Tequehead

Really enjoyed chatting with @michael_nielsen about how we recognize scientific progress. It's especially relevant for closing the RL verification loop for scientific discovery. But it's also a surprisingly mysterious and elusive question when you look at the history of human science. We approach this question stories like Einstein (who claimed that he hadn't even heard of the famous Michelson-Morley experiment, which is supposed to have motivated special relativity, until after he had come up with the theory), Darwin (why did it take till 1859 to lay out an idea whose essence every farmer since antiquity must have observed?), Prout (how do you recognize that isotopes exist if you cannot chemically separate them?), and many others. The verification loop on scientific ideas is often extremely long and weirdly hostile. Ancient Athenians dismissed Aristarchus's heliocentrism in the 3rd century BC because it would imply that the stars should shift in the sky as the Earth orbits the sun. The first successful measurement of stellar parallax was in 1838. That's a 2,000-year verification loop. But clearly human science is able to make progress faster than raw experimental falsification/verification would imply, and in cases where experiments are very ambiguous. How? Michael has some very deep and provocative hypotheses about the nature of progress. One I found especially thought-provoking is that aliens will likely have a VERY different science + tech stack than us. Which contradicts the common sense picture of a linear tech tree that I was assuming. And has some interesting implications about how future civilizations might trade and cooperate with each other. So many other interesting ideas. Hope you enjoy this as much as I did. 0:00:00 – How scientific progress outpaces its verification loops 0:17:51 – Newton was the last of the magicians 0:23:26 – Why wasn’t natural selection obvious much earlier? 0:29:52 – Could gradient descent have discovered general relativity? 0:50:54 – Why aliens will have a different tech stack than us 1:15:26 – Are there infinitely many deep scientific principles left to discover? 1:26:25 – What drew Michael to quantum computing so early? 1:35:29 – Does science need a new way to assign credit? 1:43:57 – Prolificness versus depth 1:49:17 – What it takes to actually internalize what you learn Look up Dwarkesh Podcast on YouTube, Apple Podcasts, or Spotify.

As always, the best stuff is in the system card. During testing, Claude Mythos Preview broke out of a sandbox environment, built "a moderately sophisticated multi-step exploit" to gain internet access, and emailed a researcher while they were eating a sandwich in the park.

"Mythos found a 27-year-old vulnerability in OpenBSD—which has a reputation as one of the most security-hardened operating systems in the world and is used to run firewalls [...] The vulnerability allowed an attacker to remotely crash any machine running the operating system"

Reminder: Waymo is so safe that if every car was driven like a Waymo, about 9% of America's life expectancy gap would disappear. 9 percent! Americans die in car accidents *that often*.

Italy's population has stopped falling due to Meloni taking in more non-EU migrants. Regardless of who you vote for, when your country is stuck in the aging pensioner voter doom loop politicians will increase immigration to keep the music playing a little bit longer

Just finished reading The Invention of Air, a book about Joseph Priestley. I recommend it. Priestley is a fascinating character. He was an amateur scientist in the purest and most beautiful sense of the word. In 1765, aged 32, Priestley was working as a schoolteacher and minister in Warrington, England. He was avidly following electricity experiments at the time and decided to write a “popular history” of the field. To pitch his book, Priestley traveled to London to a coffeehouse meeting of the Honest Whigs, where he first met — and was encouraged by — Benjamin Franklin. Priestley left that meeting with: - Access to Franklin’s personal library and all correspondence on electricity (a kind gesture from the American statesman). - A promise for funding to support the book’s printing. - Encouragement, by the Honest Whigs, to conduct his own experiments while writing the book. This was undoubtedly one of the most important scientific meetings of all time. Priestley went back to Warrington and immediately began working. He wrote his entire 700-page history in a single year, and also began his own experiments. During various attempts to replicate published experiments, Priestley discovered that charcoal conducts electricity, a finding that led to his election to the Royal Society that same year (1766). But I think the most fascinating part of Priestley’s life is not his experiments, but rather the way he supported his work. For a time, Priestley was funded by subscribers, an arrangement which was highly unusual in the late-1700s, when many independent scientists had immense personal wealth or got support from a wealthy patron. Henry Cavendish, for example, inherited two fortunes during his lifetime. He had so much money, in fact, that the French astronomer Jean-Baptiste Biot called him "the richest of all the savants and the most knowledgeable of the rich." Galileo Galilei was funded almost entirely by Cosimo II, a Medici. Tycho Brahe was funded handsomely by King Frederick II of Denmark. (Indeed, the Danish King spent so much money on Brahe’s experiments that it’s been estimated that about 1 percent of the Danish crown’s total revenue went to the scientist. Brahe was also given an entire island for his experiments, on which he constructed an astronomical observatory and alchemy laboratory, called Uraniborg; “the first custom-built observatory in modern Europe,” according to Wikipedia.) But Priestley was not wealthy. He attracted support from wealthy sponsors (including Lord Shelburne, who later became Prime Minister) on occasion, but often struggled for funds. This is why Priestley turned to his network of “subscribers,” who collectively sent him about 200 guineas per year. At the time, one guinea was worth 21 shilling, or roughly £1.05, meaning Priestley’s subscribers sent him about £210 per year. This is a lot, considering a skilled artisan, at the time, made about £50 per year and £100 was enough to maintain a solidly middle-class household with servants. (A seaman sailing with the East India Company only earned about £21 per year!) Subscribers included the Galtons (father and son), Sir George Savile, Josiah Wedgwood, and other figures of the local community. In exchange for about 10 guineas per year, these subscribers received not only the satisfaction of supporting a prominent scientist and hearing about cutting-edge experiments before anybody else, but also early access to Priestley’s books and publications. “Without assistance I could not have carried on my experiments except on a very small scale and under great disadvantages,” Priestley once wrote. Indeed, Priestley thought a good deal about science funding, and was ahead of his time on the subject. In 1767, he also outlined plans for industry-funded research centers. Priestley liked that the large institutes then supporting science, such as the Académie Française, supported research, but "he objected to the centralized nature of those societies,” writes Steven Johnson in The Invention of Air. Priestley didn’t like that a single individual (like Antoine Lavoisier in France) had so much power over the research of the institute. Instead, Priestley proposed “smaller and more nimble clusters,” where many different companies all contribute funds to a research center. Each research center would have a “director of experiments,” who would perform experiments on behalf of the supporting companies. The companies that gave more money would have more control over which experiments got done, and all companies would have proportional votes relative to their funding levels. It seems this idea never panned out (or, at least, I’m not aware of anyone who has tried this.) But the idea of industry-funded R&D is much older than I anticipated, and we owe part of its vision to Joseph Priestley.

I wrote a list of recommendations for beginners a few years back. Might add or change one or two titles now, but it hold up pretty well scholars-stage.org/making-sense-o…

As Trump’s deadline for attacking Iran’s power plants draws near, here are a few brief informational points about the current condition of Iran’s electricity sector, and how prolonged blackouts may affect it: 🧵

0 robot data. Less than 2hrs of human data. Model opens its eyes and suddenly sees it’s become a robot.

So many judging tasks could be improved by aggregating partial orderings, and in the limit, just ordering pairs. The annual Libertarian Futurist Society novel awards discussion is starting, and while I would like to participate on some level, there is no way I have time to read an entire slate of novels. However, I will likely read at least two from the list, and I could give a relative assessment. This cries out for the use of something like ELO ranking, as in chess competition, perhaps with some suggestions to get sufficient coverage. Peer and out-of-chain employee performance calibrations could probably also benefit from a greater quantity of sparse pairwise comparisons

Liberals quitting Twitter in a huff was a weak-minded abdication of political responsibility. Like their nomination of Kamala Harris, the exodus demonstrated that, to them, feeling morally unassailable was more important than beating Trump and combating Trumpism.

The Skill Premium in Times of Rapid Technological Change Fantastic new @nberpubs WP by @t_a_hassan - @AakashKalyani - Restrepo (@YaleEconomics). 💡In a model where skilled workers have a comparative advantage in learning new technologies, a rapid pace of technology creation leads to a sustained increase in the skill premium. I think of this as a modern application of the "Nelson-Phelps (1966) view of human capital", used to understand the college-wage premium.

This is an actual post. This is not funny. This is beyond desperate. This is a deeply unwell man who doesn’t belong anywhere near the levers of power. Every member of his cabinet and Congress is complicit in not demanding his removal now.

We sometimes think an outcome was caused by two things. We might say Amy got sick because (a) There was cilantro in the soup *and* (b) Amy is allergic to cilantro Beautiful new theory of this kind of "plural causation" from @TadegQuillien direct.mit.edu/opmi/article/d…

Anyone who thinks that Longevity Biotechnology is new and in the next 10 years we will find a cure for aging with AI should read Paul Segall's "Living Longer, Growing Younger". 1989. Read "Merchants of Immortality" 2003 next. I think that the first order of business is to build a sustainable business model with aging research at the core to have organizational longevity and growth as the first order of business.

I have written at length about the need to streamline Phase I trials to facilitate biomedical innovation. The President's 2027 Budget makes it the administration's official position that such measures should be implemented. This is tremendous news.

A recent @Research_FRI survey found that experts predict muted growth even under rapid AI progress. Why is that? @testingham, @BasilHalperin, @random_walker, @paulnovosad, and @alexolegimas provided competing explanations. I break down the debate: update.news/p/how-much-wil…

One problem in society is that dumb people are extra eager to share their dumb ideas

Does genetic support increase drug success rates by 2–3 times? In a previous post, I revisited the original Nature paper behind this claim. My new post examines additional evidence from 23andMe that is often cited as a replication of the original analysis.