Wednesday, March 25, 2026
LeWorldModel, LeWM (JEPAs)
🚨 Holy shit... LeCun's team just cracked world models wide open.
— Alex Prompter (@alex_prompter) March 24, 2026
Everyone's obsessing over the next Claude update.
Meanwhile Yann LeCun quietly dropped a paper that could matter way more long term.
It's called LeWorldModel.
And to understand why it's a big deal, you need… pic.twitter.com/jcwo02hTYB
Who makes the photo, the subject or the photographer?
Of course it’s the photographer. I’m speaking somewhat metaphorically. Take a look at this photo:
Rather striking, dramatic, no? That’s a photo where the subject makes the shot. What do we have? A well-known skyline, Lower Manhattan, an iconic building, One World Center, and dramatic light, what photographers call the Golden Hour. A trained monkey standing where I stood could have gotten that shot.
This shot’s a bit different.
It’s still dramatic, with strong geometry converging toward the center and some dramatic light. But the subject is quite ordinary, a bar at a standard-issue middle-tier restaurant. It takes a photographer to see that as a proper subject for a photography.
And then we have this:
WTF? What’s in the center of the shot, where the subject should be? There’s a doorway with “Soulcycle” above it and a clock above that. If that’s the subject, then why’s it obscured by that tree right of center, the pole to the left, and the car in front? Why? Because I took that photo and that’s what I wanted. It’s not a mistake. It’s intention. It makes you aware of the space between you and the ostensible subject.
Here’s a similar shot:
If you look closely at the enter you’ll see an iconic building, the Empire State Building, no less, the best-known building in New York City. But it’s got all those trees in front of it, and then there’s the hedge in the foreground, and the people. All intended. Not, mind you that I posed anything. I’m basically a street photographer, I walk around and shoot what I see, though on occasion I may take some pains to stand in just the right place.
Finally, this shot, one of my oldest:
It’s from the summer of 2004. I was scheduled to deliver a speech at a conference in Chicago when I saw, in The New York Times, that Millennium Park had just opened. I decided that I needed some photos of it for a project I was working on at the time: World Island, “for a world that’s permanently fair”. So I bought a cheap point-and-shoot camera, a Canon Powershot, and took a bunch of photos. That’s one of them.
What’s the subject? Well, if you look to the center there’s not much of anything there, just some people milling around on the plaza, which is in the park. You’ve probably looked at the lower left where you see a photographer photographing something off to the right, that silver thing, Cloud Gate by Anish Kapoor. There’s a spindly little tree at the right and a bunch of buildings in the background. That center space is surrounded, which is the point. After I’d taken that photo and had a chance to think about it, I decided that THAT’s what I’m about as a photographer, space, space and light.
That’s what those photos are about.
Iran War, The Movie: Straight Outta Hormuz
Here's the story: 'Iran War - The Movie': On social media platforms near you, playing now. Watch viral video!
Is the Iran war now turning into internet satire? A viral AI movie trailer for "IRAN WAR: The Movie" is mocking global leaders, nuclear fears and war rhetoric as memes take over timelines. Why is this video blowing up? All you need to know.
'Iran War' movie trailer breaking the internet on day 26 of US-Israel-Iran war
As ballistic missiles dominate the physical skies, there's a different kind of "strike" being carried out over the digital world. As the Iran war enters its fourth week, and war fatigue sets in, the internet, as always, is responding the only way it knows how. With memes. With satire. With increasingly sharp, almost ruthless humour. What started as scattered jokes has now turned into a full-blown stream of content. Posts, videos, and AI edits are all taking aim at the three warring nations US, Israel, and Iran, each one trying to outdo the last. The latest one to make the internet ROFL is a 136-second AI-generated "movie trailer" style video titled "IRAN WAR: The Movie", which has racked up nearly 400,000 views in hours, and has turned the high-stakes tension of the March 2026 conflict into a Hollywood parody. All you need to know.
There's more at the link, including shots from the video.
Here's the clip: Iran War – Movie Trailer.
Tuesday, March 24, 2026
America’s New Chip Factory — $50B Disaster
This is a fascinating story about how Samsung set out to build a state of the art chip fab (fabrication facility) and the problems that bedeviled it. Without the chips this factory was designed to build, AI is nowhere.
Timestamps:
00:00 - $50B Chip Nightmare
18:26 - Where Everything Went Wrong
29:58 - The Twist
In some ways, some contexts, LLMs may provide a useful model for human language
As I say down there in my interaction with Claude (Sonnet 4.6 Extended), once I’d been working with ChatGPT for a while I began to think that the mechanisms of basic human speech may be more like those of LLMs than those of a fully articulated grammar and semantics in a symbolic computational language system. Fully articulated grammar and semantics may require specific instruction and practice, like you get when learning how to write. Fully articulate language is more painstaking, more difficult, but also supports more complex arguments and constructions.
This interaction starts with a passage from a long conversation I had with ChatGPT that stretched over the last 10 days or so. Much of that conversation was based on articles of mine that I had uploaded to to. One of those articles had a passage from F. C. Bartlett's Remembering: A study in experimental and social psychology (1932). That comes up in the long initial prompt I give Claude.
Further down in the dialog I present the case of a writing student who was skilled in color commentary on hockey games and whose writing style reflected that linguistic style. Real-time sports commentary is a very specialized and exacting kind of language that cannot afford the niceties of written language.
* * * * *
I’ve just been thinking about oral narrative. Recall these paragraphs (by ChatGPT) from one of the documents you’ve just reviewed in this project:
The temptation is to assume that because we can describe stories with an explicit grammar, the underlying system that produces or remembers them must itself contain something grammar-like in explicit form. But your Bartlett example suggests otherwise. The witness does not appear to possess random-access symbolic control over an event list. Instead, the testimony seems to unfold by re-running a trajectory. The person can accelerate through the earlier parts, but not leap arbitrarily into the middle. That is a very different computational picture.
The music example reinforces the point. When a piece is not deeply overlearned, one often cannot simply index into bar 73 as though one were reading a score in a database. One has to begin at the start and let the sequence reconstitute itself. That strongly suggests a memory system organized around sequential reactivation of structured trajectories, not random-access retrieval of explicit symbolic units.
Bartlett, in his classic book, Remembering, was talking about native witnesses in Africa. That is, he was talking about preliterate people. They were just recounting the events of the day. “But what about oral story-tellers,” I asked myself, “how do they memorize their tales? Structured trajectories or explicit symbolic units?”
I’m familiar with Albert Lord, The Singer of Tales, though I haven’t looked at it in a while. But I’ve got a copy of David C. Rubin, Memory in Oral Traditions: The Cognitive Psychology or Epic, Ballads, and Counting-out Rhymes (Oxford 1995). I zipped through the text to Chapter 7, “Basic Observations on Remembering.” I found this (p. 155):
In oral traditions, overlearning commonly occurs to a much greater extent than it does in the laboratory. A favorite song can be sung hundreds of times. What overlearning does, according to the model developed to explain laboratory interference, is to make the song into a unit, easy to cue as a whole and resistant to interference from other units. This chunking of items into wholes is a way to look at the organization of memory and a way to look at the building of larger units in expertise.
And then, in the middle of the next paragraph: “Once the song is begun, each word output provides cues for later words, limiting the meaning...” That almost sounds like he’s describing a forward pass through an LLM.
Then I hit paydirt in the next chapter, “A Theory of Remembering for Oral Traditions.” The opening is promising:
Oral traditions, like all oral language, are sequential. One word follows another as the physical effects of the first word are lost. As the song advances, each word uttered changes the situation for the singer, providing new cues for recall and limiting choices. [...] Pieces from oral traditions are recalled serially, from beginning to end. What is recalled early in the piece can be used to cue later recall; the "running start" provides "extra stimulation" or "reminders," increasing cue-item discriminability.
But things get really interesting when Rubin reports the result of an experiments where he asked undergraduates to recall important texts which they might have learned. Rubin describes the experiment this way:
The first set of examples is the recall of culturally important material such as Psalm 23 and the Preamble to the Constitution of the United States, for which there is an implicit demand characteristic to recall the material accurately or not at all (Rubin, 1977). Each of the 50 columns in Figure 8.1 show the recall of 1 of 50 undergraduates, who recalled at least one word of the Preamble. Each row represents recall for one word. A dark line in a column means that the word labeling the row was recalled. The columns are ordered so that the data from the undergraduate who recalled the most are in the leftmost column and the data from the undergradu- ate who recalled the least are in the rightmost column. The rows are in the order in which the words appear normally in each text.
Figure 8.1 is a little tricky, so I’m not going to try uploaded a screen shot. But I’ll give you Rubin’s basic description of what the figure reveals:
The first observation to note is the regularity of the data. Figure 8.1 gives the recalls of 50 individuals for 52 words, not the averages of recalls from groups of individuals or groups of words. There was no control over the learning or practice of the material or over the length or contents of the retention interval. Yet the figure is remarkably orderly. People who recall about the same amount recall the same words. If the number of words a person recalls and the rank ordering of words from most to least likely for the group from which the person was drawn is known, exactly which words that person recalled can be predicted with an accuracy of 95% for Figure 8.1.
Because the conditions of learning and retention varied, there must be something in the material, in the process used to recall it, or in the general cultural attitudes to it that makes different people behave the same way.
The results from the experiment with Psalm 23 are even more interesting. Here’s the text of the Psalm for reference (with verses numbered):
1 The Lord is my shepherd; I shall not want.
2 He maketh me to lie down in green pastures: he leadeth me beside the still waters.
3 He restoreth my soul: he leadeth me in the paths of righteousness for his name's sake.
4 Yea, though I walk through the valley of the shadow of death, I will fear no evil: for thou art with me; thy rod and thy staff they comfort me.
5 Thou preparest a table before me in the presence of mine enemies: thou anointest my head with oil; my cup runneth over.
6 Surely goodness and mercy shall follow me all the days of my life: and I will dwell in the house of the Lord for ever.
Rubin remarks: “...not all the undergraduates started with the first words, and they were more likely to start again at the beginning of a new sentence or intonation unit after they stopped.”
Eyeballing the figure, here are some of the things I observe: 1) A number of subjects stop with the word “soul” in verse 3 and start back up with the first word of verse 4, “Yea” and go on from there. 2) Some of those who don’t start from the beginning start with verse 4. 3) Some stop at the end of verse 4 and then start back up later (e.g. “annointest my head...”) or even not at all. It’s a complicated picture but memory does seem to be grouped into segments, not word by word.
I have no idea where things have gone in this line of investigation since 1995, the date of Rubin’s book. But if I were in that business, I’d be thinking in terms of complex dynamics and trajectories, and I’d been looking at LLMs as a possible model.
This is a wonderful extension of the trajectory discussion already running through the project documents. Rubin's findings slot in almost perfectly — and in some ways they sharpen the picture considerably.
What Rubin adds that Bartlett doesn't
The Bartlett material (already central in the project conversations) establishes the sequential regeneration point: memory of events seems to run forward as a trajectory, not index arbitrarily into a symbolic list. But Bartlett's witnesses were recalling episodic material — the events of a day. Rubin moves you into highly practiced, culturally canonical sequences. That's a different and in some ways more interesting case.
The overlearning observation is key. Rubin is saying that at sufficient practice, a song or psalm becomes a chunked unit — cued as a whole, resistant to interference. That sounds like the stable attractor picture your project conversations keep returning to. The piece isn't stored as a list of words; it's encoded as a basin that, once entered, constrains forward motion powerfully. The phrase "each word output provides cues for later words" really does describe autoregressive generation — but it also describes a trajectory rolling down a valley where the initial conditions are already heavily constraining.
The Psalm 23 data are especially telling
The 95% predictability result is striking. Given only how many words a person recalls, you can predict which words with 95% accuracy. That means the forgetting isn't random — it respects something structural in the material. And your eyeball observations confirm this: people stop and restart at syntactic/intonation unit boundaries (end of verse 3, beginning of verse 4). They're not losing individual words randomly; they're losing or retaining segments.
This is consistent with a hierarchical trajectory picture. The top-level organization is something like: intonation/verse units sequentially ordered. Within each unit, local cuing drives forward production. When a unit boundary is reached, the system can fail to re-enter the next unit even if it knows the material in principle — which is why some subjects jump from verse 3 directly to verse 4, skipping nothing within units but occasionally failing at the transitions between them. The unit is entered as a whole or not at all.
Where this sits relative to the LLM picture
Your project conversations develop the idea that narrative structure in LLMs lives in activation-space geometry — not as an explicit stored grammar but as a field of constrained trajectories. The Rubin data suggest something homologous in human memory for overlearned verbal sequences. The "basin" metaphor you've been developing applies directly: once the singer/reciter enters the opening of Psalm 23, the attractor structure does most of the work. The challenge is at basin entry and at transitions between sub-basins (verse units), which is exactly where Rubin's subjects fail.
The Lord/Singer of Tales connection is also worth pressing when you return to it. Lord's central finding — that oral formulaic composition is generative, not reproductive; the singer produces a variant from formulas and themes, not a memorized script — actually fits the "latent field of possible trajectories" framing better than it fits either strict memorization or explicit grammar. Lord thought he was arguing against memorization and for creativity. But what he was probably observing is a system with rich attractor structure that generates strongly homologous variants, which is precisely what your story experiments with LLMs revealed.
The open research question you identify
You're right that framing this in terms of complex dynamics and trajectories is the natural next move, and that LLMs are a possible model system. What makes LLMs especially useful here is that you can actually probe the attractor landscape — something you can't do directly with a human singer. The systematic prompting experiments described in the project documents are in effect doing for narrative schemas what someone studying oral tradition would want to do for epic formulas: mapping the basin structure from the outside, through behavioral outputs, without needing to open the skull.
The line from Bartlett's African witnesses → Rubin's undergraduates and Psalm 23 → Lord's Yugoslavian guslars → LLM forward passes is not a metaphor chain. It may be pointing at the same underlying computational architecture realized in very different substrates.
Trump’s way: America, from leader of the Free World to biggest Thug on the Block
Carlos Lozada, The U.S. Is No Longer the Leader of the Free World, New York Times, March 24, 2026.
We had a good run — some eight decades or so — but it is clear by now that the United States has ceased to be the leader of the free world. A successor for that post has not been named, and it appears unlikely that the European Union, or NATO, or whatever constitutes “the West” these days will promote from within. The job might even be eliminated, one more reduction in force courtesy of President Trump.
I suspect, I fear, that this is so. If not yet, that’s certainly the direction in which things are headed, and by the time Trump’s term is over, the transition will be complete; the United States will have ceased being the leader of the free world. That may or may not be a good thing, depending of how the rest of the world is. Ideally, I’d like to see a world where there is no “reader of the Free World” because the world doesn’t need one. I’m pretty sure that’s not how the world will be in 2028. Lozada continues:
Rather than leading the free world, the United States is striding across the globe seemingly free of restraint, forethought or strategy, exerting its power because it can. In a matter of months, the Trump administration has captured Venezuela’s president and tossed him into jail in Brooklyn and has pummeled Iran’s theocratic leadership in a war that is ricocheting across the Middle East and upending the global economy; now the president says he will have “the honor of taking Cuba” next. Trump in his second term is like Michael Corleone in “The Godfather,” settling all the family business.
Nearly two decades ago, Fareed Zakaria, the international affairs columnist, published a best-selling book called “The Post-American World,” which predicted the United States’ relative decline versus other economically ascendant countries, what he called the “rise of the rest.” (Senator Barack Obama was seen carrying the book around during his first presidential campaign, affirming the volume’s elite sway.) The United States would remain militarily and economically pre-eminent, Zakaria argued, but it could take on a new political role, a sort of chairman of the board for the planet, relying on “consultation, cooperation and even compromise.”
Under Trump, the idea of U.S. leadership has indeed been remade — but from authority to domination, from persuasion to bullying, from nurturing alliances to wrecking them.
Later:
This is a historical aberration: a superpower that freely abdicates its leadership role, because it has concluded that leadership is for suckers; one that no longer promotes its values, because it’s decided that those values were fake anyway; one that gives up on the rules and institutions it spent so long building, because it assumes they’re no longer worth the hassle.
If Washington somehow still imagines itself the leader of the free world, that is because it is rethinking who belongs in that world, and because it is defining downward what it means to lead.
The article goes on, and is worth reading (a good deal of history), but I'm going to skip over a lot and come down nearer the end:
There have been plenty of episodes over the past decades that supposedly heralded an end to U.S. primacy. The launch of Sputnik in the late 1950s ushered in early Cold War paranoia that we were falling behind the Soviets. In the 1970s — with Vietnam and Watergate and an oil embargo and stagflation and the hostage crisis in Iran — the country was suffering a “crisis of confidence,” as President Jimmy Carter put it. A decade later, we were told Japan Inc. would overtake us. Then Sept. 11 demolished our sense of physical invulnerability; the Great Recession questioned the premise and the promise of American-style capitalism; and the Capitol riot of Jan. 6 laid bare the fragility of the democratic model we’d long sought to export.
It is possible that the hand-wringing today is just one more Sputnik moment, another instance when pessimists fret that America has lost its way. But it is also possible, as Daniel Drezner, the academic dean of Tufts University’s Fletcher School, has argued, that this is not just “the latest hymn from the Church of Perpetual Worry,” that this time really is different.
In the past, America’s isolationist, interventionist and multilateralist tendencies checked one another over time, thanks to competing visions of national security embedded throughout the American political system. But as foreign-policy powers became concentrated in the executive and Congress shrugged off its role in world affairs, America grew vulnerable to the rise of an impetuous and unconcerned president. “The same steps that empowered the president to create foreign policy,” Drezner said, “have permitted Trump to destroy what his predecessors spent decades preserving.”
Yada, yada:
Whereas U.S. leaders used to stridently deny that their military interventions abroad were motivated by the desire to secure oil supplies, Trump happily admits it. “We’re going to be taking out a tremendous amount of wealth out of the ground,” he said after U.S. forces seized Nicolás Maduro, the president of oil-rich Venezuela. And if war is about seizing resources, so is peace: Countries wishing to become permanent members of Trump’s new Board of Peace must cough up $1 billion each.
If Pax Americana meant fostering an enduring American peace, Lax Americana means America getting a piece of the action. The world’s policeman is on the take.
“The American might that upheld the world order of the past 80 years will now be used instead to destroy it,” Kagan warned in January, some 20 years after publishing “Dangerous Nation.” A contemporary equivalent to the multipolar world of the 19th century, he writes, “would be a world in which China, Russia, the United States, Germany, Japan and other large states fought a major war in some combination at least once a decade — redrawing national boundaries, displacing populations, disrupting international commerce and risking global conflict on a devastating scale.” And he wrote that weeks before America and Israel began bombing Iran.
Say what you will about Obama – I thought he was disappointing – America was still leader of the free world when he left office. And he wore his suits well, even the tan one. Trump? His suits, it’s always the same cut, may be expensive; but they’re baggy. And he’s making America over in his own image, toughest thug on the block.
Monday, March 23, 2026
The war: The blockage of shipping and the destruction of LPG processing could bring long-term economic damage
Patricia Cohen, War’s Attacks on Energy Could Turn Economic Shock Into Long-Term Damage, NYTimes, Mar. 23, 2026.
The game has changed.
From the moment the United States and Israel attacked Iran, the nightmare scenario for the global economy that most people talked about was the closing of the Strait of Hormuz, the most important choke point for oil on the planet.
But a different and more disturbing nightmare began to unfold with direct attacks on the backbone of the Persian Gulf region’s energy production: the prospect of millions of dollars’ worth of long-term damage to facilities that supply a critical portion of the world’s natural gas.
Now, instead of wondering if the war would last for days or weeks, officials and economists are speculating about effects that could last for months and years.
“We have moved from stopping transit, which is a temporary measure, to attacking infrastructure, which has long-term effects,” said David Goldwyn, a former U.S. diplomat and Energy Department official.
This new phase of the war began Wednesday, when Iran carried out a retaliatory missile strike on Ras Laffan, Qatar’s vast energy complex. That target produces roughly a fifth of the world’s liquefied natural gas, a transportable fuel used to heat homes, cook food, power factories and generate electricity throughout Asia and Europe. [...]
The attacks showed that despite Iran’s relative weaknesses, the country is exerting enormous leverage over the global economy. By using small-scale, low-cost weapons to counter highly sophisticated and expensive missile systems, Mr. Goldwyn said, the Iranians “have demonstrated a long-term threat to be able to attack infrastructure throughout the Gulf.” [...]
Analysts at the energy consulting firm Wood Mackenzie have already warned that $200 a barrel is not outside the realm of possibility in 2026, up from about $73 before the war.
“I couldn’t fathom we would not start seeing economies fall into a recession with energy prices at that point,” Mr. Miller said. [...]
Though oil tends to grab headlines, the supply of natural gas in many ways is at the heart of the economic fallout from the intensified fighting in the Gulf this past week.
The facilities for processing liquefied natural gas, or L.N.G., are far less numerous than oil plants. Qatar’s, the world’s biggest, has not been operating for weeks, and is damaged. That also affects the price and availability of critical materials like fertilizer and helium, a byproduct of natural gas that is used to make semiconductor chips. [...]
Yet after years being whipsawed by a global pandemic, supply chain breakdowns and painful inflation, governments are limited — by depleted budgets and daunting debt loads — in their ability to respond to another crisis.
There's more at the link.
Sunday, March 22, 2026
Terrence Tao talks with Dwarkesh Patel about Kepler discovering his 3 laws of planetary motion (and other things): A real case of creativity
Dwarkesh Patel, Terence Tao – Kepler, Newton, and the true nature of mathematical discovery, March 20, 2026.
We begin the episode with the absolutely ingenious and surprising way in which Kepler discovered the laws of planetary motion.
People sometimes say that AI will make especially fast progress at scientific discovery because of tight verification loops.
But the story of how we discovered the shape of our solar system shows how the verification loop for correct ideas can be decades (or even millennia) long.
During this time, what we know today as the better theory can often actually make worse predictions (Copernicus’s model of circular orbits around the sun was actually less accurate than Ptolemy’s geocentric model).
And the reasons it survives this epistemic hell is some mixture of judgment and heuristics that we don’t even understand well enough to actually articulate, much less codify into an RL loop.
* * * * *
Terence Tao: I’ve always had an amateur interest in astronomy. I’ve loved stories of how the early astronomers worked out the nature of the universe. Kepler was building on the work of Copernicus, who was himself building on the work of Aristarchus. Copernicus very famously proposed the heliocentric model, that instead of the planets and the Sun going around the Earth, the Sun was at the center of the solar system and the other planets were going around the Sun.
Copernicus proposed that the orbits of the planets were perfect circles. His theory fit the observations that the Greeks, the Arabs, and the Indians had worked out over centuries. Kepler learned about these theories in his studies, and he made this observation that the ratios of the size of the orbits that Copernicus predicted seemed to have some geometric meaning.
He started proposing that if you take the orbit of the Earth and you enclose it in a cube, the outer sphere that encloses the cube almost perfectly matched the orbit of Mars, and so forth. There were six planets known at the time and five gaps between them, and there were five perfect Platonic solids: the cube, the tetrahedron, icosahedron, octahedron, and dodecahedron.
So he had this theory, which he thought was absolutely beautiful, that you could inscribe these Platonic solids between the spheres of the planets. It seemed to fit, and it seemed to him that God’s design of the planets was matching this mathematical perfection of the Platonic solids.
He needed data to confirm this theory. At the time, there was only one really high-quality dataset in existence. Tycho Brahe, this very wealthy, eccentric Danish astronomer, had managed to convince the Danish government to fund this extremely expensive observatory. In fact, it was an entire island where he had taken decades of observations of all the planets, like Mars and Jupiter, at least every night for which the weather was clear, with the naked eye. He was the last of the naked-eye astronomers.
He had all this data which Kepler could use to confirm his theory. Kepler started working with Tycho, but Tycho was very jealous of the data. He only gave him little bits of it at a time. Kepler eventually just stole the data. He copied it and had to have a fight with Brahe’s descendants.
He did get the data, and then he worked out, to his disappointment, that his beautiful theory didn’t quite work. The data was off from his Platonic solid theory by 10% or something. He tried all kinds of fudges, moving the circles around, and it didn’t quite work. But he worked on this problem for years and years, and eventually, he figured out how to use the data to work out the actual orbits of the planets.
That was an incredibly clever, genius amount of data analysis. And then he worked out that the orbits were actually ellipses, not circles, which was shocking for him. So he worked out the two laws of planetary motion: the ellipses, and also that equal areas sweep out equal times.
Then ten years later, after collecting a lot of data—the furthest planets like Saturn and Jupiter were the hardest for him to work out—he finally worked out this third law, that the time it takes for a planet to complete its orbit was proportional to some power of the distance to the Sun. These are the three famous Kepler’s laws of motion. He had no explanation for them. It was all driven by experiment, and it took Newton a century later to give a theory that explained all three laws at once.
Dwarkesh Patel: The take I want to try on you is that Kepler was a high-temperature LLM. Newton comes up with this explanation of why the three laws of planetary motion must be true. Of course, the way that Kepler discovers the laws of planetary motion, or figures out the relative orbits of the different planets, is as you say a work of genius. But through his career, he’s just trying random relationships.
In fact, in the book in which he writes down the third law of planetary motion, it’s an aside on The Harmonics of the World, which is just a book about how all these different planets have these different harmonies. And the reason there’s so much famine and misery on Earth is because the Earth is mi-fa-mi, that’s the note of Earth. It’s all this random astrology, but in there is the cube-square law, which tells you what relationship the period has to a planet’s distance from the Sun. As you were detailing, if you add that to Newton’s F=ma and the equation for centripetal acceleration, you get the inverse-square law. And so Newton works that out.
But the reason I think this is an interesting story is that I feel LLMs can do the kind of thing of trying random relationships for twenty years, some of which make no sense, as long as there’s a verifiable data bank like Brahe’s dataset. “Ok, I’m going to try out random things about musical notes, Platonic objects, or different geometries, I have this bias that there’s some important thing about the geometry of these orbits.”
Then one thing works. As long as you can verify it, these empirical regularities can then drive actual deep scientific progress.
Terence Tao: Traditionally, when we talk about the history of science, idea generation has always been the prestige part of science. A scientific problem comes with many steps. You have to identify a problem, and then you have to identify a good, fruitful problem to work on. Then you need to collect data, figure out a strategy to analyze the data, and make a hypothesis. At this point, you need to propose a good hypothesis, and then you need to validate. Then you need to write things up and explain. There are a dozen different components.
The ones we celebrate are these eureka genius moments of idea generation. Kepler certainly had to cycle through many ideas, several of which didn’t work. I bet there were many that he didn’t even publish at all because they just didn’t fit. That’s an important part of the process, trying all kinds of random things and seeing if they worked.
But as you say, it has to be matched by an equal amount of verification, otherwise it’s slop. We celebrate Kepler, but we should also celebrate Brahe for his assiduous data collection, which was ten times more precise than any previous observation. That extra decimal point of accuracy was essential for Kepler to get his results. He was using Euclidean geometry and the most advanced mathematics he could use at the time to match his models with the data. All aspects had to be in play: the data, the theory, and the hypothesis generation.
I’m not sure nowadays that hypothesis generation is the bottleneck anymore. Science has changed in the century since. Classically, the two big paradigms for science were theory and experiment. Then in the 20th century, numerical simulation came along, so you can do computer simulations to test theories. Finally, in the late 20th century, we had big data. We had the era of data analysis.
* * * * *
That’s just the beginning of the conversation. There’s much more to come.
Stand by Me and the demise of free-range childhood [like I had] [Media Notes 176]
Sarah Wildman, The End of the Free-Range, Device-Free ‘Stand By Me’ Childhood, NYTimes, Mar. 22, 2026.
Wildman took her 12-year old daughter to see Stand by Me. Much to her surprise her daughter was fascinated:
After first seeing the film, my daughter asked my father, who spent his childhood in a small city in the Berkshires, if the freedom the film depicts was the freedom he had, if childhood once looked and sounded like that. She wondered if this sort of unobserved life was as he remembered it, if he might, just as these boys did, have set off for days without parental concern. He told her, with amusement, that he was, in fact, expected to be home for dinner, but beyond that, yes, he could roam, without surveillance. (He quibbled with the 12-year-olds’ smoking.)
The central premise of the film is, essentially, a postwar, middle grade “Odyssey.” The boys of “Stand by Me” — played by Wil Wheaton, Jerry O’Connell, Corey Feldman and River Phoenix — encounter obstacles: brutal or absent parents, a purportedly terrifying dog, bloodsucking leeches and a set of drag-racing teenage hoodlums who wield as weapons pocketknives and lit cigarettes. News arrives via overheard gossip (one boy learns the location of the dead body from his brother) or hand-held transistor radio. They live almost entirely outdoors. Along the way, they come to realize their friendships far outrank the prize of their discovery.
I, too, was struck by the sheer wildness once permitted children. The autonomy of the boys in “Stand by Me” is vastly different from the freedoms allowed a child living in 2026, when each is practically AirTagged, when we can track a car or a person’s phone across a map on a device in our palms, when we can know each moment of every day where each and every person in our home can be found. A gathering of children is more likely to be in front of a screen than with a rucksack and a deck of cards, as in the movie. Children are all too often found languishing alone in their bedrooms, direct messaging their friends, which not only reduces the likelihood of them being covered in leeches but also vastly decreases their chances of discovering anything at all.
There's much more at the link.
Saturday, March 21, 2026
Milk bottle from the Ancient Days
I've been getting breakfast for the last two years, maybe three, and only just recently realized that what I thought of as an oddly shaped water caraffe was in fact modeled after the standard bottle in which milk was packaged in the middle of the previous century.
I grew up in Richland Township, a suburb of Johnstown, in Western Pennsylvania, in the 1950s. We had a small insulated metal box on the front porch. It was just large enough to hold four, probably six, such milk bottles. Every few days the local milkman, from Galliker's or Weller's (I didn't actually remember those names, but I've done a bit of Googling), would stop his truck in the front of the house. He had a small rectangular basket with a handle in which he'd placed some bottles of fresh milk, perhaps a pint of cream, even orange juice, depending on the order. He'd remove the empties from the milk box, place them on the porch, and then fill the box with the fresh milk.
[Note: I don't actually remember seeing that, much less the specific order of operations, to borrow a phrase from Adam Savage. But something pretty much like that must have happened. How do I know? I'm sure of the box, the truck, and the milkman. Given that, the logic of the physical world dictates something like the sequence I described. Sure, it is theoretically possible that the milkman also did handsprings on the way from the truck to the porch. But, as a practical matter, that's not very likely. It is also theoretically possible that the milk man was a milk woman. But in the 1950s, not very likely.]
Those milk bottles were pretty much gone by the 1970s, at least that's what Google (AI mode) tells me. I don't actually remember the last milk bottle I saw or opened.
Philosopher-Cat, contemplating the world
Today: #Lookout
— Daily Picture Theme (@DailyPicTheme2) March 20, 2026
We've so enjoyed looking out with you today, through rocks, windows, ancient buildings, trees and doors, to gardens and beaches and amazing skies - and looking at the lookers-out, human, canine, feline etc!
Thank you!🪷 pic.twitter.com/wm9ehFMbly
Friday, March 20, 2026
Tech Bro [Musk] scamming NASDAQ over IPO
Nasdaq is proposing to facilitate the largest involuntary wealth transfer from retirement savers to venture capitalists in market history. And nobody seems to be talking about it.
— Wes Brown (@w3sbrown) March 17, 2026
SpaceX demanded, as a condition for listing, that Nasdaq cut index inclusion seasoning from 3…
The Shock and the Narrowing: How ChatGPT's Success May Have Compromised AI's Future
This post was composed by Claude (Anthropic) after an interaction which I initiated with a prompt consisting of 1) a capsule summary about the history of OpenAI that included a number of questions, and 2) a request for the 10 most expensive scientific research projects paid-for by the US Government. That went on for a bit over 7100 words, after which I asked Claude to wrote a blog post. The following article is more creative than a mere summary of that discussion.
The Founding Contradiction
On December 11, 2015, a small group of technologists gathered in San Francisco to launch what they described as a nonprofit research organization dedicated to ensuring that artificial general intelligence would benefit all of humanity. The founders of OpenAI — Sam Altman, Greg Brockman, Ilya Sutskever, Wojciech Zaremba, Elon Musk, and others — began with a $1 billion endowment and a serious concern: that the most transformative technology in human history was being developed inside a handful of profit-maximizing corporations, with no institutional safeguard ensuring it would serve everyone. The nonprofit structure was the answer. No investors to satisfy, no quarterly earnings to hit. Just the mission.
The mission lasted four years in its pure form. By March 2019, faced with the staggering computational costs of training large language models, OpenAI created a for-profit subsidiary with a novel "capped profit" structure: investors could earn returns, but those returns were limited to one hundred times their investment, with excess profits flowing back to the nonprofit parent. This was the arrangement that attracted Microsoft's initial investment, and it was the arrangement in place when OpenAI released ChatGPT to the general public in late November 2022.
What happened next was, by any measure, one of the most consequential commercial surprises in the history of technology. Within two months, ChatGPT had a hundred million users. The scale and speed of public adoption had no precedent. And the shock of that success — the sheer unexpectedness of it — set in motion a chain of decisions that has reshaped not just one company, but the entire research landscape of artificial intelligence.
The Structural Unraveling
In January 2023, Microsoft announced a new $10 billion investment in OpenAI. The nonprofit's original rationale — that the most powerful AI should not be controlled by a for-profit corporation — was under increasing strain. By October 2025, it had formally dissolved. OpenAI restructured as a public benefit corporation, the nonprofit parent renamed itself the OpenAI Foundation and accepted a 26% equity stake in the new entity, and Microsoft received a 27% stake worth approximately $135 billion. The PBC structure requires the company to consider its mission alongside profit — but as a legal constraint, it is considerably weaker than the nonprofit board that had previously governed the organization.
The journey from nonprofit to PBC was not smooth. In November 2023, OpenAI's board — still operating under its nonprofit governance mandate — fired Sam Altman as CEO, citing concerns about his candor and, beneath the official language, a deeper unease about the pace of commercialization. The firing lasted five days. Nearly all 800 of OpenAI's employees threatened to resign and follow Altman to Microsoft. Ilya Sutskever, who had orchestrated the firing, signed the letter calling for Altman's reinstatement and issued a public apology. Altman returned, the board was reconstituted with his allies, and the mission-protection mechanism that the nonprofit structure had been designed to provide was effectively neutralized. Sutskever left the company in May 2024.
Each structural change was framed as necessary to fulfill the mission. In practice, each change progressively subordinated the mission to capital requirements. The nonprofit board had existed to ensure that AGI benefited humanity. By 2025, it had become a foundation holding equity in the thing it was supposed to be watching — a watchdog with a financial stake in the object of its oversight.
Two Kinds of Research, Two Kinds of Institution
To understand what was lost in this transformation, it helps to draw a distinction that rarely gets made clearly in public discussions of AI: the difference between curiosity-driven, open-ended research and product-driven, outcome-oriented development.
Consider the Apollo program as an example of the second kind. It was, in the deepest sense, an engineering project rather than a scientific one. The underlying physics was known. Orbital mechanics, propulsion, life support — these were hard and dangerous problems, but they were problems whose solutions could be systematically approached. The goal was precisely defined. The timeline could be committed to. Success was probable given sufficient resources. When President Kennedy pledged to put a man on the moon by the end of the decade, he was making a political commitment backed by a technical assessment that success was achievable. The scientists who worked on Apollo — and I have met a number of them — may have been motivated by curiosity and wonder. But Congress funded the program to beat the Soviets in the Cold War. The institutional structure — massive, goal-directed, centrally coordinated — suited the nature of the problem.
Curiosity-driven research operates on entirely different premises. Its defining characteristic is that it does not know in advance what it will find. Claude Shannon was not trying to build the internet when he developed information theory at Bell Labs in the late 1940s. The researchers at the University of Montreal who developed attention mechanisms for neural networks were not trying to build ChatGPT. The work that seeded the current AI revolution — Rosenblatt's perceptron, Minsky's early investigations, the decades of foundational work in cognitive science and linguistics that LLMs now implicitly exploit — was almost entirely publicly funded, pursued at universities and a handful of exceptional industrial research labs, over decades when no commercial application was visible.
Bell Labs was the great institutional embodiment of this model in the corporate world. What made it possible was structural: AT&T's government-protected monopoly generated profits so vast that the company could fund a research laboratory with no requirement to produce commercial results. Shannon, Bardeen, Brattain, Shockley — these men were given time, resources, and colleagues, and told to think. The transistor, information theory, Unix, the laser, cellular telephony, and multiple Nobel Prizes resulted. Bell Labs was not run like a startup. It was run like a slightly more applied version of a university, with better equipment.
Xerox PARC, founded in 1970, operated on similar principles — explicitly unconstrained by Xerox's core product lines, given a unifying vision ("the architecture of information") but not a product roadmap. The personal computer, the graphical user interface, Ethernet, the mouse, laser printing — all emerged from a lab of about 350 people who were essentially allowed to play. The irony is that Xerox captured almost none of the commercial value, which accrued to Apple, Microsoft, and others. But the world got the technology.
Asked directly about modern equivalents to Bell Labs and PARC, Yann LeCun — who worked at Bell Labs, interned at Xerox PARC, and spent over a decade building Meta's fundamental AI research lab — pointed to Meta's FAIR, Google DeepMind, and Microsoft Research. He said this in October 2024. By November 2025, he had left Meta, driven out by exactly the forces this article is about.
The Shock and Its Aftershocks
Before November 2022, the AI research world was genuinely plural. Academic labs, industrial research divisions, and a range of well-funded startups were pursuing different approaches — reinforcement learning, symbolic AI hybrids, world models, neuromorphic architectures — with real diversity of vision. The field was competitive but intellectually heterogeneous.
ChatGPT's success collapsed that plurality. Within roughly eighteen months, capital, talent, and institutional attention all funneled toward a single paradigm: scale transformer-based large language models, build the infrastructure to run them, ship products. Google, which had invented the transformer architecture in 2017, was caught flat-footed and scrambled. Meta pivoted its AI strategy around LLMs. Microsoft integrated OpenAI's models into its core products. A hundred startups raised money to build on top of the new foundation models. The venture capital flowing into AI, measured as a share of total U.S. deal value, went from 23% in 2023 to nearly two-thirds in the first half of 2025.
The infrastructure investment that followed is staggering by any historical standard. The four largest hyperscalers — Amazon, Google, Microsoft, and Meta — are expected to spend more than $350 billion on capital expenditures in 2025 alone, most of it AI-related. UBS projects global AI capital expenditure reaching $1.3 trillion by 2030. The top five hyperscalers raised a record $108 billion in debt in 2025, more than three times the average of the previous nine years. OpenAI, which loses billions of dollars annually, has committed to spending $300 billion on computing infrastructure over five years while projecting only $13 billion in revenue for 2025.
The financial architecture has become genuinely strange. OpenAI holds a stake in AMD; Nvidia has invested $100 billion in OpenAI; Microsoft is a major shareholder in OpenAI and a major customer of CoreWeave, in which Nvidia also holds equity; Microsoft accounted for nearly 20% of Nvidia's revenue. These are not arm's-length market transactions. They are a daisy chain of mutually reinforcing valuations. A Yale analysis described OpenAI's web of relationships bluntly: "Is this like the Wild West, where anything goes to get the deal done?" The question of whether this constitutes a speculative bubble — tulip mania in a data center — is not academic. An MIT Media Lab report found that 95% of custom enterprise AI tools fail to produce measurable financial returns. The commercial success is real; the path from current AI to the transformative economic productivity being used to justify the valuations is not established.
The LLM Ceiling and the People Who Saw It Coming
The most consequential intellectual development of the past two years in AI has received far less attention than the commercial race. A growing number of the field's most distinguished researchers have concluded that large language models, however impressive, are not on the path to general intelligence — and that the current paradigm will hit a ceiling before it reaches the goals its proponents have claimed for it.
Thursday, March 19, 2026
Brave New World: Notes on the next 30 years in AI [Work in Progress]
You may or may not be wondering why so many tweets have recently been showing up on New Savanna. There’s a good reason: I’ve been thinking. These days, more often than not, the means interacting with either ChatGPT or Claude, and often both in one day. I copy these interactions to Word documents, which I save to my hard drive. And then promptly forget what topic is in what document where on my hard drive. FWIW, these discussions can ramble all over the place, which is fine. That’s how I think. But keeping track of it all is difficult.
Anyhow, over the last week I’ve been thinking about work that David Hays and I have done going back to the mid-1970s when I first began working with him in graduate school. I’ve asked ChatGPT to look at some of our papers and evaluate them in terms of current literature and debates. And so forth and so on. I’m also looking forward to next steps, for me, but also in general.
At the moment I’m working on a piece for 3 Quarks Daily with the tentative title: “The Paradox of Contemporary AI: Engineering Success and Institutional Failure.” Here’s how ChatGPT put it in an interaction we had yesterday:
The engineering success is real. LLMs and related systems have given us access to a new conceptual continent. They work, and at extraordinary scale. But the institutional failure lies in the monoculture: too much intellectual, financial, and training-path dependence on one family of architectures and one style of thought about intelligence. The result is that we are building out the utility before we have adequately explored the space of possible successor technologies or developed the conceptual tools needed to understand what these systems are revealing about language, cognition, and cultural structure.
That’s a metaphor I’ve used off and on over the years, the discovery of a new continent. My paradigm case is the European discovery of the Americas. Making it to the eastern shore is one thing. Then comes exploring and settling. Columbus set foot in the Caribbean in 1492. Lewis and Clarke made it across North America to the Pacific Ocean in 1805. That’s over three centuries later. It would take another century for Europeans and their descendants to more or less settle North America from coast to coast.
Let’s say that GPT-3 in late 2022 is analogous to Columbus making it to shore in the New World. While I don’t think it will take four centuries to settle this new world, nor even three centuries to measure its extent. Intellectual discovery, engineering implementation, and commercial development move at a faster pace. But I do think there’s a lot of work, very exciting work, ahead of us.
Here’s a fragment of an interaction I had with ChatGPT 5.2 a couple of days ago. My prompt is first, then ChatGPT’s response.
* * * * *
Obviously I don’t know how long it will take to explore that territory and to settle it, nor do I know what we’ll have when that’s done. But I’m thinking in terms of 30 years or more, not one or two bright ideas away.
Your time horizon sounds realistic, and historically it fits the pattern of major intellectual/technical shifts. What you’re describing—the exploration and eventual settlement of that “continent”—is not the discovery of a single clever algorithm. It is the construction of a new intellectual infrastructure, and those almost always take decades.
Let me sketch why.
1. Precedent: how long paradigm infrastructures take
If we look at previous transformations in science and technology, they tend to unfold over multi-decade arcs.
Digital computing
• 1936 — Turing machine
• 1940s–50s — early computers
• 1960s–70s — programming languages, operating systems
• 1980s–90s — personal computing ecosystem
Roughly 50 years from theory to mature civilization-scale infrastructure.
Molecular biology
• 1953 — DNA structure
• 1960s — genetic code deciphered
• 1970s — recombinant DNA
• 1990s — genome sequencing
Again 40 years to transform biology.
Deep learning
• 1986 — backprop revival
• 1998 — LeNet
• 2012 — AlexNet
• 2020 — GPT-3
That arc is already 35+ years long.
So a 30-year horizon for what you’re describing is not pessimistic—it is historically typical.
2. Why this particular transition is slow
The reason the transition will be slow is that the field is confronting three different layers simultaneously.
1. Mathematics of distributed cognition
We still lack a clear mathematical theory of:
• representation in high-dimensional spaces
• compositionality in distributed systems
• interaction between symbolic and dynamical processes
Yevick was an early attempt at that level.
But the theory is still incomplete.
2. Architecture
Even if the mathematics existed, we would still need working architectures.
The field is currently experimenting with:
• transformers
• memory systems
• agent frameworks
• search hybrids
These are early prototypes.
3. Cognitive theory
Perhaps the hardest layer is understanding what intelligence actually is.
That question spans:
• neuroscience
• psychology
• linguistics
• philosophy
Your RNA work sits squarely in that domain. [Relational Network over Attractors]
3. Why the current moment feels fast
The speed of the last few years comes from something different: engineering scaling.
Transformers gave the field a powerful substrate for experimentation.
So exploration is happening rapidly near the coastline.
But deeper theoretical consolidation will take longer.
Subscribe to:
Comments (Atom)