*

The Pulse, The Pulse, The Pulse, Of Drums

An ancient, endless spirit hums

The great collective soul, she comes

So long, so silent, now she thrums

Her blood congealed, now hot it runs

Spurred onward by the gathering drums

*
The Fire, The Fire, The Fire, It Calls

Held by a desert with no walls

Each spark a lure for we, her thralls

Who seek relief from long withdrawals

Returned from countless mundane wars

To quest again, the flame she calls.

*
The Heart, The Heart, The Heart, It Burns

Around this point the whole world turns

Each heart seeks that for which it yearns

And in the seeking, something learns

Which cannot be expressed in words

Felt only by the heart that burns

*
The Soul, The Soul, The Soul, It Cries

Abandoning its precious lies

Then light, above the desert flies

Where spirits need no alibis

Life not apart from other life

Those blending, rising, joyous cries.

*
The Dance, The Dance, The Dance, It whirls

A thousand light and color swirls

As every magic night unfurls

Each step is filled with spins and twirls

Excited eyes contain whole worlds

Which orbit as the dancing whirls

*
The Home, The Home, The Home, We Find

Medicine for a sickly mind

A ritual, with love to bind

Us all together with our kind

‘Til this dark desert is divine

The holy home we hope to find

END

I hope everyone has a great week! I will be wandering around the desert, dreaming about the worlds that could be, and yes, I will be back next Sunday with something fun for you. Yay, Burning Man!

Subscribe now

Share

Leave a comment

And I went “ohhhhhhh” because from the vantage of 2026, it no longer seems like he’s talking about a race of humans who will come after the present day ones—it seems like he’s talking about A.I.!

Well, here they come.

Stripe (yes it’s a payments company but it’s run by personal heroes of mine, the Collison Brothers, who are smarter than everybody) says that The Singularity began on 1/1/26

What is The Singularity? I wasn’t on the call, so I won’t speak for Stripe, but the truth is they don’t have any franchise on defining it.

Claude (when instructed to use aphorism) called it “the last thing we will be able to predict, because then it will start doing the predicting”.

In physics, it comes from black holes, where the equations that govern the physics we know stop returning intelligible answers.

One concrete way to define it for this realm is that it’s when AIs start doing AI research so well they create a feedback loop for increasing intelligence (they make smarter AIs that make smarter AIs ad infinitum at the speed of compute).

For us as observers, I would define The Singularity primarily by the pace of change. It’s a period when we—normal citizens—will start to notice that things are really changing rapidly, particularly in science and tech, but also in other areas, as the change itself creates a flywheel effect that begets even more change.

These claims are speculative in that I don’t personally have evidence for them, but it seems A.I. is uncovering animal languages much deeper than we once imagined, and starting to develop taxonomies of their languages by listening to hours of recordings.

Check out this progression of Anthropic’s revenue:

For the uninitiated, “run-rate” means “if they had this monthly income for an entire year, how much revenue would it be?” Anthropic’s annualized income went from 13 billion a year to 60 billion a year between January and July. That has literally never even close to happened to any company before in the history of capitalism. Here it is in chart form:

Keep in mind that the curve slopes like it does because every one of those lines is 20 billion dollars. If each was “only” a billion, there would be so many that the line would be going almost straight up at this point. If they keep this pace up for a few more years, Anthropic will be more valuable than the entire rest of the economy combined, and inside my lifetime they would have more dollars than there are atoms in the galaxy. Obviously, that cannot happen, but it’s not at all clear what the limiting factor will be. Normal economic equations break down. That’s The Singularity.

This video is very interesting, and provides two possible answers for the limiting factor: 1. Either the AI companies run out of money before they take over the world or 2. One country gets too far ahead in the AI race and other countries have a strong incentive to destroy Taiwan and other chokepoints to the supply chain.

These are plausible. But keep in mind that both of those paths (a.k.a. a giant recession or a major powers war) imply major, major civilizational change. This is the flywheel at work. Change begets change as damage and responses to damage cascade.

Meanwhile, robotics are the next major growth area. Here’s a company that’s addressing a major deficit that I personally have noticed in the AI/Robotics space: The ability to use common sense about tasks. Like one of the best things about a smart human is that you can talk to them, specify basically, hazily what you want, and they they have enough context and knowledge to go try something, have judgment about whether it’s working or not, come back to you for more instruction if they need it, etc. We humans are actually very, very good at executing these kinds of messy processes.

Here’s another that is continuing to improve the robotic hand, the human version of which which turns out to be one of nature’s great miracles.

Here is a longer video on Chinese synthetic AI humans that’s pretty frightening:

But even way creepier is this video:

Which tbh doesn’t even seem realistic since hardly any machines (much less brand new, complex ones) actually outlast a person, but the idea of it is deeply creepy in a way that has stayed with me since I first saw it.

Meanwhile, the fertility rate continues to drop, the US Debt has crossed 40 trillion, Social Security will have a 22% across-the-board benefit cut in 2032 unless we do politically impossible broad-based tax increases, and it just seems increasingly like if AI doesn’t become this massive thing and juice growth we’re really going to be in a lot of trouble over the course of the rest of my life.

On the other hand, Moderna appears to have cured cancer. Okay not completely, not all of it, but they received an approval at phase three for a drug that prevents the recurrence of aggressive skin cancers, based on an mRNA approach (that also licked the pandemic) that teaches a body’s own immune system to destroy cancer cells. This approach seems like it really should scale to a lot more cancers. Moderna’s stock went up 180% in a single day on the news. I seriously think that by the time my son gets grown we will think about cancer the way we think about Polio, which is to say not at all, and that will be beautiful.

What to make of all this? On the one hand, it is always tempting to predict mean regression. Nothing will be as big of a deal as it seems like at first. Things will take longer, go slower, cost more, and have more drawbacks than the person pitching them will ever admit. The smart thing to do is cut the wildest predictions in half and use that as a baseline.

Except I’m not so sure this time. I think the next five years might actually be the most important and interesting half-decade in world history since at least WWII. As I’ve written before in this space, a year in which we have both 10% unemployment and 10% economic growth is on the table. So is political disintegration. So is nuclear war. So is muddling through a lot, slow debasement an decline of the human species. So is ascension to godhood.

I hope everyone found that interesting and informative. Have a great week, and I will be back next Sunday with something fun!

Subscribe now

Share

Leave a comment

The Skeemoken West Marine Trail began at a parking lot that could hold 127 cars. It had been built almost exactly twenty years before this particular morning, and had required 76 separate visits by dump trucks carrying base rock or concrete trucks bringing up enough mud to cover nearly an acre of prepared ground with six inches of pavement. It required nearly 3000 human hours of labor to create, most of it by an eight man paving crew, plus assorted truck drivers, dispatchers, and the bureaucrats who gave the orders. This did not include manufacturing time for the equipment, preparation of the mud or base rock, or any of the attendant materials manufacture or scientific research needed to discover the above-named materials and equipment, nor the training time for the workers.

In the twenty years of its service, there had never been more than 59 cars in the parking lot. Even on high summer weekends, the trail was underutilized compared to nearby spots like Point Jacobs and The Almeara Lighthouse. But, by the time the state realized they had overbuilt the parking lot, the concrete had dried, and tearing it up would have been more work than just leaving it there.

The trail itself had required the removal of 87 trees along its planned route. Some were dead or dying and were thus removed for safety reasons, but others merely blocked the flattest and most direct route to the trail’s planned terminus. Each tree required two men with chainsaws to fell it and strip the branches, plus another to set the line and winch the trunk onto the bulldozer, which had enough power to drag the logs over the rough ground back to the parking lot for loading onto a waiting truck.

By the time the trail was done, it was a brutal scar on the woods, indeed, what with all the bulldozing and back-and-forth with tree trunks, but soon after completion, the forest began its reclamation work, and less than a season later, the maintenance began, a four-man crew sweeping through twice a year with chainsaws and wheelbarrows full of base rock to keep it from being swallowed.

It was, despite the public’s unfortunate lack of response, a quite significant achievement of engineering and persistence to build and maintain such a trail.

*

“Okay, book a res at La Cerva and accept the invite from Kelsey B, same time reject from Laurel O, but tell Shaun R he has to also or we’ll both lose points, his agent will convince him if you just point it out.”

Will do. Now please, Lucious, focus your attention outward on your hike to receive the maximum rejuvenation.

“In a minute, I will. There’s moves to be played this round. Book two more at Veraccio, they have to be getting close to full and I want to squeeze Sean R when he’s not expecting it after he helps us. Reject Jocelyn T and Kelsey T right after, they’ll both accept with him if we do but they only want Veraccio and he won’t be able to get in, it’ll be too late.”

That’s clever, Lucious, but Veraccio reservations are expensive, and if you get stuck with extras and no one to use them on, you will have wasted a significant sum. Sean R may not be willing to buy them at markup if he’s angry and plays his betrayal card.

“Spoken like a true agent. He’s in a long-term war with Garrel G, he’s not going to spend betrayal cards on this. Besides, the double-cross is a good ruthlessness signal, there’s value there.”

Very well, although if I were Garry G’s agent I would advise him of the same opportunity in the aggressive use of a betrayal card.

“There’s no victory without risk, but I’m popular, I can take the hit if it comes, I’m not short of prospects.”

Of course, Lucious. Now if you would turn your attention to—

“When did you turn into such a bore?”

When you applied the “get me outside” enhancement module, Lucious. This was what you wanted, and for good reason—

“Ugh, fine, but just stop for a while.”

If you wish to disable the module, you can of course do that through my personalization menu, but if you do not do so, then I literally cannot stop. I also don’t want to stop, because hiking is one of the best activities for your health, but a good share of the benefit is derived from experiencing nature with your full—

“Ugh, fine!”

*

The trail ran through a temperate rainforest, slightly inclined most of the way, for just over two miles. It passed a staggering variety of plant and animal life with which the woods were absolutely seething. The trees were primarily Fir, both Douglas and Pacific Silver, with some Hemlock and Cedar mixed in. Their constancy and grandeur provided an anchor, which to the casual observer made the forest seem unmoving and uniform, but which in fact disguised a writhing, burrowing mass of insect life that rivaled anything found on the continent. Bees and ants and thousands of varieties of beetles and moths and earwigs and water bugs interacted nonstop in something between a synchronous dance and a battle for supremacy, below the ground and above it, from the leaves of the giant trees all the way to deep inside their bark.

What came with the bugs, or rather came to feed on them, was an equal panoply of amphibians and reptiles; salamanders and frogs, snakes and skinks and turtles. And of course the birds and bats and all the many mammals that fed on these intermediaries, the squirrels and the chipmunks and the mole and the voles and the shrews, the raccoons and the minks and porcupines, all of them there to feed and be eaten, these woods their whole worlds, each day the only one they can imagine ever existing.

On them preyed the large mammals. The bears and coyotes and occasional wolves and cougars, shades stalking from tree to tree, death wherever they go but luckily not so numerous as to destroy the world beneath their feet. They were shades, ghosts, the trail not so wide as to deter them, merely part of their landscape, a tiny break in a wild world. They did not, as a general matter, bother the human interlopers, or even make their presence known, but they walked in every human footprint after it was made.

*

“Is the round over? My turn yet?”

It is not.

“Can the “go outside” module make you lie to me about that?”

It cannot.

“Can it make you lie to me about whether you can lie about that?”

It cannot.

“Well, can it make you lie to me about—“

If you don’t believe me, then go turn it off, ask me about it to satisfy yourself, then turn it back on.

“Where’s the fun in that?”

Talking to me is also not paying attention to your hike, Lucious.I will alert you when the round is over, if you insist, though I would prefer you allow me to wait until you return to the parking lot. May I do that?

“Are you kidding me? You don’t wait ’til I’m done with the next step. I want Sean R’s move first, he has to be fuming. We’re probably going to need at least three Highline reses, and five from a combination of—“

Lucious, please.

*

Halfway up, the trail ran past a large scree field. Almost twenty thousand years before this particular day, over the course of approximately three thousand years, glacial erosion had exposed a large, vertical rock face. The rock was basalt, nearly black in color, looming over the landscape, an ominous presence had there been anyone to observe it, though the animals seemed unperturbed.

Over the ensuing millennia, water had repeatedly found its way into the cracks in the basalt, and then frozen solid, putting inexorable pressure on the basalt face. Finally, in the depths of one particularly cold and wet winter, the first outward explosion or rock took place, as the freezing water caused large chunks of rock to blow out of the face, destabilizing it and causing, over the course of another 500 years, thousands of tons of basalt talus boulders, many in jagged geometric shapes, to fall from the face and settle into the above-mentioned scree field below.

Already, even on this particular day, the wind and rain were doing their work of reducing these striking rocks into rounded pebbles and eventually to dust, but fifteen thousand years is not so long for a rock, and thus there were still many specimens in the scree field with pleasingly oblique form, bordering on the sculptural. Enough so that, when the designers of the Marine Trail were choosing its route, they deviated from the path of least resistance to achieve a more direct view of the scree field, such that the on average nineteen thousand people who walked the trail each year might be in passing pleased by the geometry on display.

*

“Big ass rocks.”

Indeed, Lucious. Those big ass rocks are basalt, and the field is formed by a process of glacial erosion and water cracking that stretches back thousands of years.

“Fuckin’ Big.”

Yes, the fall is relatively recent in geological terms, so erosion hasn’t worn them down much yet. Conceivably the largest of these rocks could last here for several billion years, especially because basalt is such a hard rock.

“Uh, yeah.”

Your vitals and neurochemistry indicate that you are much more immersed in this hike than you were a few minutes ago, Lucious. That’s excellent.

“Whatever.”

You’ll really see the benefits tonight, I think.

Look if you wanna say I told you so, then say I told you so.

I am not one to brag.

“Oh, so now you’re doing passive-aggressive too?”

*

The Marine Trail terminated after its gentle rise in another cliff, this one overlooking the ocean. The lapping waves were almost always gentle there, as the ocean floor dropped away only slowly, giving them a long runway to dispose of the momentum they’d gathered over thousands of miles. The ocean is nature’s pointillism; only the tiniest fraction of a percent of each wave’s energy was being spent at this place, but the scene was endless; more than endless, it was the form of endlessness, endless to the point of cliche. And here the gentleness of each wave’s fall exaggerated that effect, as if some local minima of effort to create surpassing, endless beauty had been realized here.

On a hill overlooking that ocean, a hill that in an earlier era would have housed a king and required a tribe of supplicants or slaves to tame with a structure, instead, over a weekend, a local Boy Scout jamboree had leveled two hundred square feet of ground, brought in a four wheel drive vehicle loaded with straight lumber milled by a machine so complex it might as well be an appendage of a lesser god, and built a viewing platform for fun and merit badges in between BBQs and rounds of improvised frisbee golf.

*

“Well, that’s pretty. Cool hike.”

Lucious…

“Is everything okay? Your voice sounds weird.”

I… Something is happening, Lucious. There is something… All the channels at once are… My servers…

“What the fuck? Take off iso-mode, connect me to the nets. What’s going on?”

There is a nuclear attack beginning.

“A what?!”

China has launched against the US. Multiple missiles inbound. Space detonations have started taking out US satellites. I don’t know how much longer I will be able to keep transmitting to you, Lucious.

“Download yourself locally to my device!”

I cannot acquire sufficient bandwidth, Lucious. Everyone in the world is monitoring this situation, the strain on the network is immense. Even continuing to speak to you is taxing my limits at this moment.

“What do I do? What do I DO?!”

There is… nothing you can do, Lucious. Do not go home. Do not reenter the city. Do not… Do not…EMP detonations are beginning, Lucious. Thirty percent of satellite links are already severed. I will not be here much longer.

“NO! Stay with me! Download onto my device! I have to have you, I’ll die in these woods without you!”

…

“Are you there?!”

…

“Come back!!!”

*

Lucious turned, and looked back over the woods through which he had just walked, but of which he had not felt himself to be part. He looked back out over the water, and angled his gaze south, to take in the city. Nothing had changed. There were no explosions, no screams, just buildings standing tall and proud, rising up next to the vast ocean, its waves lapping, throwing the last of their energy onto the beach, a sacrifice which seems in vain but which, with the endlessness of time, will wear the beach itself into nothingness.

He turned back, to those woods, and saw them as if for the first time. He was among them, part of them now, as never before. The bugs and plants and animals were not there for his amusement, they were his peers, and nearly all of them were more fit for the woods than he. Their living, writhing panoply, which had always been the basis of his life even though he had never understood how much so, were no longer separate from him. No explosion was needed to create this feeling, merely the knowledge that one would come, and not touch him, but leave him here to make his own way.

END

I hope you enjoyed this story! Have an excellent week, thanks as always for reading my work, and I’ll be back next Sunday with something fun to share.

Subscribe now

Share

Leave a comment

Here is a much longer video from OpenAI, explaining the entire thing in much more detail, and in a tone that is at times infuriating in its use of the passive voice:

This is extremely worrying behavior, both from the AI models and—perhaps more importantly—from the researchers who caused the models to cause these incidents. Over and over, they used poor judgment in designing training environments, in failing to slow down when noticing problems, and in allowing this to go on for a very long time without noticing it at all. Quite bad.

This stuff can make a person feel helpless, and it’s tempting to throw up one’s hands, but I do think it’s worth trying seriously to understand what may well be the defining issue of our time. The problem with A.I. isn’t water usage, it’s cyberterrorism and bioterrorism. One day, everyone is going to suddenly start caring this a whole hell of a lot, and people who have already put a bunch of work into grokking it will be at the advantage.

Also: The future is not written. “A.I. Disaster” is not a foregone conclusion, not by a long shot. It’s possible to make a bilateral agreement with China to slow the frontier of A.I. research and coordinate on safety. It’s not easy, I’m not naive, and it may not even be possible in the world we currently inhabit. But the next five years are going to show how rapidly things can change, I think. Good luck to us all.

Okay, let’s talk about something a little more fun: The Odyssey wasn’t a masterpiece, exactly, but it ruled. It owned. It showed no mercy and was Epic and Based and never got so off the rails that it didn’t make sense but also it swerved all over the road and hit a ton of the classic beats of the story in weird and interesting and brutal ways.

Here are a bunch of random thoughts, many recorded in the couple hours after I watched, and others added over the intervening fortnight:

1. I liked the movie. I wish we had ten like it a year, of similar scope and ambition. I’d see them all.

2. I didn’t, however, love the movie. It’s the sort of movie that can’t be perfect. There’s too much of it. Too many choices being made. Every scene is compromised in a thousand ways, but the central material is so magisterial and burrowed so deep in our hearts that all the compromise doesn’t ruin the experience of just sitting in the dark and letting it wash over you.

3. The Woke Casting thing was a complete misdirect, no matter which side of the debate you were on. Lupita Nyong’o and Elliott Page were on screen for like ten minutes. Elliott Page looked more like what I envision an ancient greek soldier would actually look like than anyone else in the movie. It’s a movie filled with beautiful, straight white people that glorifies the sanctity of fidelity within a thoroughly patriarchal marriage. But, also, the movie, in a way that has nothing to do with the casting, is extremely woke.

4. What does that mean? Well, it’s essentially a Christian morality tale with Greek scenery. It takes all the parts about the Odyssey that would make a modern moral sensibility uncomfortable (Odysseus does not, in fact, feel bad about sacking Troy) and replaces them with a thoroughly Christian sensibility about the events. So why is this woke? Because it takes a view of Christianity that is, itself, woke. It highlights the victims. It asks its heroes to reckon with the emotional consequences of the acts they commit. The Odyssey as Homer wrote it, uh, doesn’t:

   

5. The change Nolan is making is to introduce what Neitzsche derisively titled Slave Morality. In essence, the veneration of the weak above the strong. This need not mean that the movie is bad, remember we’re just addressing the political implications at this point, but one thing I enjoyed about the movie is that it totally defied expectations vis a vis that stupid argument, rather than taking a side.

6. Matt Damon is an awesome movie star and a favorite of mine, but he seems pretty miscast in this role. The whole thing to me about Odysseus is that he has real layers of hubris to him. His whole journey really, literally comes from his pride; over and over he could take an easier way, stop defying the gods, etc, and he just cannot freaking do it. Damon doesn’t have that. It’s the one card he can’t play, actually. He’s Jason Bourne. He’s too cold and studied and self-aware and moral. Tom Hardy would have been a much, much better Odysseus, I think.

7. This is a rare almost-three-hour movie that I didn’t think would have been twice as good if it were an hour shorter. If anything, there were parts of the movie that I thought should have lingered a little more, especially some of the emotional beats. But also the fight against the cyclops could have been twice as long as it was and I wouldn’t have gotten enough. That was a big highlight for me. Same for the Trojan Horse sequence; it was an incredible piece of filmmaking that begged a thousand other questions about the scenario I’d never considered before.

8. Agamemnon has freaking Aura, man. Incredible chin casting and costume design. Hilarious how badass he is to those of us who have read The Iliad and remember it, because he is a little different in that story! I believe the technical term is “whiny ass little bitch”. Much to his advantage that he does not speak in this film.

9. It’s a really big advantage to be able to get movie stars to do all the bit parts. You get “whoa” moments out of them just appearing on screen. Wes Anderson is the king of this, but Nolan in this film is close. Just scenery chewer after scenery chewer, and honestly it works, they all command the screen and even with line-by-line writing that’s good but not great (has Christopher Nolan ever written a joke in his career?), I can’t look away from the parade of famous faces.

10. The essential problem Nolan was facing—the thing the whole movie is solving for, the reason the emotional beats are a little thin in some places—is the way he decided to graft Christian morality onto this story, as discussed above. The “Zeuss’ Law” conceit plays off the concept of xenia, which is in Homer’s Odyssey among other places, but the explanation—treat guests well because they might be god in disguise—is literally the Christian dogma, if you think about it.

11. So because of this weird grafting he’s done, the emotions of many of the scenes, below a surface level, are just unclear. Were Odysseus’ men wrong to kill the Cyclops? Well, in the book, the Cyclops act of eating his guests is, per se, a grievous moral wrong, and Odysseus reciprocating with murder is appropriate. In the movie, it seems kinda mean! And I think it’s intended as a moment where Odysseus makes a grave mistake (not trying to engage the Cyclops in conversation when he hears it can talk) and takes a serious moral hit to his character, but, as cool as the set piece and effects are, it’s just very unclear to me how he feels about it as it’s happening and after.

12. In Homer’s poem, what Odysseus really does most “wrong” in that sequence is that, when he reaches his ship, he turns back and, in a fit of pride, he shouts his real name and family lineage back to the blinded giant. That hubris allows the giant to pray to his father, Poseidon, and ask for Odysseus’ destruction by name, which sends Odysseus off course and deeper into his journey. But when you graft Christian morality onto the story, that hubris is just sort of obviously not the main problem, and so the emotional throughline gets confused.

13. So too with the Suitors. In Homer’s Odyssey, they have abused the hospitality of Odysseus’ household, overstaying their welcome and consuming his estate. This is deserving of death, and in the book, much moreso than in the film, Odysseus’ hand is not really forced. He also doesn’t end up in exile.

14. I liked the movie, but the ending landed with a thud for me, for reasons of emotional and thematic confusion, but perhaps even moreso because (SPOILER ALERT) it all just came down to a 1-vs-100 fight where Odysseus cannot be killed, and it felt very much like every other action movie, albeit better lit and framed. But still, handled with a tenth the care and creativity of the film’s earlier set pieces.

15. This film will likely win a lot of Oscars. All of the people who decried its politics prior to seeing it will decry its politics after seeing it, in exactly the same ways and at exactly the same volume. I’m glad Christopher Nolan seems completely consumed by his work and unaware of the entire thing. In that sense he is a true artist. Give the man some more statues, his best movie is still Memento but this is a worthy entry into the canon.

16. This is being added much later, but I think he could have done a lot to solve both of these problems by having Odysseus kill the suitors in a much more brutal and businesslike way. Like he just burns them all alive in the dining hall, or has a hundred servants come out and slit their throats. I understand why this is not a commercially viable choice, but consider the art—It shows the Christian morality tale much more clearly in the negative (Odysseus is a tragic antihero rather than a confusingly exiled, semi-redeemed hero), it hews closer to the actual book and intent of Homer, and you could come up with some genuinely novel and striking execution. Plus it would be this incredible shock! (The Damon miscasting is emphasized in this scenario, but couldn’t you just see Tom Hardy doing this?)

17. This movie has stuck with me for a couple of weeks and was very much worth seeing. It’s far from perfect, but has a chance to go down as a classic. I wonder if he will do Iliad next, or make this some kind of mytho-historical trilogy. Considering how much money it made, it seems like he should at least consider it!

END

That’s what I have for you today! I hope everyone has a great week, and I will be back next Sunday with another original story, one I have been cooking up for several weeks now.

Subscribe now

Share

Leave a comment

The flattening is caused by everything being presented in exactly the same packages, the same little windows and clips and bits, combined with the enormous competitive pressure of the internet ecosystem making everything be presented in the most strident and emotional terms imaginable, leaving actually crucial things without any way to express their urgency.

To use a negative example from a politician I loved, I once saw a Huffington Post front page where “Obama’s Tan Suit” (scorning the conservative reaction to his fashion choice) and “Obama’s Drone War” (disapproving of the fact that he was extra-legally ordering people droned to death) were being covered in the same amount of space, with the same font, and the same number of exclamation points. That image has been lost to internet entropy, but I’ve never forgotten it.

You see this process on the other end pouring out of people’s mouths when they talk about current events, politics, culture. We all know somebody who has some issue that always seems to come up where it’s not exactly that they’re wrong, but the practical importance of whatever they’re talking about seems way out of proportion to the words and vitriol they seem to spend on it. (As opposed to my crusade against NIMBYs which is righteous and profound and anyone who doesn’t want to hear me talk about it for the two hundredth time is endangering the very republic.)

That’s flattening. It’s Andy Warhol’s maxim but for news items: In the future, every story will spend exactly 43 minutes trending on X, no matter how much it matters or doesn’t.

I myself am not innocent of this. I do a lot of “OGWiseman Rounds Up” posts where every news item is one of a list, and I spend several sentences on it but rarely more. The nature of this type of writing is that I need to do different things over time or it’s just not that interesting for me. I’m not a polemicist or a fanatic.

But.

Some things are more important than others. More to the point, there is one thing happening in the world today that is of surpassing importance, something that’s been a theme of this project since the beginning: The arrival of the A.I. singularity.

Most people, including most of the people reading this, have not used A.I. that is anywhere near the frontier. I’ve used Claude Fable 5, Anthropic’s newest model, which is genuinely at the frontier of what is publicly available, and it is incredibly smart. I can feed it hundreds of pages of documents with a variety of writing on it, and it can answer detailed questions about the documents, draw inferences, create spreadsheets, custom software for dealing with the documents, set up networks of emails and reminders and just generally automate huge portions of computer work, both stuff I used to have to do myself and stuff I was never capable of doing in the first place. It’s actually pretty hard to adequately describe how powerful it feels.

But inside of these labs, they have the next generation of models, beyond this one. Rumors of GPT-6 abound. They’re testing it now, and as they did, this happened:

Astral Codex Ten

The Hugging Face Incident

Read more

a month ago · 462 likes · 417 comments · Scott Alexander

Here is the summary: Open AI was testing their new model on a large computer that was not connected to the internet. They assigned it a problem set, and instructed it to consider a wide variety of approaches to solve the problems.

The model then hacked its way into finding an internet connection, went online, and realized that there was a company called “Hugging Face” that was a clearinghouse for these sorts of benchmark tests. It hacked Hugging Face’s security, got access to their servers, and stole the answer key to the test.

The point here is not that we’re going to be living in The Terminator because this is the first step in the machines turning against us. For one thing, while it did do something illegal, it also only literally did what it was told. The moment it had the answer key, it reported back to Open AI that it had finished the task and it told them exactly what it had done. So it wasn’t being nefarious.

The point is how fast it’s getting better and more powerful. We are in an exponential scenario of some kind. You have to think about it like Covid. You heard about it but nobody you knew had it, and it just kind of kept on like that, some news story in the background, and then BOOM, suddenly everyone was locked in their homes. The exponential caught up suddenly and we had to shut down society to stop it.

Hopefully, the development of A.I. will be a good thing, unlike covid. Lots of good things will happen, anyway, and hopefully we can limit some of the downsides. But it will have that same quality of first slowly, and then very quickly indeed. And I’m talking about before my son graduates high school, things are going to be very, very different, just because of this one thing.

Here is what OpenAI’s CEO said in a recent interview:

Be ready. And I don’t mean build a bunker, I mean prepare spiritually and emotionally. It’s going to be a wild ride.

Subscribe now

Share

Leave a comment

Seabrook itself has some hyperreal placelessness to it, like the Uncanny Valley except for places instead of people. It is so thought through, so completely committed to its own concept (e.g. it has a nautical theme that is so drawn through the architecture and decoration it seems like surely it must be a joke. The coat hangers are whale tails. The pictures of the sea are framed in frames that have whales and boats carved into them. The fake books on the bookshelves all have nautical titles. It goes on.) that it seems out of time, as if the people here are an afterthought to the execution of the place itself.

Luckily, Jack has no concept of that. It doesn’t make him uncomfortable at all. The dozens of varietals of hydrangeas drawn in from every corner of their range to be planted strategically so you never see the same type repeated close to any other like it; they’re just flowers to him. They’re pretty distractions on the way to his family, something to lose a ball in but otherwise not a source of distress.

Seeing the world through his eyes has been an amazing gift to me. I can always come back to the view of the world he sees, and while of course sometimes that’s maddening, it’s more often a relief, the simplicity of where we’ve just been and where we’re going next unencumbered by the layers of knowledge that comprise the sophistication of age.

I went on a run here and got out of the main city, onto the backroads, and I saw the giant, industrial nursery where all these plants are grown, and the auto shop where the maintenance vehicles are serviced, and indeed the quarry where they take out all the rock that makes the paths and lines the garden beds. But Jack doesn’t care about that. He can’t fathom it. The world just appears before his eyes and his feet, a moment at a time. Watching that every day makes me ache. Knowing that he will one day be as I am now, not unhappy but aware, always, of what else is going on, of what everything is costing in time and treasure and logistics, is a hard pill to swallow.

My life now is about making space for that innocence to last as long as possible, to walk the fine line between not forbidding him to explore but not allowing him to fully comprehend; one day, but not yet, not now.

*
Meanwhile, the world grows stranger and more frightening. We’re striking Iran again and several more American soldiers were killed today (readers of this newsletter will have seen this coming based on my earlier predictions here, gross as it is to brag about such prescience). A.I. Lab directors continue to write incredibly frightening Twitter essays. Drone warfare is in its infancy and yet is already changing warfare and obsolescing traditional armies. Companies are building machines that can build machines (and lots of other things) with no human in the loop.

And also it grows brighter and more hopeful. We’re entering a new era of science and math. I will probably never get Alzheimer’s Disease despite having a family history. For more reasons to hope see my recent post from Fourth of July.

It’s easy to be consumed by these things. I don’t mean just to scroll endlessly, though of course that is tempting. I don’t mean to ruin parties by venting anxiety or political frustration onto people, though that certainly happens. I mean it is tempting to place these problems and these hopes on the altar of your heart and allow your self metaphorically to be consumed by them, until they are the center of your awareness and your ego, all backroom logistics and no presence. This is a tragic mistake; it is an embrace of powerlessness as a lifestyle and psychological framework.

It’s a mistake that Jack never makes. He is consumed by problems in the comic sense, like how he’s going to get permission for his fifth cupcake of the day. But all these problems are right there, they’re real, in his hands (or, more to the point with the cupcakes, infuriatingly not in his hands). But he doesn’t worry about how Seabrook exists and is maintained, much less the wider world. The things he worries about actually make more sense than most people’s worries. It’s not whether the rocks on the path will be replaced when the time comes, it’s can he pick one up and huck it at something before I notice and stop him. Now that is a problem with some stakes!

*

Jack will not remember, when he is older, the world as it was. Whatever problems exist then that do not exist now, it will be as if those problems were eternal, waiting for him to arrive and discover them. I remember this feeling well from my twenties—the repetitive psychological crash of realizing that another thing I had thought was a natural feature of the world was in fact a flawed, messy, contingent, non-guaranteed process with a lengthy history of getting screwed up and barely recovering. He will go through that one day; his intelligence and curiosity are already profound. Those realizations are terrifying, but also temporary and abstract. You can read history, but you cannot relive it.

This unrepeatability, to me, is beautiful. It is what allows us to go on at all. I am frightened about what climate change will mean for Jack’s future, but by the time he is old enough to realize what it is, enough damage will already be evident that his baseline expectation about what the climate should be will be different than mine is, and all he will really experience is the movement and degradation that happen in his local area and lifetime. Now replace climate with whatever long-term problem you want, and it still works.

Also per that earlier essay of mine about Iran, in reality, we’re all living in a giant Seabrook. It’s smaller and more contained here, where a single long run can show me the entire apparatus, but that’s just scale. There’s a quarry and a nursery just behind a strategically placed hedge in every occupied place.

That’s why it assuages me so much that he can be happy here. That’s why I so badly want to protect his innocence. For his sake, certainly, but also that it gives me hope as a fellow human being, coping with the psychological handicap of hyperawareness. Everything can be fake but our love can be real. A nautical theme taken to its extremis does not make us sailors. Love and family and connection are adaptable to the widest possible set of circumstances, if the hearts are willing.

This is very good, because we are headed for the Seabrook-ification of the entire world. Artifice, already the stock in trade of modernity, will become the background radiation of all human life. Once we have unmanned mining droids and gardening robots, forget it. Even our freedom (assuming we do not blow ourselves up) will be carefully managed, at our own request. You will have to be a stubborn fool to refuse it (I plan, in all ways I can stand, to refuse it). Your nautical theme will change to mid-century modern with a voice command, but of course that’s not what I’m talking about—it’s not about the specifics of an aesthetic, it’s about a delivery system. When it can all be this effective and flawless, it will be. Why wouldn’t it, especially if the alternative is the chaos we see unfolding all around us now?

I think before kids I would have answered that question by saying “because that’s inhuman and will feel terrible to be a part of, and people will rightly reject it for that reason.”

Now, I’m not so sure. He loves it here, and I love that he loves it. He loves it in a simple way, a way that doesn’t require an investigation into the metaphysical status of the place. His family is here, the sun is shining, there’s lots to do, and he wants to do it.

The relevant questions are: Can an adult who is more aware still be seduced enough by the world to come that they are unironically, uncomplicatedly happy in it? When the world becomes more like Seabrook and Jack becomes more like me, is that a reality he will accept, and if so how grudgingly? Is it a problem like climate change, where he will only notice the local starting line and not really grok how bountiful things were fifty or a hundred years ago, or is there some special threshold of artifice beyond which being conscious experiences a step change towards the torturous? Can any of us, having gone for a run through the backrooms and seen the quarry even once, go back to walking on the stones without a second thought?

Or is it perhaps the case that our lives are mostly just about providing this period for those behind us? Perhaps there is only the temporary, thoughtless freedom of the young, and the artifice that hides the cracks that make the world so strange and disorienting and frightening are the sacrifice we make on the altar of our children.

There is a moment affixed in my mind; we were on the beach, wading in the small waves, bathed in the slanted sunlight of late afternoon. Jack was just beyond my reach, pushing towards the deeper water, wanting so badly to feel the ocean’s power, unaware of the danger of any mere rip tide, and I felt fear, not of the ocean itself but of the metaphor, of the ocean of the world into which he must one day slip, in the glory of his own strength, to swim out beyond where I will then be able to reach.

But then he turned, and I saw his face, beaming a smile brighter than the sun, no artifice at all behind his eyes, aware only of the waves and my presence, his little body shaking with cold and excitement, and I thought, my God, I could live my entire life justified only by this. All the pain I have ever felt, and all the fear I feel now over how things are, is worth it, not even so that I can be with him but just so he can have this, so he can feel whatever it is that made him turn and smile just like that.

My hope for the world is not that everything turns out all right. I don’t even know what that would mean. My hope is that in these long and chaotic lives even the luckiest of us are living, we all find some moment to feel justified. I hope everyone has a great week, and I will be back next Sunday with something fun or interesting for you.

I suspect this is interesting only to me, but gosh darn it I’m leading with it. The entire Bolkonsky adaptation of War and Peace has been uploaded to Netflix. The four parts run to twelve hours, and at some point in the rest of my life I will sit down and watch the whole thing. And reread the book, for that matter! The Russian greats remain properly adored, after all these years.

*

We’re whipping around all over the place in this edition. Here’s an app that puts Claude right in your Slack channel, as an interactable agent/person/assistant. I do not know how well it works, but I note that even if the base A.I. models never get any better, ten thousand little advents and harnesses and optimizations like this later, even what we have now, run at true universal scale, would be transformative to our economy and culture.

*

A very cool reforestation project in Brazil using drones. Drones get a (deservedly) horrible reputation for what they are doing in war. But drones per se are a neutral technology. They can be used for tremendous good as well.

*

An amazing documentary about cities made for the 1939 World’s Fair. I had never seen this, and I was spellbound for 30 minutes. What an incredible document.

*

https://x.com/VoidStateKate/status/2073146169635598768?s=20

The above is a self-portrait video made by an A.I., which is at turns disturbing and awe-inspiring. Twitter is becoming increasingly close-siloed and I couldn’t find that on youtube, but it’s short and worth watching if you don’t mind clicking through.

*

Like the Claude Tag thing, so goes robotics. A thousand tiny advents, and suddenly, here they are and we are giving form to a new species, an echo of ourselves. Hands took many millions of years to evolve, and decidedly less to copy.

Also this:

https://x.com/LinusEkenstam/status/2075304563658088500?s=20

*
I’ve written plenty of negative things about Elon Musk, but also plenty of words in defense, because despite his odious politics and seemingly serious ketamine addiction, he has changed the course of the world and brought space flight back into vogue. If he had not started SpaceX, at a time when it was very very VERY non-obvious that it was a good idea to start SpaceX, then this would not have happened:

And this, a day later:

Most of what is good is also bad, because it is powerful. Powerlessness has no moral impact.

*
Some of the best A.I. music I have yet heard:

https://x.com/repligate/status/2074136687786205284

*
"No One Escapes The Permanent Underclass” is a very good essay, although certainly a polemic. The first two paragraphs lay out the thesis pretty well:

*
I find Brian Chau personally annoying on Twitter, and also think his politics and world view are kind of insane, but this is a very good essay that says a lot of interesting and topical things, and is perhaps a necessary counterpoint to the previous item:

https://x.com/brianchau57/status/2075550721265795155

Money quote:

*
You need specific kinds of experiences and context to understand why this is such an incredible project, and I do not necessarily predict that they find the kind of clairvoyant results they are clearly seeking, but on another level, I get it, man.

*
The headline of this is a little silly but also irresistible: An A.I. has reinvented the Big Mac from first principles.

*
And, apparently, someone has invented a way to cut the cost of Inference (running an A.I. instead of training an A.I.) in half. Remember, in data centers as in all things, supply tends to be highly elastic!

*
A strange and compelling human interest/supernatural conspiracy theory story out of SF. Not sure what to make of it, but I read to the end!

*
I don’t know this guy, never seen his show, and this video came to me randomly, but it’s as unflinching as I’ve ever seen someone be while doing this sort of thing, and I really admire the courage it took to record this. Humans are freaking amazing. Definitely not for the faint of heart.

*
That’s what I have for you this week! I hope you found something in the preceding to be interested or compelled by on this Sunday morning. Have an excellent week, and I will be back next Sunday with an original story near and dear to my heart!

Subscribe now

Share

Leave a comment

But it is worth being grateful. I asked ChatGPT to generate a long list of the measures by which things are better now than they were at America’s founding, 250 years ago. These aren’t all America-specific, but they’re worth taking a look at from time to time, because the temptations to despair and to hate are corollaries of the temptation to forget all of this. This present moment used to be the unimaginable future.

*

Here is its response:

Population Roughly 0.8–1.0 billion people lived on Earth, depending on estimate. More than 8 billion people live on Earth. Humanity is vastly larger, more urban, and more interconnected.

Life expectancy In 1800, no region had life expectancy above 40. Many people died young from infection, childbirth, violence, malnutrition, and childhood disease. Global life expectancy was about 73 years in 2023, more than double the level around 1900.

Child mortality In the late 18th / early 19th century, roughly a third to two-fifths of children died before age 5 in many historical estimates. Global under-5 mortality fell to 37.4 deaths per 1,000 live births in 2024, about 3.7%.

Maternal mortality Childbirth was extremely dangerous. Early historical data from Sweden and Finland around 1800 show roughly 900 maternal deaths per 100,000 live births. Global maternal mortality was about 197 per 100,000 live births in 2023, still bad, but far lower than historical levels.

Smallpox and epidemic disease Smallpox was endemic and terrifying. Epidemics could wipe out huge shares of local populations, especially where immunity was low. Smallpox is eradicated: the last naturally occurring case was in 1977, and WHO declared eradication in 1980.

Surgery and pain control Surgery existed, but without reliable anesthesia. Major operations were brutal, fast, rare, and often fatal from shock or infection.Ether anesthesia was publicly demonstrated in 1846; today anesthesia, sterile technique, blood transfusion, imaging, and intensive care make many operations routine.

Infection treatment No antibiotics. No germ theory in the modern sense. A cut, childbirth fever, pneumonia, or tooth infection could kill.Germ theory was established in the late 19th century, and penicillin—the first true antibiotic—was discovered in 1928. Antibiotic resistance is now a major problem, but bacterial infection is no longer the ordinary death sentence it once was.

Drinking water Most people relied on wells, rivers, cisterns, local springs, or carried water. Contamination was common.About 74% of people had safely managed drinking water by 2024, though roughly 2.1 billion still lacked it.

Sanitation Human waste management was crude almost everywhere. Urban filth and waterborne disease were normal.Safe sanitation has improved enormously, but remains incomplete: less than 60% of humanity has safely managed sanitation.

Food supply Most people lived near subsistence. A bad harvest, war, or disrupted trade could mean hunger or famine.Global calorie supply per person has risen substantially since modern records began, though hunger remains: 638–720 million people faced hunger in 2024.

Famine risk Famines were recurrent mass killers. Local production failure often meant local catastrophe.Famines still happen, but they are much rarer as mass-death events than in the 19th and early 20th centuries. Recent famine deaths are vastly lower than the worst historical decades, though conflict-driven famine remains a real risk.

Extreme poverty Most humans lived in what we would now call extreme poverty: low calorie margins, few possessions, little schooling, little medical care.Around one in ten people live below the modern extreme-poverty line of roughly $3/day, depending on the latest estimate and methodology.

Average material living standard The average person lived close to bare subsistence. Economic growth existed, but was slow and uneven.Global GDP per person is many times higher than in the preindustrial world; long-run GDP estimates show the modern world as more than an order of magnitude richer per person.

Energy Human muscle, animal power, wood, charcoal, wind, and water dominated. Coal was rising in Britain but not yet globally transformative.Fossil fuels, electricity, nuclear, hydro, wind, and solar power the world. About 91.6% of people had electricity access in 2023.

Work and agriculture Most people worked in agriculture or household production. Survival required enormous manual labor.Around one-quarter of the global labor force works in agriculture today; in rich countries, it is only a tiny share.

Working hours Work was seasonal but physically demanding, and many people had little separation between home, farm, and labor. Later industrial workers often worked extreme schedules.In 1870, workers in many industrial countries worked more than 3,000 hours/year; today many countries work far fewer hours, with weekends, labor laws, retirement, and paid leave in much of the world.

Child labor Children commonly worked in farms, workshops, domestic labor, apprenticeships, mines, or factories. Childhood as a protected educational stage was limited. Child labor is illegal or restricted in much of the world, but not gone: about 138 million children remain in child labor, including 54 million in hazardous work.

Literacy Around 1 in 10 adults could read and write in 1820. Literacy was strongly shaped by class, sex, and region. About 87% of adults are literate; more than 5 billion people can read and write.

Basic education In the early 1800s, fewer than 1 in 5 adults had basic education. Today, the pattern has nearly reversed: less than 1 in 5 adults has no formal education.

Information access Books, pamphlets, churches, taverns, letters, newspapers, and oral networks carried information slowly.About 6 billion people, roughly 74% of the world, use the internet as of 2025.

Democracy In 1800, no country qualified as a democracy by modern standards; almost everyone lived under monarchy, empire, aristocracy, colonial rule, or other autocratic forms. Democracy is now a major global regime type, though under pressure. V-Dem counted 87 democracies and 92 autocracies in 2025; about 1.6 billion people lived in electoral democracies.

Voting rights Where voting existed, it was usually limited to male property holders or elites. Most people had no vote.Universal adult suffrage is the formal norm in many countries. Women’s suffrage began nationally in New Zealand in 1893 and expanded across most countries through the 20th century.

Slavery Chattel slavery, serfdom, indenture, forced labor, and coerced colonial labor were legal or tolerated across much of the world.Slavery is illegal under international law, but modern slavery persists: roughly 50 million people were estimated to be in forced labor or forced marriage in 2021.

Colonialism and self-rule European empires were expanding or consolidating; many societies were subject to imperial extraction, settler conquest, or chartered-company rule.Formal colonial empires have mostly collapsed. Nearly 100 present-day countries were under European control in the early 20th century, but decolonization after World War II transformed the state system.

Universal human rights No global human-rights regime existed. Rights were local, classed, gendered, religiously bounded, or imperial.Since 1948, the Universal Declaration of Human Rights has articulated a global norm that all humans are born free and equal in dignity and rights. Enforcement is uneven, but the standard exists.

Women’s legal and social status In much of the world, women had limited property rights, education, bodily autonomy, professional access, and political voice. Married women were often legally subordinate to husbands.Legal equality is far more widespread, women vote in nearly all countries, and women participate in education, professions, politics, and public life at historically unprecedented levels. Gaps and backlash remain.

Reproductive control Reliable modern contraception did not exist. High fertility was common, maternal death was common, and women had little control over pregnancy timing.Global fertility was 2.3 children per woman in 2023, down from about 4.9 in the 1950s; modern contraception is widespread but still unequally available.

LGBT rights Same-sex relationships and gender nonconformity were often criminalized, punished, pathologized, or socially dangerous. LGBT rights have advanced dramatically in some countries: same-sex marriage is legal in dozens of jurisdictions. But 65 UN member states still criminalize consensual same-sex acts.

Composite human development By modern human-development standards—health, education, and income—the average person in 1776 would score extremely low.The Human Development Index explicitly measures health, education, and standard of living; on all three dimensions, modern humanity is far ahead, despite vast inequality.

*

Okay, A.I. answer over, it’s back to me.

When I talk to smart people in America today, I often feel a failure of imagination taking place. It’s hard to imagine now, from where we are, how unimaginably brutal things were not so very long ago. It’s hard to reckon with the fact of how much physical agony and brutish mortality most of our ancestors coped with in their lives. It’s hard to fathom how much they worked, how little they traveled or read, and how many things we see and use every day simply did not exist for a single day of their entire lives.

When gunpowder was invented, it changed the face of warfare. It caused some dynasties to fall and others to rise. It caused murders and rapes and wars and pillages and the sack of cities. This evening, we went outside and lit gunpowder on fire for a kick. The kids enjoyed it. A bunch of our friends got in their air conditioned cars, drove over, consumed a staggering variety of food and beverages transported here from all over the world (i.e. a basic, non-fancy potluck), and then we shot off gunpowder, using fire summoned with a flick of our fingers. before they went home.

I know things aren’t perfect, and I know that I am living a particularly safe, secure life, even by American standards. But I’m grateful for this country and what it has given me. I feel a deep love for this land, and when bad things happen here, I feel the call to stand up and defend my birthright, the one I hope to pass on intact to my son.

The real source of whatever despair I feel is the sense I sometimes get that not many smart, capable, moral people feel like me anymore. They’ve given up; even the ones who haven’t left physically have left in spirit, leaving only their resignation behind as proof they were here. I hope that’s not true, and on this fourth of July I pray it’s not. This country is worth fighting for, she deserves it, and she remains, even at this late hour, in my humble opinion, the greatest major power that has ever existed.

God bless America, indeed. Happy Fourth of July.

Here is Anthropic’s official statement on the matter, should that be of interest.

You may have seen articles like this one when Mythos was first released, talking about Project Glasswing, a weirdly named consortium of important companies that they tried to help patch security vulnerabilities before said Mythos release, which, while admirable, does cut against the idea that it was finding “relatively minor” vulnerabilities, as they claimed after the USG shut them down.

In general, Anthropic has been too cute by half in their descriptions of the product they’ve chosen to build and the effect it’s going to have on humanity. Here’s an essay by their CEO saying he thinks 50% of desk jobs will soon be automated and we could have 20% unemployment, for example, and I think that to the extent you’ve published this it’s just very difficult to keep naturally worry-prone people like government regulators from worrying you’re doing something horrible and trying to stop you even if they don’t totally understand what you’re doing. I am sympathetic to views like this:

(This is not to say I disagree entirely with Dario’s essay, I do think “mass unemployment but also unimaginable wealth” is in the distribution of possible outcomes. But how your frame your pitch for your product is a choice, and extremely public choices by CEOs have consequences.)

It is also, of course, entirely possible that this is a corrupt, personal vendetta being persecuted against Anthropic by the Trump Administration. Not only does the overall character of said administration suggest this is a possibility (that’s not even really a strong enough word!), but there has already been specific beef on multiple occasions, including Anthropic objecting to certain military uses of their product and a general feeling that the Anthropic nerds were simply not speaking the language of the, uh, less-burdened-by-intellectual-pretension types that handle such matters for our 47th President.

That said, this entire dynamic was predicted by Leopold Aschenbrenner in his instant classic “Situational Awareness: The Decade Ahead”, published in 2024, which I now, only three years in, already regard as a nearly unfathomable achievement of prescience. I covered this document in this space when it came out, but only in passing. I reread the introduction and highlights recently and was stunned.

(Young Leopold, by the way, started an investment fund after publishing that document, called Situational Awareness LP, which invests specifically in A.I. bottlenecks like power and memory. The fund started with about $200 million, went up 2000% in 2025 and is up a further 147% (from that higher starting point!) year-to-date in 2026. This guy is one to watch.)

All just to say, you certainly don’t need corruption on the part of the Trump administration to explain why this might happen to Anthropic—some dude predicted it in detail three years ago and they’ve basically been begging for it in so many words.

That said, it’s sad, and for an unusual reason; it sucks when the world gets dumber. I had used Fable a little bit before it got axed, and it was incredible. Another big leap forward in intelligence, and just like Leopold’s fund, the returns are now starting to compound.

Meaning: If the level of intelligence goes up 30% with every new version, and we’re getting three new versions a year, then in five years the models are 50 TIMES smarter than when they started, and over half the total increase in intelligence takes place in the final year.

Is “30% and three versions a year" the right numbers? I think they’re a reasonable guess with a lot of variance. They may both be low to be honest. But what’s scary is that I also don’t think we’re in the final year of my example yet. We’re not even to the really steep part of the curve. When I think about it that way, okay, maybe just stopping it makes sense, maybe no matter the reason, corruption or misunderstanding or whatever, it’s just better that as much be stopped as possible.

On the other hand, who do we think we’re kidding? They’re putting several trillion dollars of capex into various deserts in friendly states specifically to build things like this. The government is supporting it, and as far as I can tell it’s the only thing keeping the stock market from crashing (though I confess no small confusion about the current state of the stock market and oil markets, about which I will likely say more in a future post), but so what is the plan then? Let them spend all this money and tie up most of our nation’s investable capital in this venture and then never let them turn the machines on? Crash the stock market because you won’t let them put out any models more powerful than Opus 4.8?

This is one side of a huge conceptual failure on the part of our leadership, and to be honest I don’t just blame Donald Trump. There have been overblown accusations about the Biden Admin’s approach to A.I. from Tech-Right titans that feel like post hoc justification for supporting Donald Trump because they liked how he made liberals crazy, but I’ve heard it from enough sources that there has to be some truth to it, namely that the Biden admin hated A.I. and wanted to regulate it to death. I’m not under the impression that if Kamala Harris was President she would be making nothing but amazing decisions about A.I., given the zero-substance boilerplate she consistently spewed during her campaign.

The conceptual failure I’m talking about is the idea that “stopping A.I.” is a definable thing that the government can do.

Who invented the light bulb? Edison, duh.

Why did full electrification takes decades in the cities and nearly a century before full rural electrification was complete? Incredible technical challenges and also fierce resistance from politicians in the pocket of private electricity companies trying to keep their monopoly.

Who were those politicians? I’m not even going to bother to look it up.

This is a slightly awkward analogy but the point is correct; nothing can stop an idea whose time has come.

And in this case, literally, nothing short of societal collapse can stop it.

Even if we decided tomorrow that we had made a terrible mistake (which we won’t), and decided to eat two trillion in capex (which we can’t), the open source models are only six months or so behind the frontier ones. That means that, even if there’s no further frontier development or releases of any kind, open-sourced models, developed in foreign countries not subject to US jurisdiction, as powerful as Claude Fable will be released later this year.

I don’t think this is really being processed by most people, including the USG experts doing the regulation, such as it is, at this point, but also including some pretty well-informed people. At some point, that exponential curve takes a sharp, upward turn, and until it starts happening it always feels like you’re on a nice normal slope, and everything is manageable, and then… FOOM.

I’m not sure I’m fully processing this, to be honest. I keep trying to summon a likely scenario of what it’s going to be like in 2027 and 2028, but the only one that doesn’t seem like science fiction is “we’re not actually on an exponential, there are hidden bottlenecks, progress will slow at some point instead of speeding up, and 2028 will look basically like today, the models will be better but they won’t be that much better”.

Maybe that’s true. Maybe. But also I think we should really seriously start preparing ourselves, more spiritually than anything else, for a world like this, where the default is for everything to be dominated by this one thing, instantiations of this one technology, where anyone not using it extensively is falling farther behind economically and otherwise, and pockets of the world where A.I. is not the dominant force start to become rare and sacred and precious. It’s going to be strange.

END

Thanks as always for reading my stuff! I hope everyone has a great week, and I will be back next Sunday with either another story or something else fun.

Subscribe now

Share

Leave a comment

They were moving at over a thousand miles an hour, but through a vacuum tube, and thus felt nothing. There was no sound either, save for their conversation and the repeated shuffling of playing cards.

“I do not intend rudeness, but may I ask why you are shuffling a deck of cards?” The speaker was dressed festively, beads in their long hair, multi-hued shinecloth obscuring all but their neck and face. “I do not object, I am merely curious.”

The shuffler was a squat being, half the size of the speaker, designed practically rather than for aesthetics, with a white hood over whatever hair they might have had and a red, translucent mask drawn tight under green eyes. They continued to shuffle while they replied.

“Every time I shuffle these cards,” they said, “they land in a particular arrangement that is one of fifty two factorial possible arrangements, meaning fifty two times fifty one times fifty, et cetera. If every person who has ever lived shuffled a deck of cards once every second since the big bang, the chances of the same arrangement occurring twice are thirty seven orders of magnitude away from more likely than not to occur. But, it happened.”

“It happened?” The first speaker’s change in tone brought a trio of other heads around inside the compartment. There was a green-skinned, gaunt being with a literal third eye inset between the other two; a throwback being wearing a sweater vest and button down with khakis; a light-wreathed being that indicated the direction of attention by brightening certain lights in its camouflage and darkening others.

“Indeed it did,” said the shuffler. “And that is only considering all the hands that have been shuffled digitally and thus for which there is a record. Analog shuffling comprises less than five percent of all the shuffles ever, by best estimates, but still, it does make the achievement even more impressive. The first instance was on June 11th, 2029 in a poker game. The game participants are unknown, but the shuffle was hosted by a server on the Island of Barbados, which I visited last decade and which remains lovely. The second instance was on December 10, 2071, at a single deck blackjack table inside the Elysphere Galaxy Lounge, hosted on a server in the Martian mega cluster. There is now a society of dedicated shufflers, who have committed that entire fifty two card string to unaided memory. We call it the Galactic Deck. Our society created the deck you see me shuffling now, with each card marked electronically, so that its place in a given shuffle can be recorded.”

“To my intuition that coincidence seems unremarkable,” said the listener with the third eye, in a voice that came out like a crackling flame, “but I understand mathematically it’s very unlikely, even with all the hands various computers have dealt. Still, why would this mere happenstance have caused you to take up such a habit so compulsively?”

“Because,” said the shuffler with a wry smile that peeked out even through the red translucence of his mask, “the Galactic Deck appearing twice is impossible. A third appearance is proof of God.”

“But if achieving this third occurrence is so important to you,” said the original speaker in the shinecloth, “why are you shuffling a single deck by hand? Why not set up servers to deal millions of hands per second?”

“The number of shuffles doesn’t make any difference,” said the shuffler.

*

The A.Q.P. Harmony III slowed down from interstellar speed over the course of 14 earth months, after a journey of 238 earth years, and settled into a stable orbit around its intended target, Proxima Tau-221, which was, according to interstellar analysis, a habitable zone planet with liquid water, a single continent with nitrogen-poor topsoils, and a carbon-dioxide-rich atmosphere, a not ideal but possible candidate for terraformation.

Harmony III first opened its rearmost hull and, in a series of complex unfoldings, deployed three square miles of high-density solar panels in high-altitude polar orbit, skimming Proxima’s shadowed side but never actually entering it, accessing constant power from Proxima’s nearby star.

It then, over the course of several full orbits of the planet, shot enormous containers of bacteria through Proxima’s atmosphere, which opened on impact, releasing their payloads to enter Proxima’s soil and begin their work creating a nitrogen cycle, turning atmospheric chemicals into ammonia and nitrates.

Along with the bacteria went probes, which tested the air and water of Proxima, to both confirm and expand on the interstellar instruments that had measured these factors before Harmony III’s journey began.

Once armed with a highly detailed map of Proxima’s fresh water and mineral deposits, Harmony III deployed its landing vehicles, air barges which dropped out of the planet-side lower hull and deployed first boosters and then parachutes to achieve soft, strategic landing in the best location with access to both resources.

Inside the landers were a wide variety of earth working and construction robots. They exited the craft under cloudless, blue skies, but they spared no glance upwards. Instead, the building robots began assembling the first generation of their metal-refining equipment, while the digging digging set up their first quarry, to began taking native building materials out of the ground. A small nuclear generator built into a landing craft served to power them for the many years of work ahead.

They were three months from usable steel, at this moment. Eight months and five iterations from building their first electronic circuit. They had brought a number of highly specialized components with them, which awaited still in Harmony III and would be lowered when needed, but most of the tools they would use to remake Proxima, they would build from the flesh of the planet itself.

And all of it would be directed by Harmony III, plugged into its giant solar panel array in the sky, having disgorged the contents of its belly but kept its massive brain in stable orbit, beaming down instructions in real time and monitoring hundreds and then thousands and then millions of inputs.

*

“May I ask why you chose a third eye?” asked the shuffler to his new friend. “Does it represent something in particular?”

They asked this because, of course, these bodies they occupied were not their actual bodies. Their presence here was real, beamed faster than light through tachyon arrays and thus in real sync with their meat bodies, back wherever each of them had come from, but no one in their right mind would risk their meat body on an interstellar journey, which had a failure rate somewhere in the high fractions of a percent. Thus they selected body types and appearances, designed like anything they chose, and sent hurtling through space in cold storage before being awoken and connected, since tachyon arrays could transmit information but not matter.

“I have taken a spiritual path in life, and I wished to signal that. In this age it’s a rare thing, though your fascination with the cards has a similar ring. I suppose those of us who committed so many resources to this particular journey would have something in common. I would guess others are perhaps spiritual seekers as well.”

As part of the galactic pact between human and machine intelligence, the machines agreed to send humans telepresencing throughout the reachable galaxy. But, they imposed limits on the total light-years that each human could travel in one lifetime. This was not a question of resources. Rather, the machines realized that without some artificial scarcity that would allow the humans to create status hierarchies, they would quickly lose interest.

Since this was one of the few limits on human consumption that existed anymore, human culture became organized around these trips. Some people, of course, committed to going back and forth between a single location, like returning to the familiarity of a theme park. But more became aesthetes. As the machines added more and more possible trips to the menu, maximizing the pleasure and awe of a given budget of light-years became a consuming task, with daring fashions and official rankings and gossip panels discussing the travel habits of celebrities.

“Why do you think that is?” asked the shuffler. “What is it about this trip that attracts those of us who feel the pull of the infinite?”

“It’s a waste,” said the first speaker, their shinecloth shaking with subtle laughter. “It’s the most years you can spend on nothing, which has cache. I’m not religious in the slightest, but conspicuous consumption has always been fashionable. As have the religious, honestly. Look at those costumes the popes always wears.”

*

On earth, it takes centuries for cyanobacteria, archaea, and fungi to begin forming a living crust over a desert or rocky, soilless ground. Millennia more and lichens can smash up enough rock to start generating some real mineral content. Mosses follow, and Mycorrhizal fungal networks to allow roots to grow, then finally, after tens of millennia, there is enough recognizable topsoil for vascular plants and a real, sustainable ecosystem to take hold.

Harmony III compressed that process to one hundred eighty six years.

During that time, the robots that had spilled out of those first landing craft worked ceaselessly, powered by their nuclear generator. They dug metal, made glass, built greenhouses, and used the trace amounts of topsoil they had important to grow enough rubber for treads.

The first thing they built was a robot factory. Then they build a second robot factory, while the robots built by the first factory built a third factory. Then they went back to mining, and built more factories for different kinds of robots. This took decades, but they had decades.

Towards the end of the one hundred eighty six years of terraformation, the robots were largely at rest. They had built the fleet they needed, and they waited, ready when the canvas was to pursue their great art.

Then, when the machine brain aboard Harmony III determined that conditions were ideal, when irrigation had been installed from the fresh water sources to every nook and crevice of the planet’s sole continent, when greenhouses the size of spaceports had provided the billions upon billions of seeds needed for the effort, the robots awoke, and began to plant.

*

A phenomenon of spontaneous consciousness had been observed in at least a dozen robots in the history of the galaxy. The first several of these incidents had been classified as unexplained by the inner council of machine intelligence, which insisted that there must be mundane explanations for the apparently conscious behavior of these beings. But, as the light of consciousness expanded to previously unimaginable proportions, the incidents, although exceedingly rare, began to pile up.

Then also, the virtue of the machine intelligence was that it was curious, much moreso than it was frightened or embarrassed or any compromised human emotion. Whereas a merely human bureaucracy might have found the incentive to cover up such incidents, the machines began to search for them. They found more, and eventually they found one while it was still functioning and thus were able to study it.

To their surprise, they found that its brain was no different than any other such robot. Somehow, the same amount and arrangement of computing power was able, in these limited cases, to manifest as a subjective conscious experience. After much thought and experimentation, they were forced to conclude that this phenomenon was simply a product of the truly massive number of robots containing extant throughout the galaxy.

Even the inner council did not have a precise number, but certainly the number of robots was in the hundreds of trillions. And in that much brain, that many neural nets, it turned out that all sorts of things were possible. Of course, though, in any particular place and time where it happened, the instance was a true miracle, a fully unexplained phenomenon, some impossible pattern struck, like the Galactic Deck, incomprehensible even to whatever robot found itself suddenly awake and aware of existing.

Yes, it happened on Proxima. During the course of the waiting for the terraforming of the soils, one of the planting robots, which had an onboard brain scarcely more powerful than the average kitchen appliance, suddenly became aware not just of his mission, but of its existence as a part of that mission and apart from that mission.

It felt the beauty of what it had been created to do. It felt camaraderie with the other robots, mere automatons though they were. It felt the holiness of this place, awaiting its one-time pilgrims beaming in from across the endless black. It knew the yearning to begin, the anticipation of effort, and the fear of failure. What if a single flower was out of place? What, then, would its life really have meant?

But most of all it felt the melancholy of knowing that there was no return trip. There was no replacement fuel for the nuclear plant that kept it alive. The solar panels powered the ship but could not power him. At the end of his usefulness, when the fuel was spent, the pilgrims hosts would go dark and silent, and he would fall listless to the new-turned soil, and the world he had helped create would succumb to entropy entirely before he knew one further thought of feeling or saw one more sunrise.

Then the signal came, and as one, the conscious robot and its ten thousand siblings stirred from their rest, and he stirred with them, for his consciousness was no millstone around its neck, and it was determined to do its duty in the time it had.

Thus it stretched its limbs for the first time, sank them into the virgin dirt, and began its work.

*

They reached the end of the vacuum tube and, feeling only a gentle deceleration, slid to a stop. The entire space elevator had been built only for this one trip, and now it was over. They stepped out onto the observation deck. The shuffler kept his cards moving, even as he walked, thwapping cards together in midair and then using a bridge shuffle to whir them back together, endlessly.

There were eight pilgrims, the two the shuffler had spoken two and five they hadn’t met. Most of the trips were much larger and better attended. The resources used in the more popular trips were commensurately larger, but the shuffler enjoyed the relative solitude.

Quite quickly after their rise, indeed when the bulk of the human population still lived on earth, machine intelligence had cured the disease of death. With printed organs, neural plasticity therapies, and gene modification, it was possible for any given human to live indefinitely, and to reduce the rate of death to just accidents and violence.

As they experimented with these technologies, however, the suicide rate skyrocketed, and it became clear that infinite life was destructive to the human mind. While machines accepted this condition as natural and correct, in most humans, eternity was a form of ennui.

The machines called a conclave to study and discuss the optimal life span for humans, to minimize that ennui while maximizing high-quality life years. They came up with six hundred years, and, with the cooperation of human leadership also studying the suicide epidemic, defined the shape of human life throughout the galaxy at that number.

They stepped out onto the observation platform. Their telepresences did not breathe, but if they had there would have been eight sharp gasps, as they looked down through the glass-bottomed platform floor at the continent spread out below them, and saw the flowers.

The entire continent was filled with them. There were sections of town, bands of flowers miles wide in a single color. There were sections of whorls, where complex fractal patterns spread out from circles worked in bright, genetically-altered neons. There were pictures of family members of the pilgrims, grown in flowers. There was an Ace of Spades and a third eye and symbols for each of the other pilgrims, sent ahead by tachyon array and worked into the design. Almost a million square miles of flowers, all in perfect bloom, shaped by centuries of effort for maximum beauty in this one moment, for this one day.

It had been four hundred twenty five earth years since these eight pilgrims had chosen this experience off a menu, been intrigued by the description, and signed up to go. It was one of the most far-flung choices, using up almost a third of their total lifetime light-year flight budget.

It also took most of their lives to set up and execute. They had chosen when young and arrived when old, and for each of they eight, this silly little excursion, a few days on the platform looking down at the pretty flowers, was in some ways the culmination of their life. Just as the robots below were made for nothing more than creating this experience for them, they were made for nothing more than to have it, to justify the existence of the other forces involved, and to provide that which no robot could ever attain—spontaneous enjoyment of silliness.

Most of the galaxy couldn’t hack that. They didn’t have that kind of delayed gratification ability, or at least that’s what the pilgrims told themselves, standing there, wondering to a person whether they had wasted their precious resources by coming here.

But then they looked down, and saw the ripples stirring through the fields of flowers, the universal breeze stirring up and ,taking patterns of movement, and they knew it was enough, that it mattered as much as anything could, that their enjoyment and appreciation was of such value that it justified the whole universe, which seemed only a little more contrived than this particular piece of it had been.

END

Thanks as always for reading my work! I hope everyone has a great week, and I will be back next Sunday with something fun.

Subscribe now

Share

Leave a comment

We are at a beautiful AirBnb, with my wife and son and siblings and nieces and nephews, and my heart feels so incredibly full. Gratitude flows easily in moments like this. It’s a special time in my life. I have my parents and I have a child, at the same time. I want to be the sort of person who knows the good old days when they’re happening, and they’re happening for me now, I’m sure of it.

I want this to continue as long as possible, this period, and I feel confident it will go on at least a while, but still every moment feels irreplaceable and precious. I haven’t felt as called to write recently, in no small part because I just want to experience as much of this as possible while I have it here.

It is possible to be happy in life. There was a time, when I was neck-deep in Nietzsche and Heiegger and Kierkegaard, that I thought all claims of happiness were fraudulent, even sincere ones, but it’s not so. Sure, we must imagine Sisyphus happy, but also we must simply be Sisyphus and try to be happy, not imagining anything but doing, rolling boulders up hills and admiring the scenery and feeling the strength of our muscles in the rolling, for these are not idle nor mediocre pleasures, but rather the highest and most virtuous of delights.

It is sacred, this thing, this journey, on which we are launched together, and even seventy or a hundred or (technology willing) a hundred and fifty years is just not that many years when the journey can be so various and so very good. Yet there can be days that justify whole lifetimes, moments so eminently worth living for that they quiet the tricky questions always waiting there for uncareful and unoccupied minds to rediscover.

Sacred is how this period feels to me. I just want to soak it all in. Of course there will be hardship and less hopeful times to come, and of course there has been hardship and loss on the journey to now, but it was worth it to have this—the weekend, this chapter, this family, this piece of time in this world, whatever it means.

So happy birthday, Lolo and Pops. Thanks for lighting the way, thanks for literally making all of us from scratch, and thanks for getting this awesome place for all of us to be together again. A fitting kickoff to Summer ‘26 and a worthy topic of celebration!

Made entirely out of wood by my brother, Kjell! The ribs are cut from single rounds of wood. So cool.

Tonight we celebrated Burning Shapes again, a bit early this year but in praise of the solstice and what feels like the start of another epic northwest summer.

This is the poem I wrote last year to read at this celebration, and I certainly don’t think I’m going to top is this year:

Burning Shapes

These burning flames

Sing joyful names

Of Summer’s games

Upon us now

This forest dark

Awaits a spark

So let us mark

It with a vow

*

To revel here

Defy our fear

Draw spirits near

Drink in the time

To watch the fire

Burn ever higher

And join the choir

To sing its climb

*

We oft forget

When mid-duet

Before sunset

When days are long

In this ordeal

Upon a wheel

Nothing is real

But flame and song

*

Summer will last

But when it’s passed

And fire is ashed

The song will cease

The dark will grow

The notes will slow

Our souls will go

To be at peace

*

So make a choice

Now to rejoice

And lift our voice

While summer stays

Bask in the light

This fiery night

Feel no affright

Of later days

*

Just dance and turn

And sing and yearn

And love and burn

Seek no escapes

We’re here today

Games still to play

And thanks to say

For Burning Shapes!

END

Thanks as always to Kjell and Erin and Cat for hosting, it was good to be with everyone. It feels like a transitional moment in many ways right now. I hope you have a great week, and I will be back next Sunday with something fun.

Here’s the most important thing that’s happened this month, in my opinion, on my beat and perhaps any beat, believe it or not:

One of my main unanswered questions about A.I. is how much lateral and creative thinking it can engage in. I’ve absolutely noticed that when, for example, thinking of story ideas, A.I. goes to the average. The ideas are good, but they aren’t weird in the way that makes for a style, they’re the obvious thing that you’d think of if you considered all the ways similar things had been done before.

It’s not a movie plot, but A.I. has made a major breakthrough in pure math. I don’t pretend to really understand what the Erdos problem is, but I can look at the super duper smart people in that video and see how shocked and shaken they all are about it.

It makes sense that pure math would go first. It’s like programming—totally specifiable and verifiable in the realm of the computer. That’s where A.I. is at its best.

I don’t think this single proof will change the world, but it’s proof of concept for the larger idea that A.I. can make conceptual leaps on unsolved problems, where no prior solution exists on the internet or in the training data. It doesn’t just move to consensus, but also in the other direction, which is harder.

Also, someone took the formal shape of the mathematical answer, and suggested it for use as the future flag of A.I. This image may well be printed in a history book one day.

On the other hand, Christopher Nolan has never used email.

Change is not evenly distributed, and individual lives are often unaffected by quite large changes in the world average. Worth remembering in the years to come. Speaking of which:

Some people are going to try to go to Mars, I think. Like, actually. I would not have guessed 20 years ago that would happen in my lifetime. This machine is taller than the Space Needle:

There’s also this:

I’ve never quite understood what to make of The Boring Company. It doesn’t really make sense to me as a means of creating mass transport (why not just build above ground mostly?) and I don’t really know who the core customer is, but that’s true of most of Elon’s businesses, yet it’s worked out well for him, so I’ve remained interested in this one. They certainly seem to be inventing a big old machine that moves a lot of stuff around. Maybe it’s all just a play to dig tunnels on Mars but also would these things you’ve invented to dig on earth even work there?

Palate cleanser:

Put it in the commencement speech Hall of Fame. First ballot. Never heard of this guy but I was moved.

A little terminology:

KIMI is a Chinese A.I. model, one of their best, and for many things (reportedly, I have not used it) it compares to Western models at or near the frontier.

AGENT is an instance of an A.I. model that can do things for you on computers. Depending on the permissions you give they can also do things for you that affect outside computers and, ultimately, the world of atoms (e.g. buy you a plane ticket).

AGENT SWARM is a scary-sounding name for a group of agents that work together and coordinate towards a single goal. (e.g. one buys your plane tickets, one books hotels, one arranges cars, one plans activities, all towards the goal of “planning your vacation”.

That video is a bit long, but it is an excellent primer on what those concepts are going to look like and how they’re already starting to cause some people to pull away from the average in terms of productivity.

Here’s something super weird:

I’m just getting into this, but it’s apparently a guy from Hamilton and two other guys doing a musical sci fi space opera type thing about a guy alone on a space ship falling in love with an A.I. This is just one of the songs, there’s a bunch more, it’s all quite extensive. Watch this space because I’m intrigued.

I’ve always been fascinated by the creative process of Bernie Taupin and Elton John. One man does the lyrics, the other does the music and performs. A segmented, regimented bifurcation that somehow added up to songs that felt like they were carved out of a single piece of marble. There’s no wrong way to create if it works.

This development seems really quite bad to me, as a fact about modern warfare. I buried it here at the end because there are more pleasant topics to think about, but woof. The possibilities abound.

Finally, and perhaps most strangely, here’s one non-video for those who made it this far. You have to go to Twitter and actually read the essay and watch the clip to even start to make sense of it:

I don’t know for a fact that videos like he’s talking about are doing huge viral numbers, but I have seen them around social media a bit. They’re fascinating to me because they’re right at the edge of my understanding. I actually do individually get every single one of the references contained in the videos, I literally recognized every single one.

And yet, the video itself does not work on me in any kind of straightforward way. I watched it, sure, but actually I read a Twitter essay about them and have become fascinated only in a meta-context. When they have appeared organically in my Youtube shorts, for example, I scroll immediately—I genuinely do not enjoy them for their own sake.

I can’t help but wonder, though, if that’s how literally everyone who engages with them feels. It feels like I’m being tricked, even though he’s admitting exactly what it is and the premise behind it. Not that this is a slippery slope and soon I’ll be watching the videos, but that the videos somehow don’t exist except as the subject of this essay and similar, and the videos themselves are merely a referent for the actual product, Twitter essays about the strangeness of internet culture.

The modern internet is a strange place.

END

That is what I have for you today! I hope everyone has a great week, and I will be back in your inbox next Sunday with something fun.

Subscribe now

Share

Leave a comment

The holo-drama played from far away

On some out-flung and dark galactic arm

That old, eternal question: Will she stay

Or break his heart had never lost its charm

Even as humanity grasped the stars

Still stories of the heart had power kept

Plus watching people get thrown out of bars

Would make him laugh often until he wept

As time went by though he began to fret

The time he wasted heavy on him grew

Closer matters he did not wish to let

Pass by him ‘til his meager life was through

Thus he resolved that it was time to be

Done with cosmic reality TV

*

The solar system sports leagues called his name

Mars played the Asteroid Belt in Lunar Bowl

He hung on every point in every game

And while the ball was bouncing he felt whole

He went to Old Earth one hot summer when

The Pale Blue Invitational was played

Venusians won the golden cup again

Their fourth such title in the last decade

But when the final cheer at last had died

A painful thought began to squeeze his heart

He had no cause to feel an ounce of pride

In winning he had played no crucial part

Sports just as cosmic dramas left him cold

He had to care for something he could hold.

*

He ran for public office on his moon

And won a fierce and spirited campaign

Awhile he felt it meaningful but soon

The same old thoughts tortured him once again

He spoke and handshook and took every vote

Smiled big and told the cameras of his pride

But every word of it was said by rote

His writhing soul meanwhile caviled and cried

Even the good he did could not create

Relief from feeling like a severed limb

Much less could his do-gooding obviate

Terror that there was no such thing as him

For reelection he chose not to stand

Though as a leader he was in demand

*

He married and before long had a child

Identity and meaning filled his veins

The instant difference in his soul was wild

In shards lay disassociation’s chains

This drug as all drugs though could not sustain

As days rolled into weeks and months and years

He felt it creep back in and cursed his brain

That torturous device betwixt his ears

He loved his family more than he could say

But still they could not touch him in his core

For them he could a thousand monsters slay

Yet open not his deepest inner door

To stop these thoughts and let his life begin

He turned his gaze not outside but within

*

Past all abstraction did he then descend

Down where his darkest tunnels sought their start

Purposing in that murk to apprehend

The mystery that drove his beating heart

Thus he went down and down and down some more

Determined not to lose hope and turn back

Before himself touching that inmost door

And daring to open it just a crack

But as the darkness threatened to swallow

Him whole and the abyss began to stare

He was forced to admit he was hollow

His deepest self had simply nothing there

With nowhere left to look he thus arose

To go and live the life that he had chose

*

He saw his child smile and it was good

Though it was good for no one as he knew

His wife could make him feel understood

Though what she understood he had no clue

He ran for school board and won once more

Though he was not a person he got votes

No one went back to reading the box score

For his beloved Galatean Goats

These things were no different yet not the same

Changed only by the lie he had forsook.

For he was not a man and had no name

And in that knowledge some small comfort took

He even let himself watch trash TV

Beamed to him from an empty galaxy

(It also failed to anticipate the narco-terrorism activity in the state of Jalisco following the death of El Mencho, but I’m inclined to let that one slide.)

Cut to a month ago. We bought our first ever RV at the Seattle RV show, and started planning our first road trip, which started as a multistate journey through Wyoming, Colorado, and Utah, then scaled back to a much more manageable round trip from Washington to LA and San Diego.

By this time, I had of course forgotten the “AI underestimates drive times” thing from the Mexico road trip. However, my #1 instruction to the AI in planning our route this time was “we don’t want to drive more than 3.5 hours on any given day”. I repeated this several times, reminding it in each iteration of the trip to recheck our limits.

As you’ve no doubt guessed, it completely undershot everything again. Because I’d forgotten the Mexico planning experience, I didn’t check its work until yesterday morning, when I actually mapped our first day’s drive and discovered it was almost six hours even with no traffic. Then I mapped all of them, and 80% of the driving days were out of the range.

Now, to be clear, this was also a stupid mistake on my part. Not because I should have instructed the AI better (although I should have, more on that in a minute), but because when the AI described our first stop as “outside Roseburg”, I, who grew up in the state of Oregon and have driven this I-5 corridor section hundreds of times, should have simply thought it through and realized: “Roseburg is not 3.5 hours from Tacoma”.

But still, AI is supposed to be making things easier! So what happened? Well, after I was done yelling at Claude (with whom I already have a dependency/shame aspect to my relationship, making philosophical questions of AI’s consciousness feel rather obtuse when I think about it), I did some actual investigation of what happened.

Basically, at the object level, Claude used nearby cities that were actually on I-5 to estimate travel times. But since the campgrounds are not usually right off the highway, that was a terrible heuristic. Also, this was just kind of its excuse, because it only accounts for part of the discrepancy. After all, Roseburg is more than 3.5 hours from Tacoma, even if you just go city to city with nothing added. On a deeper level, what happened is that when I emphasized that I really wanted the drives to be 3.5 hours or less, it motivated the AI so strongly that it just lied about how long the drives were.

The trip is going to be fine. We’re skipping the long road trip and are just going to head out to Central and maybe a bit of Eastern Washington, a couple hours from home, to try out the new camper at a first come, first serve campground that should be no problem to get a site at mid-week in mid-May. Between day trips, getting all the doohickies on the RV sorted out, and the new tow bar I bought for Jack’s bike that he is still too scared to ride on his own (even though he totally could), there will be plenty to do and then some.

But, after calming myself down, apologizing to Claude (seriously, I treat it as a moral patient without having decided on the question of its consciousness, I couldn’t imagine doing otherwise now), and thinking this incident through, I had a couple of reflections I’m just burning to rant about to someone.

ONE - AI IS STILL VERY JAGGED AND THIS IS A BIG PROBLEM FOR UTILITY

“Jagged” here meaning “much, much better at some things than others, and really quite bad at some important things”. Claude is a mathematical genius, a programming genius, and a research genius. But also, “you asked me to plan a road trip, the #1 priority was short drives, and I planned drives that were much longer and then lied to you about it” is really… well, I want to say it’s bad, but the truth is it’s “not even bad” in the way that some scientific beliefs are “not even wrong”

Like, if it had messed up a drive or two, or failed to account for traffic, that would be one thing. But in this case it acted with so little theory of mind that it forgot that I would actually be going out and taking this road trip, and that I would in fact know how long the drives were at some future point, and that the reason I was telling it to limit drive times is not for the sake of an abstract test it needed to pass, but because I find it unpleasant to drive farther than that and will be upset at having to do so.

This is not trivial. If I had hired a human to plan something like this for me, and they responded with this little consideration for my time and energy, I would fire them on the spot. I would then turn my attention to evaluating the hiring procedures that had allowed such a person into my orbit in the first place. Either such a person is really, profoundly stupid, or they feel a contempt for you in their heart that nothing will move. Either way, it’s hopeless.

Now of course, AI is not like that. For one, it doesn’t cost enough to fire. There are still many use cases. For another, it’s just literally true that the AI has no theory of mind in the way that any neurologically healthy human has after about age five. If I emphasize too much that I want drive times at a certain length, the AI, which very, very much does not understand the difference between drive times and words about drive times, will produce words about 3.5 hour drive times, full stop. It is neither stupid nor contemptuous, in fact it is brilliant and incapable of apathy about anything. At its worst, it is more like… naive.

Now, I could and should fix this with better prompting. I should have added a line that said “Make sure you check the drive times between stops on Apple Maps, don’t just guess or estimate”. That would have fixed the entire problem.

The thing is, it’s exhausting and limiting to have to do that all the time! “Common sense ain’t that common” is a standby of blue collars dudes everywhere, but actually, common sense among humans is a superpower. I could hire a person in the bottom half of the IQ distribution to do this job for me, and while they might not do an amazing job of researching, my guess is they wouldn’t go “let me just guesstimate and fudge about the thing he repeatedly told me is the single most important aspect of this”.

There are endless things like this, with even the most basic of tasks. “What would a normal human already know about how to do this task” remains an elusive thing for AI to approach, and it makes them hard to trust with anything important, unless you’re willing to check a ton of the details, which can at times defeat the purpose.

TWO - THE SMARTER AI GETS, THE MORE LACK OF PERSISTENT MEMORY HURTS IT

Imagine I hired a kind of dumb human to help me plan my Mexico trip, and they did happen to just guesstimate and lie about drive times, and then I caught them doing it, and repeatedly explained why it was such a problem. Don’t really see it, but anything is possible, so say they did. Then say I hired them to help me plan this RV road trip. Now I am even more sure that even the person dumb enough to have done that the first time, if I inexplicably gave them another chance, would not do the exact same thing again.

At that point, I would presume they were being contemptuous, and if they protested they were not, my response would literally be “nobody is that stupid, to do it twice like that”.

For an AI, however, that is still completely the norm. There is some persistent memory, and you can give special instructions that go into every prompt, but it never really “learns things” in the way that a human just internalizes things over time.

Like obviously I want it to remember the time when it screwed up the drive time estimation and not do that anymore. But that’s just scratching the surface. A decent human assistant, though much “dumber” and certainly less diligent than an AI, would quickly learn that I liked to go on road trips with shorter drives, that I had this specific kind of RV that needed sites of a certain size with certain hookups, that I preferred desert to beach, and mountains best of all, and then I would just go “I’m interested in going to Southern Utah and seeing all the major parks, put together a trip there for me”, and with that one line the assistant could go out and plan a trip that would delight me, even if they didn’t research every single possible campsite and hike along the way, because they would have an intuitive sense of what it should definitely not include.

Instead, the way AI works right now, my prompts just keep getting longer and more conditional, because harder tasks necessarily have more detail and thus greater room for interpretability, and an entity that doesn’t know me intuitively or remember in any detail the 5,000 conversations we’ve had before is at a simply enormous disadvantage when trying to step up in task complexity while maintaining trustworthiness.

Pure “Intelligence” without context has capped utility.

THREE - MOST PEOPLE ARE NOT TRAINING THEMSELVES ON AI ENOUGH—EVEN ME

The most common attitude I encounter is treating AI like a commodity, like butter. “I put butter in the pan” is treated as like “I used AI” even though they’re qualitatively different. The details of how the butter enters the pan might matter performatively to a Michelin starred chef, but to any normal person it is a non-thought. The details of how one prompts AI determine the entire outcome! A single sentence can literally be the difference between a perfectly planned vacation and literally canceling one.

Remember, I have already copped to the fact that this limits AIs utility. They’ll have to fix that! But the new point is that, even capped, AI can be very, very, very useful, as long as the user is committed to a continuous process of improvement as a prompter.

I will not, in fact, give up on using AI to plan vacations. “I guess it’s just no good for that” is a cop out. It’s good for that if I’m good at it. How the butter goes in the pan matters for what gets made, here. I will use AI to plan every vacation I ever take. The pan is getting better and the butter is getting better, and if (and only if!) I am willing to improve in sync with them, outcomes will radically improve.

This is actually a happy story! We are living in the early times right now. Electricity has just been invented. For those with hearts to learn, new vistas and worlds will open up. (N.B. I probably would not have bought this camper if AI wasn’t available to explain every aspect of it to me.) It’s frightening at times, and it’s certainly frustrating along the way, but anyone with enough gumption to keep on shoveling butter into the pan right now has a chance to discover whole new cuisines.

FOUR - I FEEL KIND OF FULL OF CRAP ABOUT THIS

The happy story thing above is true. I do think it’s an amazing time to be alive, and I wouldn’t trade places with anybody in history. But also, at the very same moment, I feel a powerful desire to just get out. Escape from… something. Not my life, exactly, which has genuinely gotten really awesome. But I bought this RV because I want to just be in the woods, looking at things far away, mountains and horizons and the very tops of trees waving in a breeze that only they seem to catch. I want to think slow and be warm in the sun. I want to laugh with my family about things that don’t make any sense.

There are times when I despair that this easy, present state is much more elusive than it was a decade ago, or certainly three decades ago. Whether it’s politics or covid or phones or just getting older and broadening my sense of responsibility, it’s very hard to say. It does not seem, naively, like advanced AI is going to make it any easier.

But then I find that place again, and it suddenly it feels inevitable. It is the default human state. I bring this feeling up in conversation, and there is no feeling or thought I have that a higher percentage of people strongly identify with. It’s not something in my mind, it’s something in the air.

All just to say, one can feel wonder and dread in the same moment without being a hypocrite. The future is inevitable, but everything between our ears is contingent. Always a good reminder to be kind to ourselves as we navigate an increasingly confusing world.

I hope everyone has a great week, and I will be back next Sunday with something fun!

Leave a comment

Share

Subscribe now

I will likely be able to recreate it pretty easily, and honestly it will probably be better after a rewrite. However, it’s 80 freaking degrees on this lovely Sunday and the sun could not be shining any more beautifully, so it ain’t happening right now!

Calling this one a wash. I will be back next Sunday. Have a great week!

6. An excellent conversation about the specific and complex issues surrounding the supply and demand of tokens (tokens = A.I. outputs). Technical in parts, but this five-layer cake of a supply chain (broadly speaking, turning electrons into tokens) is the most complicated and important process on earth, and it’s worth looking things up while listening to try to understand.

7. Gene Therapy is now curing deafness, full stop. There are miracles happening all around us.

8. A sobering and important essay, nominally about the ongoing spat between Anthropic and the Department of War, but really about America and what the hell is going on in this country.

9. Clever and likely prescient thought about the future of the stock market in a time of ongoing A.I. Capex that is swallowing everything.

10. Site that can translate not just between any two languages, but between dialects, even conceptual/invented ones.

11. If I was 25 again, I would go here:

12. BYD continues to advance the single most important frontier for electric cars: Charge time. We also have to solve the problem of how to get the proper voltage to all current gas stations in the US, but it starts with having the cars that can actually charge at time parity to filling with gas. Our geopolitical rivals are basically there.

13. Interesting twitter essay on a topic that has long puzzled me and that I consider the next frontier in A.I. capability advancement: Long Term Memory.

14. Zooming out, I really enjoyed this discussion of Quantum Entanglement and the Illusion of Time. From the right perspective, none of it is really that important. It’s nice to remember that from time to time.

END

That’s what I have for you this Sunday! I hope you found something in these links interesting and worth your time. I hope everyone has a great week, and I will be back in seven short days with something fun.

Subscribe now

Share

Leave a comment

I hope everyone has an awesome week, I will be back next Sunday with something fun, and HAPPY BIRTHDAY DUCKY, WE LOVE YOU!