signs1 savers
- The Book of T1 savers
- Ban the H20: Competing in the Inference Age1 savers
- Post-racial nations are the future1 savers
- The Tinkerings of Robert Noyce8 savers
- A New, Chemical View of Ecosystems1 savers
- Reading the Economic History Reading List of the Very Sharp Pseudoerasmus1 savers
- So You Want to Learn About Economics3 savers
- governance ideology1 savers
- Three Wikipedia Pages For Great Economists1 savers
- Writing an LLM from scratch, part 11 savers
- We Live Like Royalty and Don’t Know It — The New Atlantis8 savers
- What fully automated firms will look like6 savers
- Making Deep Learning Go Faster29 savers
- The future belongs to those who prepare like Dwarkesh Patel | Meridian22 savers
- An open letter to 56 people I know - by Sasha Chapin20 savers
- How To Understand Things - Nabeel S. Qureshi18 savers
- How to Build a $20 Billion Semiconductor Fab15 savers
- Reflections on Palantir - Nabeel S. Qureshi13 savers
- On Medici and Thiel - by Rohit - Strange Loop Canon12 savers
- Strategies for Learning - by Andy Masley11 savers
- Anatomy of a credit card rewards program9 savers
- Be patient with problems - by Henrik Karlsson8 savers
- Attention Is Off By One – Evan Miller7 savers
- Error6 savers
- $2 H100s: How the GPU Bubble Burst - by Eugene Cheah6 savers
- KYC and AML: beyond the acronyms3 savers
highlights — 45
Lin Ostrom, Governing the Commons.
A socratic dialogue over the utility of DNA language models (Part 1 of 2)There’s a bedtime, after which you can’t be loud. There are taxes on drug sales. In the bigger prison systems there’s regulation of who lives in the community, with a particular eye to excluding specific people, usually sex offenders and former law enforcement officers.
Why We Have Prison GangsSo there’s a pretty standard set of procedures. That’s partly because prisoners in general are subject to inconveniences surrounding sight, smell, and sound.
Why We Have Prison GangsBut there are no gangs emerging in the small ones. The social order that exists in the small ones actually looks a lot like the convict code that existed when California had a smaller prison system. They’re also not as racially segregated in those smaller prisons.
Why We Have Prison GangsBut it’s also true that there are ways in which gangs stabilize communities. Large-scale disruptions are bad for gang business.
Why We Have Prison GangsIt’s clear that prison governance has gotten a lot better. For lack of a better term, the silver lining of mass incarceration is that corrections became professionalized and the technology to assist with it improved.
Why We Have Prison GangsI would say it’s not that they don't matter, it’s that they don’t provide the totality of the governance in a way that many of us would assume. Guards of course regulate the perimeter and are responsible for preventing contraband and things like that from coming in. They will also quell large-scale disruptions — through tear gas, etc. And they will investigate assaults and homicides that happen inside prison. So they do govern in all of these ways. But it’s clear that they don’t govern in their official capacity when it comes to daily social interactions.
Why We Have Prison GangsBut as prison populations in California grew through the ’50s and into the ’60s, it became too costly to know everyone’s standing. Reputation became a much less effective driver of behavior. And in the process of becoming less effective, we begin to see a lot of outbursts of serious acts of violence. During the ’50s and ’60s, there were large increases in homicides, stabbings, and riots.
Why We Have Prison GangsBut because of their control of prisons, they’ve been able to leverage a credible threat of violence that generates a tremendous amount of power and influence, both within prisons but also among those dealing drugs outside of prison.
Why We Have Prison GangsCalifornia had prisons for more than 100 years with no prison gangs. But today, when someone enters prison in California, they have almost no choice but to affiliate with racially segregated groups that operate under umbrellas of larger, more established, and very powerful prison gangs. And these gangs provide rules on how people can interact in social and communal life, as well as regulating the underground economy.
Why We Have Prison GangsThe financial logic requires AGI to parse.
$2 H100s: How the GPU Bubble Burst - by Eugene CheahIt is now consensus that the capex on foundation model training is the “fastest depreciating asset in history”, but the jury on GPU infra spend is still out and the GPU Rich Wars are raging.
$2 H100s: How the GPU Bubble Burst - by Eugene CheahThese include a priority on high technology, attention to geopolitical rivalry’s political-economic dimensions, and using the government for developmental investment. But to an extent with which Lichtenstein and Stein do not fully engage, the liberal advocates of the 1980s–1990s failed to build an actual political constituency for their vision of industrial policy and “managed capitalism.” Industrial policy remained, from the early 1980s through its reflorescence in the early 1990s, an overwhelmingly elite preoccupation.
Back to the ’90sThe decision to prioritize deficit reduction in 1993—in the wake of the 1990–1991 recession that had helped get Democrats elected—rearranged the zone of long-term political decision-making about a changing US economy away from investment in “reindustrialization” and towards investor “confidence.” These political conditions were conducive to the policies of the later 1990s and early 2000s that continued the shift of capital towards finance and real estate.
Back to the ’90sLichtenstein and Stein’s policy entrepreneurs not only admired the East Asian economic miracle but suggested—in influential books such as Ezra Vogel’s Japan as Number One (1979) and Chalmers Johnson’s MITI and the Japanese Miracle (1982)—that the institutions and high-tech orientation of Japanese industrial policy could be adapted for a new industrial policy, American-style.
Back to the ’90sYet Japan’s role in the political imaginary was, for many American liberals, dichotomous. It was not merely a threat but a model for economic policy. Reconsidering the industrial policy moment of the 1980s–1990s in A Fabulous Failure is one of Lichtenstein and Stein’s most significant historiographic contributions.
Back to the ’90sThe end of inflation not only threw many thrifts into bankruptcy, but precipitated the deflation of a commercial real-estate bubble in many growth centers. The increasing interrelation of domestic and foreign economies exacerbated the downturn. Famously, throughout the 1980s, Japanese investment had poured into US government securities and high-profile commercial real estate. When Japan’s stock market began to decline in 1989–1990, Japanese companies’ stateside investment fell precipitously—a collapse that would be particularly pronounced in early-1990s Southern California.
Back to the ’90sBecause of the collapse of employment in the aerospace industry and defense-related construction, Southern California encompassed more than a quarter of all job losses in 1990–1991.
Back to the ’90sMore than 90 percent of the job losses in those two years were concentrated in manufacturing.
Back to the ’90sNew accounts of the 1990s locate the origins of today’s politics not in the accommodations and defeats faced by labor and liberals in the 1970s and 1980s and their consolidation under Democratic Party leadership, but in the unfinished struggle that defined federal policy in the Clinton years over the transition from a productive industrial economy to the services-based postindustrial economy.
Back to the ’90sThe Negro American revolution is rightly regarded as the most important domestic event of the postwar period in the United States.
The Moynihan Report: The Negro Family, the Case for National ActionBut it almost certainly offers the only possibility of resolving in our time what is, after all, the nation’s oldest, and most intransigent, and now its most dangerous social problem. What Gunnar Myrdal said in An American Dilemma remains true today: “America is free to chose whether the Negro shall remain her liability or become her opportunity.”
The Moynihan Report: The Negro Family, the Case for National ActionA national effort is required that will give a unity of purpose to the many activities of the Federal government in this area, directed to a new kind of national goal: the establishment of a stable Negro family structure.
The Moynihan Report: The Negro Family, the Case for National ActionThe fundamental problem, in which this is most clearly the case, is that of family structure. The evidence — not final, but powerfully persuasive — is that the Negro family in the urban ghettos is crumbling. A middle-class group has managed to save itself, but for vast numbers of the unskilled, poorly educated city working class the fabric of conventional social relationships has all but disintegrated.
The Moynihan Report: The Negro Family, the Case for National ActionIndices of dollars of income, standards of living, and years of education deceive. The gap between the Negro and most other groups in American society is widening.
The Moynihan Report: The Negro Family, the Case for National ActionIn this new period the expectations of the Negro Americans will go beyond civil rights.
The Moynihan Report: The Negro Family, the Case for National ActionIn the decade that began with the school desegregation decision of the Supreme Court, and ended with the passage of the Civil Rights Act of 1964, the demand of Negro Americans for full recognition of their civil rights was finally met.
The Moynihan Report: The Negro Family, the Case for National ActionThe United States is approaching a new crisis in race relations.
The Moynihan Report: The Negro Family, the Case for National ActionEchoing civil-rights leaders of the time, Johnson declared, “Freedom is not enough”: Equal citizenship for African Americans was incomplete without the ability to make a decent living.
The Moynihan Report: An Annotated EditionBest known as the “Moynihan Report,” it launched the career of its author, who became a professor at Harvard University, a top adviser to President Nixon, and a four-term U.S. senator representing New York.
The Moynihan Report: An Annotated Editionover-reliance on LinesOfCode
Soviet Shoe Factory PrincipleThe factors that are easiest to measure or most visible are the ones that get the most attention, regardless of their importance in comparison to other factors. Another way of saying this is that factors that are easiest to measure or model are not necessarily the most important ones and/or a complete picture, but human nature often forgets this.
Soviet Shoe Factory PrincipleI think measurable incentives are a bit overrated and more suited for endeavours where the relationship between input and output is linear. Science is the opposite of that, with Pareto outcomes. Great Science is transformative, rare and happens when serendipidity and the will and passion of a few meet.
The Strong Researcher HypothesisGiven “strong” enough researchers, metascience may be left with only really being able to accelerate researchers in doing the work they were going to do anyways, an approach taken by programs like FastGrants.
The Strong Researcher HypothesisIf the hypothesis is true, then frequently the apparent influence of funders over researcher behavior is typically more a demonstration of researchers finding ways of manipulating funders to meet their own interests, rather than the other way around.
The Strong Researcher HypothesisSimply stated, it is the presumption that researchers are mostly driven by strong intrinsic motivations.
The Strong Researcher HypothesisIn other words, the very researchers metascience romanticizes the most are weird characters who are apparently deeply immune to the lure of external incentives.
The Strong Researcher HypothesisThe fields are instead products of an intense martial combat between communities of researchers for money, minds, and prestige.
Field Essentialism, Field Realism, Field PragmatismAt the end of the day, fields represent a specific kind of research machinery: a collection of rallying cries, norms, funders, and bureaucratic arrangements that are designed to output new insights about the world at large.
Field Essentialism, Field Realism, Field PragmatismMachine learning provides a useful illustration. At its heart, much of the core methodology of machine learning emerges from statistics. But in practice, it has emerged as a subfield of computer science. Why? The realist view is that computer science departments are more well-resourced, more prestigious, and were able to capture a commercially valuable domain as their own before fields like statistics were able to exploit the opportunity.
Field Essentialism, Field Realism, Field PragmatismPerhaps the greatest argument against field essentialism is the history of science itself. The boundaries between fields have always been fuzzy, and fields rise and fall over time. “Political economy” no longer exists, because political science and economics have emerged as more sharply defined fields in the 20th century. These changes don’t reflect some higher-order change in the nature of the universe, but something quite a bit closer to earth.
Field Essentialism, Field Realism, Field PragmatismIt also explains in part why efforts to redirect and reshape the energies of the scientific community through “field creation” often fail. You cannot define fields arbitrarily because certain things simply do not make sense to study together in a focused manner.
Field Essentialism, Field Realism, Field PragmatismThe collection of fields we have are, on some level, representative of reality itself. The study of people is simply different from the study of celestial bodies, so it is no surprise that we have fields that divide into something like sociology and something like astrophysics.
Field Essentialism, Field Realism, Field PragmatismWhy is it exactly that we end up with mathematics being separate from biology, both as separate departments and as distinct bodies of knowledge? Why do we end up with the specific set of fields that we have?
Field Essentialism, Field Realism, Field PragmatismThe field is to science as the nation-state is to geopolitics.
Field Essentialism, Field Realism, Field Pragmatism