Unambitious (v3)

Note: Final version. I’ve added 1,600 words onto yesterday’s version. I am publishing this work as a new post to respect the Inkhaven contract, but I think logically & aesthetically all this content belongs in one place rather than as a series of posts. New work is below the divider.

Warning: Navel-gazing wank!

I am unambitious in a way that is hard to describe. When I try to explain this to people who know me, I come away feeling like I was subconsciously humble-bragging, fishing for compliments, or just generally being false and tedious. Friends are typically confused and start reflexively praising me (which is pleasant but still distressing). To them I seem pretty complete: I have a lot of hobbies that I’m good at, I read a lot, and I’ve worked some competitive, high-status jobs. Nonetheless, I think I am significantly less ambitious than many in my reference class. I’ll describe this in a couple ways.

Brazilian Jiujitsu

I’ve made enough money such that I don’t feel the need to earn (if AGI-driven super growth doesn’t happen, maybe I’ll have to work eventually), so I’ve been unemployed and doing a lot of bullshit. The main bullshit I’m doing is training bjj six days a week, frequently twice a day. I’m now into year 4 of jiujitsu and I’m a good blue belt hobbyist. I’ve successfully recovered from some serious injuries without quitting, so I’m now pretty certain that I will ride this hobby out long enough to at least get a black belt, which will probably take me 7 more years.

When I first started doing a lot of training, I had an idée fixe that I was optimizing to be really good. The whole appeal of jiujitsu is that it is real: you can pressure-test it at more-or-less full force, all the time. This weeds out trash moves, and prevents the art from becoming aikido (which due to its lack of pressure testing has degenerated into make-believe for adults). Being really good means that my grappling needs to keep up with gym-mates who are training for amateur competition. It means studying the top athletes: Gordon Ryan, Roger Gracie, Marcelo Garcia, and their ilk. It means willing my jiujitsu into becoming the blue-belt version of theirs: sharp, tough, and overwhelming.

I watched the Craig Jones Invitation 2 tournament (day 1, day 2) — the biggest nogi bjj tournament of 2025 — and had a revelation: competition jiujitsu sucks ass. Competitive wrestling, judo, and jiujitsu suffer from the problem that the matches themselves are transcendentally boring. These are all “participant sports”, meaning that the only people watching the competitive circuits are the people who do the sport (and their families)¹. Competitive jiujitsu does not have a standardized ruleset, so there is a lot of experimentation done that is meant to make the sport exciting. Every year some big new idea comes out that is going to fix everything — and then it sucks ass. Craig Jones — who individually is responsible for many of jiujitsu’s most exciting matches — put a lot of thought and effort into making an exciting event, and he failed overwhelmingly². But this isn’t his fault! Competition generally optimizes strongly for being boring: two evenly-matched competitors are not going to get a chance to display the fun stuff. Most of the most entertaining bjj matches of recent memory have been either great grapplers getting a chance to stunt on absolute tomato cans, or young prodigies mutually agreeing to maximize entertainment value. What made CJI 2 especially sad to watch was seeing how upset all the competitors were with themselves: everyone knew they were being boring and weren’t showing why the sport is beautiful.

This morale implosion got me talking to my jiujitsu friends, and we’ve come to the conclusion that bjj is fundamentally not about competing or about “being good”. It’s about trolling the fuck out of people worse than you.

This is the Craig Jones mindset.

Here’s a nice video of Craig Jones clowning on people at a seminar. At the 5:50 mark, he rolls with a young kid who starts going way too hard. Craig Jones puts him in a silly submission from bottom called a buggy choke. The kid tries to escape by putting his knee on Jones’s neck (this is rude). Jones then wags his finger at the camera and proceeds to unleash the most diabolic move I have ever seen.

Moments before disaster

He unlocks his buggy choke, pushes the kid’s arm back, and laces his leg under his arm into what I suppose you could call a “clothesline buggy choke” (this move has no name because no one besides Jones would even think to do this). He then sweeps him, and punch chokes the kid’s trachea with his free hand.

What in the fuck is this

This is peak

All I want to do now is nicer versions of this. I have been training leglocking a lot recently, which by reputation is an especially complex and arcane subsystem of nogi grappling. Leglocks are fun because they can really confuse people who don’t train them enough. I have started submitting people from bizarre leglocking positions that aren’t good and don’t have names. Here’s a move I do:

What now tough guy?

I call it “under the sea” because it forces my opponent’s knees together like they are a mermaid. I wrap and lock my legs below and around the outside of your legs, which knocks your knees together. It is a surprisingly tight hold: it pits my leg adductors against your leg abductors, which is a pretty rare combination. I then am free to split your feet with my arms, grab the outside of your heel, and rotate it to torque the ligaments in your knee. I sometimes get people to tap here prior to submission because the situation itself is so absurd. I want all of jiujitsu to feel like this.

So I go to jiujitsu a lot, to enjoy the feeling of learning. But I don’t see much value in “the peak”; I can only see the margin.

Roonian ambition

Roon, when asked by Liron Shapira about what is exciting about working in AI, said:

For me personally I want to see works of excellence. Just, more amazing things in the world. One of the dominant frames of why AI progress is exciting is “we’re going to automate so many things, and GDP will go up…” and it’s true but it’s also super boring. I don’t actually think anybody ever has been truly stirred by the idea of increasing GDP — although maybe some people have. I think when they say that, they really mean is something else. Advancing the frontier of human greatness, or something. Lifting people out of poverty, it’s a thing to do. It’s almost like an achievement, an accomplishment.

Oftentimes it’s less about the poor people than it is about — like, “I eradicated polio”. That’s a pretty cool thing to say. There’s an incredible, unmeasurable amount of human suffering associated with that statement that was erased from the world. I think Bill Gates once said “the legacy of ‘I erased polio’”. I think it’s okay to admit that.

My transcript makes it sound like Roon is being flippant, but in context it’s clear that he is searchingly describing a mood. Nevertheless, it is plain to me that I have no access to said mood and I really don’t see the appeal. The glory of Norman Borlaug is, to me, completely coextensive with the prosperity he wrought, and has nothing to do with the fact that he did it or with the idea that his actions constitute an “accomplishment”. The false promise of Roonian ambition (i.e. the pursuit of great ends qua personal accomplishment) is that advancing the frontier of human greatness will cause humanity to bequeath to you an elated status that will outlive you. Where does this ambition come from? Perhaps in the Bay Area in the 21st century, it comes from reading The Fountainhead at a critical developmental stage. Or, perhaps more generically, it is the reification of our ordinary status instincts: we feel there must be more to life than quotidian status games, so we put all our hopes into winning the eternal status game.

There is no such eternal status. One should make peace with the notion that though you may not be forgotten, you certainly will be misunderstood. I remember when my girlfriend’s father died, his brother gave the following eulogy:

He was the joy of the table. He would order for everyone and made sure no one left hungry. Sometimes though he would order a little too much! **knowing laughter**

One time he took my wife and our three kids out to dim sum and he ordered for the table. We ate great, but long after we were all full we still had round after round after round of food coming out, and we were all laughing hysterically, not knowing what to do — laughing at the absurdity of it all.

That’s what he thought was happening. What actually happened was the following. This uncle’s family is full of picky eaters: his daughter would want har gow but would only eat the skins; one of the boys was vegan but would eat any “leftovers” (so they had to strategically over-order); the other boy ate no seafood; others would only eat seafood; some would refuse to eat chicken feet, while others needed it for a collagen infusion. My girlfriend’s father was left to do all the ordering (because he’s a fancy food critic), so predictably he ordered way too much. Instead of being pleasant company about this, the family laughed at him and mocked him so much that he had to exit the restaurant, go to his truck, drop his forehead onto the steering wheel, and bawl his eyes out for ten minutes.

I couldn’t believe what I was hearing at this funeral. My girlfriend’s uncle had misunderstood his brother so profoundly, and titrated his life down to a falsehood.

Perhaps time is kinder to those who are great. I suspect not. We quickly start misunderstanding them. And really, we reduce them to cartoons. George Washington’s great wooden teeth, and his cherry tree; Napoleon’s height, his hat, his wife…

I like Roon enormously. He’s funny, insightful, and sensitive. He’s willing to go on Shapira’s “Doom Debates” despite working at OpenAI because he is capable of holding multiple perspectives in his head at one time. But really I do marvel at how he and the rest of the AI capabilities folks can go about cavalierly pushing forward the frontier of intelligence knowing that they are tempting fate. I suspect it’s because the Roonian superego has been chafing against two decades of ad-tech, gambling, b2b SaaS, crypto pump-and-dumps, social media eyeball-algo-maxxing, div aligning pull requests, yet another Forbes 30-under-30 recipient fraud trial, Nick Land’s yapping, Sam Bankman-Fried’s blogging, Elon Musk retweeting Catturd, Vanity Fair’s Bay Area sex party exposés, jhana retreats, and tradwife longhouse discourse. What it wants is to come into contact again with the open sea John McCarthy saw when he wrote the Lisp meta-circular evaluator — some sign of transcendence. They want to be there, and to be with other people who want to be there, tossing the biased-coin of Armageddon without modelling the odds (as Roon says, he “studiously avoids having a P(Doom)”). This is how you can get the sort of situation David Duvenaud describes where you can go around Anthropic asking the tech leads “so let’s say we succeed in building superintelligence — what will human civilization look like ten years after that?”, and they can respond with a full heart, “I don’t know. I don’t know how to think about that.”

Puzzles, researchers, and hackers

I was laid off from my Google machine learning team two years ago. My direct managers were happy with me, I think. I was in the middle of pushing out a PR at 7 PM in the cafeteria when my terminal died and I got locked out of my email. I thought at first that some backend service had crapped out, but quickly realized I was being canned. I rushed back to my desk before my badge access died to say goodbye to my team, but no one was there. I considered absconding with an expensive monitor or two, but didn’t.

Google gave me one month to interview internally for a transfer. Surprisingly, I received two offers. One would have been a lateral move, the other downwards. I realized I was mentally stuck and just scared, and walked away. My manager video-called me to present a goodbye slide-deck that ended with a GIF of a cat waving goodbye. There was a very moving note somewhere in there from an old manager. I asked for a copy of the presentation but they never emailed it to me.

Since then, I have just been playing with puzzles. Jack Lance created beautiful things which you also should play with. I have been doing contest math questions from the Art of Problem Solving meant for 9th graders. I like Israel Gelfand’s book Algebra, which is intended for smart children. He teaches you what multiplication is and then straight away asks questions like this:

Which is nice.

It would be natural at this point in life to veer toward acquiring an advanced degree in math. But grad school seems so awful. And what’d be the point anyway? My unambition tells me not to push the edge of mathematical research. The wonderful thing about math is that there’s already so much of it at every conceivable level of difficulty. Without the burden of school, you can just choose whatever mathematical stimulus is optimal for your skill level and learn that — as if you’re Mark Rippetoe judiciously meditating on whether to add two 1/2 pound plates onto a 495 pound deadlift.

I’ve been reading a lot about the MIT AI Lab in the 1960s:

Hacking had replaced sex in their lives.

“The people were just so interested in computers and that kind of stuff that they just really didn’t have time [for women],” Kotok would later reflect. “And as they got older, everyone sort of had the view that one day some woman would come along and sort of plunk you over the head and say, you!” That was more or less what happened to Kotok, though not until his late thirties. Meanwhile, hackers acted as if sex didn’t exist. They wouldn’t notice some gorgeous woman at the table next to them in the Chinese restaurant, because “the concept of gorgeous woman wasn’t in the vocabulary,” hacker David Silver later explained. When a woman did come into the life of a serious hacker, there might be some discussion “What’s happened to so-and-so ... the guy’s just completely falling apart ...” But generally that kind of thing was not so much disdained as it was shrugged off. You couldn’t dwell on those who might have fallen by the wayside, because you were involved in the most important thing in the world hacking.

…

After he dropped out of school, Greenblatt had taken a job at a firm called Charles Adams Associates, which was in the process of buying and setting up a PDP-1. Greenblatt would work at their offices near Boston’s “Technology Highway” outside the city during the day and drive thirty miles back to MIT after work for some all-night hacking. Originally he moved from the dorms to the Cambridge YMCA, but they booted him out because he wouldn’t keep his room clean. After his stint at Adams, he got rehired at the AI Lab, and though he had a stable living situation as a boarder in a Belmont house owned by a retired dentist and his wife he would often sleep on a cot on the ninth floor.

…

He was turning out an incredible amount of code, hacking as much as he could, or sitting with a stack of printouts, marking them up. He’d shuttle between the PDP-1 and TMRC, with his head fantastically wired with the structures of the program he was working on, or the system of relays he’d hacked under the TMRC layout. To hold that concentration for a long period of time, he lived, as did several of his peers, the thirty-hour day. It was conducive to intense hacking, since you had an extended block of waking hours to get going on a program, and, once you were really rolling, little annoyances like sleep need not bother you. The idea was to burn away for thirty hours, reach total exhaustion, then go home and collapse for twelve hours. An alternative would be to collapse right there in the lab. A minor drawback of this sort of schedule was that it put you at odds with the routines which everyone else in the world used to do things like keep appointments, eat, and go to classes. Hackers could accommodate this one would commonly ask questions like “What phase is Greenblatt in?” and someone who had seen him recently would say, “I think he’s in a night phase now, and should be in around nine or so.” Professors did not adjust to those phases so easily, and Greenblatt “zorched” his classes.

All of this is so romantic and unambitious to me. John McCarthy and Marvin Minsky definitely had sweeping visions for how technology would steer humanity, but their “coolies” Greenblatt and Gosper seem to me to be all about life at the margin. Like, read a bit of Gosper’s monograph on the arithmetic of continued fractions. Yes it’s true, this stuff is useful for hacking, and so it did “push the frontier of human achievement” in some sense, but really it’s just interesting because it’s interesting. Gosper is still out there giving talks at the Gathering 4 Gardner about dragon curves, a subject which no one on Earth has a single reason to care about outside of the fact that they are beautiful and beguiling and they are still hiding mysteries from us 60 years after Martin Gardner first described them.

My ambition is to be invited to the Gathering 4 Gardner, and to be able to show off something beautiful and beguiling.

---

Leaving

After leaving Google, my parents gently encouraged me to leave Los Angeles.

I could return to New York City — the city I love most — and look for a job. At the time, I didn’t have a driver’s license (an absurd folly for someone who had at that point lived in Los Angeles for five years), so a move to NYC would allow me to, you know, actually walk and see people. Jonathan Gold said that Los Angeles is the greatest food city in the world, and I believe him. But I wanted to eat like I did growing up: gripping a hefty Sicilian slice while walking, sitting down for a real plate of mofongo, tearing at a fresh Jamaican beef patty, and (as corny as this is) ordering a bacon egg and cheese on a shitty kaiser roll with ketchup. I wanted to have a stack of that crappy salmon & avocado makizushi they sell at the Deluxe Food Market in Chinatown, which my mom used to buy twice a week. Or, get that absurd eighty-piece Canto-fusion sushi-platter from Kam Man Market on Canal street. I could turn my nose up at Katz’s Deli, and preach to mono-coastal nudniks about the superiority of Langer’s pastrami on rye. I probably wouldn’t get to live in the Manhattan of my childhood though — I’d rent a unit near my parents in Elmhurst, Queens. So I’d get to have some nice char kway teow at Good Taste or Coco’s (R.I.P.), and walk down to Rego Park to study the Uzbek plov. I could visit my friends up in Astoria, have the best Greek food in the Western hemisphere, and finally learn how to pronounce the name of the world’s most perfect dessert: galaktoboureko.

I also had friends and acquaintances inviting me up to live in the Bay Area. It would have been an ideal opportunity to take a pay-cut and do whatever needed doing in the EA/rationality cliques. Maybe I could hang around some MATS scholars and fake it till I make it. Or I could go all-in on becoming absolutely cracked by working 15 hours a day and diving head-first into a series of poorly-conceived startups.

Or — my parents reasoned — if I wanted some solitude I could return to our home in Minnesota. Reclaim my birthright: the Midwestern identity that only had 8 years in gestation. I could fix up the home and stare at the deer. Learn how to fish — learn how to ice fish! Minnesota millennials probably don’t make all the casseroles their parents did, but they definitely still make hot-dish. Apparently Minnesotans don’t make new friends, but I think I could crack through. Reddit tells me the jiujitsu gym near me is run by a former Proud Boy. I could go and see what that’s all about.

So there were all these futures available to me. And there were so many reasons to leave Los Angeles: my romantic relationships had wound down, so I was living alone in a home meant for two people; the rationality meetup I ran felt ready for a baton-passing; some previously intense friendships had transitioned to a maintenance-phase; a bunch of drama caused the instructors at my old bjj gym to turn on each other; and my landlord was creatively testing the limits of L.A. County’s rent control laws. I felt everywhere this ambient mood of pre-finality.

Staying; or: the Lie of the “Dojo”.

But the thought of leaving my friends and starting over again was too agonizing. The downside of running a weekly rationality meetup is that you have to get used to interfacing with tedious people³; the upside is that every 1.5 years, you form a life-trajectory-changing friendship. I still had four such friendships (among them Abner, who still had much grappling arcana in store to pass down to me), and I hated the idea of deprecating them.

For my money, friendship needs frequent contact to sustain. Every time I come to the Bay Area I meet multiple rationalists and effective altruists who don’t think this is true (or think it’s true but unimportant). This is the most prominent anti-wise character trait of this crowd that I can detect. I have three memories from the pre-FTX-collapse era that illustrate this attitude.

I remember having a discussion with a well-networked rationalist about my lack of desire to uproot my life to take up a high-impact job. He said:

Those friendships you leave behind are still there, even if they are latent. I think of it like this: Imagine a dojo. You train very hard for years and develop this very close relationship with your sensei. You are with him every day in the dojo. Then one day you earn your black belt, and it is time for you to go out into the world. Perhaps to start a dojo of your own. But no matter where you are in the world, you will always have that relationship.

Very picturesque. Unfortunately, I’ve actually met martial artists. The modal relationship between black belts and their teachers is characterized by mutual hatred, distrust, and envy. Either that, or apathy. I know multiple gym-owners right now who are hesitating to give black belts to certain students they know will immediately branch off to start their own gyms nearby to undercut the parent business. The Gracie family has completely balkanized, and they all hate each other, due to dynamics like this. In any case, it is rare for these relationships to remain close, even in some abstract and metaphorical sense.

Another memory was from a young, twinkish EA ops person who had just finished doing a three-month stint in London, following a three-month stint in Sydney, following a three-month stint in NYC. He was now struggling to find deep friendships in Berkeley. Instead, he was receiving more sexual attention than he knew what to do with. But he was happy to help. Maybe working 14 hours per day would compensate (impact-wise) for his sub-par mathematical ability.⁴

The last memory was talking with a woman who just moved down from Seattle to take a job with Anthropic. She seemed bored when I said I’d have a hard time leaving my friends to execute a career-move. “Well, I’ll still see those friends once or twice a year. That’s all I really need to keep those relationships alive. We could be living in the most pivotal years in human history. How could I be anywhere but here?” Got to see what happens next, from the best seat in the house. I was reminded of this exchange this year reading this blog post from Trenton Bricken (another Anthropic employee):

It may be hubristic to say, but at times it feels like unless you’re doing tech in SF then you aren’t serious about it.

…

I plan to keep living here for the foreseeable future. The combination of being at the heart of the AI takeoff and incredible nature makes it a great place to be for now.

Today, I’m back in Berkeley. A few days back, an old friend pitched me on working on AI alignment. “Better late than never. Who knows?” He was hesitant though: “Smart people tend to move here to work on X-risk reduction, and then a year later they are working at a capabilities lab. Some of them do safety-washing. Others do ‘alignment research’ that consists of trying to build baby-versions of the torment nexus, and then stress-testing it like the scientists in Starship Troopers. Most end up just working on straight capabilities. Their co-workers are smart and inspiring to be around, I guess.”

Unambition may be good for something.

Rebirth

For better or worse, I stayed in Los Angeles. I did a bunch of travelling (Paris twice, Tokyo twice) and trained jiujitsu casually to pass the time. Then towards the end of 2024 something wonderful happened: one of the best jiujitsu instructors in America was moving to Los Angeles, and was going to open his gym a block away from my house. Construction was delayed for months. It is excruciatingly embarrassing to say, but sometimes I would wake up at 3AM, walk over to the construction site, press my face against the glassy darkness, and have reveries about what sort of life I would have on the mats. Surely, I was setting myself up for disappointment?

No, it was as wonderful as I thought it would be.

My life is replete now with unambitious activity, highly decorrelated in their variety. And all the signs of Los Angeles ending for me? All reversed. My romances have gained a second-wind. My friendships have deepened. And I am calm.

Appendix: List of Activities for the Unambitious

• Do some Open University modules

• Make a topology and differential geometry reading list and road-map

• Get a bjj purple belt

• Do Inkhaven

• Read Philip Larkin

• Join inaugural cohorts

1. Sumo is a notable exception to this among grappling arts.

2. The way CJI 2 ended was one of the funniest pieces of drama I’ve ever seen, and the highlight of 2025 for me. A prime example of reality being funnier than fiction. Nonetheless, watching most of the event live was punishingly dull.

3. This is a minority, to be clear! But rationalists tend to maintain social norms that result in conversations being easily hijacked by tedious people.

4. I was reminded of this after reading Harri Besceli, who is writing for Inkhaven a series of funny, penetrating, and grotesque blog posts about social anti-patterns in Effective Altruism. Read first his piece “Promises”. Then read this and this and this.