Lost betwixt the narratives of Skynet rendering humans useless and algorithms poisoning the youth is that computation was always supposed to replace human activity. More than replace, it was an effort to mimic it — the history of computation is essentially a meditation on how humans work. By replicating enough of our processes, perhaps we gain insight into the biggest black box of all: the human brain, and how intuition is formed.
While Becoming One with the Machine focused on the parallel processing relationship between eyes, brain, and body,
If the brain is the supercomputer of pattern recognition, the eyes are supercomputers of data processing. Critically, these are two separate-but-related instances, as eyes are capable of independently processing and driving an action without necessarily consciously processing it. We can call the independent subconscious action “reaction”, but obviously when the brain and the eye merge, that subconscious activity is where we get “intuition” from.
this is almost a mature state, if not an end state, of a thinking machine.
Something I have been thinking about recently is what “mature” really means. While I have a hyper-generalist, “everything connects” view of the world, frequent readers will note that I lapse into parallels with chess all the time (as I’m about to do now.) Indeed, for me, chess is essentially the Rosetta Stone for interpreting the modern world, from playing the 12-D version of it by solving markets to framing and interpreting developments in tech.
Creativity is generally associated with youth, and it’s no different with chess. If you look at Magnus Carlsen’s childhood matches, they’re filled with speculative sacrifices and deranged attacks. At a much lower level, my “rise” mimicked the same pattern — I would dive into ultra complex, highly tactical positions and rely on my intuition and speed of thought to wipe opponents off the board. I wasn’t concerned with playing accurately — I was trying to have fun and put pressure on my opponent the only way I knew how, by making their clock tick instead of mine, confident and assured that my speed of thought was much, much faster than everyone else’s.
What happened as Magnus aged? Upon reaching his 20s, stylistically, all the speculative complexities suddenly vanished. His games evoked pure simplicity, where keen judgment and evaluation reigned supreme. Why the shift? Well, he realized that, rather than creating complexity for himself, he could take it away from his opponent. He spent enough time in his head understanding what made him good at chess, and realized that his ability to evaluate a position’s strengths and weaknesses for both sides was better than anyone else’s. So, he would aim for positions where accuracy in evaluation, rather than calculation, would be rewarded the most. No doubt he’s up there with the best of both, but he’s completely unmatched in evaluation, and his results from 2013 onward reflect pure dominance. There will never be a human that surpasses Magnus in classical chess because there simply isn’t enough room for humans to improve. The limit has been approached.
In a chess sense, maturing is realizing that the possibility of variance extends beyond the board. When you’re a kid, all you’re doing is playing the game. The universe fits in an 8 x 8 realm when it’s your turn to move. When you’ve played enough, and made enough adults mad by beating them, this perspective shifts. Kids do not have a predictable skill distribution — adults with thousands of games in their belt do. Chess rewards stylistic advantages in the same way MMA does — reducing the complexity of a position is not an attempt to take advantage of your opponent, but to reduce the odds of running into a massive surprise, a deep intuitive insight that someone without 10,000 games of biases could think is a good idea. The game becomes about prophylaxis, about playing your opponent and pushing them out of their comfort zones, rather than winning on the board itself.
A key thing to understand is that being in your 20s is kind of old for a strong chess player. Certainly, through studious practice and repetition, one can start at any age and get their rating to a respectable level. But for the speed of thought and deep intuitive understanding to develop, one basically has to start playing by 7 years old to have any chance at getting very, very good. (Just think about how hard it is to become fluent in another language as an adult.)
“Deep learning” sounds like a spooky term, but the idea is very simple. If you define a universe with distinct boundaries and immerse a program with information contained within, without direction, the program will start to form connections and “learn” how to operate in the universe. If this sounds familiar, it should: teaching a kid a language follows the exact same concept.
Chess is how kids unlock deep learning as a generalizable concept beyond language. While the concept is found throughout sports, the bounded universe is that of an octagon or a ballpark, games that were created as a relief from boredom. Chess falls into the theoretical realm: games modeled after a real world process to develop and hone a skillset. (Hence Von Neumann’s insistence on developing game theory as a mathematical field, and why self-driving could be described as “a flow game created for automated vehicles to test their pattern-recognition abilities on real roads.”) It trains the skills of pattern recognition, foresight, and eventually operates as a test of one’s referential memory at the highest skill levels. It’s also deceptively simple to learn, as there’s only a few more unique moves than checkers. While I consider chess skill to be a symptom of intelligence, rather than a sign of it — egghead types will naturally be drawn to a cerebral game — chess skill in youth is almost certainly an indicator of an elite memory and visuospatial ability.
It is also a massive, massive waste of time.
Language is properly understood as a base of understanding so that you can interact with an expanded universe of individuals and ideas. Nobody spends their time reciting the alphabet or memorizing vocabulary off flashcards (the worst “aid” on the planet, but I digress), you write sentences and absorb context-dependent vocabulary so it properly stores in your internal “hard drive”.
With chess, on the other hand, you’re just playing a game. The only people who understand the language you speak are other chess players, and even then, there’s vanishingly few people who “get it” as you improve.
There are many people who purport to have the definitive solution to improving at chess, but this simply isn’t possible with deep learning problems. The inherent property of any learning algorithm is that you don’t know how the connections are made, hence my adage that any learning algorithm must solve a defined input, verifiable output problem to be useful. Again, the chess framing is delightfully simple: you have a defined input — pieces with static properties that can be maneuvered in a limited, 8x8 space — and a verifiable output — a hard win condition (checkmate) and draw conditions. Accordingly, the only way to improve at chess is to deeply understand your own metacognitive patterns. I don’t know how I’m coming up with my moves so quickly, but by scrutinizing the right and wrong moves and solving backwards to explain your own reasoning, I can dynamically adjust my weights for future games. It’s almost meditative, and it’s part of the reason why I have such an uncanny ability to rapidly verbalize and frame my own stream of consciousness. I’ve been doing it in some form or another since I was 4.
Intuition is only one part of the game, however. There is some rote memorization involved in openings (my least favorite part, and the reason I stopped playing, was the opening theory arms race) and endgames, and raw calculation ability. Sometimes, you just do have to sit down and do the work, and these are the parts that can be trained reliably.
This is the crux of why chess (and its variants), in particular, is so addicting and interesting even to the casual player, as opposed to a game like go: there are virtually no other games that enable such a high level of multithreading with no mechanics required. It’s why online chess and speed chess are a pale imitation of classical chess — I might be lower rated than many online specialists, but my understanding of the complete game is far beyond what theirs is — because so much of the complexity is stripped down due to computers and convenience. As a result, playing a great classical game is pure flow: it’s remembering your opening lines, figuring out if your opponent knows theirs, warming up your calculation brain, letting your brain wander and spot fresh ideas, and managing your clock, with the unique caveat that it is a pure skill game. There is simply no other game on earth like that — the closest thing I’ve seen is Starcraft 2, but there is an intense mechanical floor to enjoy the game on a comparable level, and the opening complexity of chess can’t be matched.
Naturally, this strains your brain quite a bit. We only have one core processing unit, as brilliant as it is, in our head — it’s the most powerful pattern recognition machine in existence,
The older I get, the more I realize that chess was the best training ground possible to learn to abstract the idea of man versus machine…
My own philosophy is that the human brain is the most powerful pattern recognition machine to ever exist. Of course, engines are significantly better at chess and are unbeatable at this point. (I call it “soft-solved” — chess mathematically cannot be “solved”, but the system design is limited such that a computer can never lose a game unless it is forced to play certain openings that allow for variance.) But how much processing power, code, and data is required to program a fork into Fritz or teach AlphaZero a pin? A moderately bright 5 year old could figure out the concept seeing it just twice and start to utilize it in their own games. They wouldn’t even need the full game data, just the positions.
Where the human starts to unravel is when they have to put a bunch of complex processes together.
but a computer is not limited in such a fashion. The human limit of chess is precisely the point where multiprocessing becomes necessary to improve. Accordingly, there is zero chance that Demis Hassabis’s chess background is a coincidence, and that a deep learning engine (AlphaZero) is what bridged the gap between brute force and becoming unbeatable. But more on that another time.
There are severe risks and tradeoffs associated with every addiction. The human brain is designed to maximize its reward function — walking away is one of the most powerful skills anyone can have. While I knew my intuition and referential memory was (and is still!) world-class, a shock to the system as to how outclassed I was in the realm of chess made sure I entered the real world and treated chess as a base, rather than a lifestyle.
And that’s the crux of the Morphy quote: chess should not be both a vocation and an avocation. Most of the greats in the 1900s had lives and professions: Emanuel Lasker was a brilliant mathematician, Mikhail Botvinnik was an engineer, and Capablanca enjoyed the high life. On the other hand, a singular obsession with chess can propel you to the top, but essentially overclocks your brain, best exemplified by Fischer. Certainly, the lack of state sinecures played a role, but much like chess thought was the precursor to AI thought, “LLM psychosis” is a direct descendant of “chess madness”.
The brutal reality of chess is that there is a hard cap on your skill ceiling. For Fischer and Carlsen, this is the limit of the human brain itself — for mere mortals, it’s when we run up against individuals with higher chess-specific memory ceilings than ours. No matter how much I study or practice, I will simply never come close to playing a dozen blindfold games at once.
This is a harsh thing to explain to a kid, and their abilities develop at different rates, so you don’t. Playing chess as a kid is about having fun, that dream that you can play in the NFL as well. The market and biological realities of pro sports make it evidently clear when you’re capped out in a different way — it’s not fun to be consistently outmatched, and when you see the physical abilities of a pro-caliber player, it’s much easier to “give up” and play casually. Plus, most sports are not entirely isolating — even tennis has doubles, and there is a social element to games like golf. After all, sports are a function of leisure first and foremost. It’s easy to modify these games to control for skill gaps.
There is no such thing for chess, and the tail is so exponential that it’s downright cruel to have to explain to a prodigy US junior champion that they’re not going to ever genuinely compete for the top rank. Fortunately, those kids are usually bright enough to do something else. Tech and trading are filled with former chess masters who learned to use their analytic capabilities elsewhere.
But for those that remain and persist, they almost get stuck in a loop, and the overclocked playing state takes its toll. There are only so many ways you can dynamically adjust your weights — you can’t rewire your playing style like you can modify a golf swing, one of the most programmed, fluid motions in all of sports. And, of course, you have to make money, which requires further entrenching yourself in the “chess world”. (After all, improving at chess is more than a full time job — at my peak, I was probably spending 50 hours a week outside of school studying and playing. For top players, I would venture that this hovers closer to 80.)
The chess world is not a particularly pleasant place. I have made many, many friends throughout my life in the chess world, but the fact of the matter is that such a cerebral, introverted game is bound to exist as a honeypot and an escape for certain types of individuals. You’re around a lot of people who struggle with social environments, are totally incapable of building social skills, and who experience stunted adolescence due to the sheer time it takes to build skill properly at chess. This only gets worse the higher you go, which is why so many of my friends stopped playing seriously when they went to college or got their first job. The “internetification” of chess made this worse — sure, more money started flowing in, but this created a whole new parasocial culture: one where 99.9% of people won’t have a clue what’s actually going on in your head, but where 100% of them have access to the tools that have solved the game, allowing an endless stream of cacophony to erupt from the peanut gallery.
Improving at chess is a Sisyphean struggle at every level past expert. Consider my experiences: I was on the brink of hitting Master elo just before I turned 15 years old. Not prodigy pace, but certainly within the reach of GM. I was stuck at that tranche for two years. Imagine that: hours after school and sports each day spent training, every weekend gone for tournaments, training sessions with a revolving cast of characters who promise to help you break through. Nobody knew what the problem was, so I had to explore my own black box as to what the barrier was. The lack of results or any sort of positive feedback make this a particularly soul-crushing experience: you start to question if there’s actually any skill at all, if all the prowess you made was just chance. Doubt starts entering everything you do, as you figure out how to start over in an area where you can’t, because it’s so deeply wired in. With a black box, as soon as it starts to recursively iterate, it becomes worse than useless — it starts spitting out such faulty output that it corrupts the thinking machine itself. (In Fischer’s case, this was “how do I prove that I’ve gotten better after reaching the top?) One negative weight can completely brick the system — one negative recurring thought that the ego is unable to expel can ruin the confidence of a chess player forever, rendering them unable to play properly. If you dig, there are plenty of instances of this, both in individual matches and in the ends of careers — the recursion begins over-the-board when someone doubts their calculation abilities and starts obsessively re-checking everything they’re doing. An old adage: A Grandmaster is not playing like a Grandmaster if he’s checking his variations twice.
What cracked me out of this doom loop was maturing — I simplified my play style dramatically. On a much lower level, the insight that Magnus had was properly explained to me — I didn’t need to plunge the game into wild complications, especially because I refused to memorize the opening lines that properly led to them. If I stripped out complexity and played simple, clear, open lines, I’d be able to find the drop of poison in a plain glass of water well before my opponent. After cracking that nut, I promptly crossed the Master barrier and was content to walk away.
The dampener of recursion corruption is ego. An indefatigable belief in my talent and abilities across everything I did made me confident that I could keep going. (Indeed, it’s the same mindset I apply to trading, and why it’s impossible to convince me I’m wrong about anything until I’m intuitively sure that I am.) There is a huge difference between narcissism and ego which is most apparent in people who were exceptionally good at something while growing up versus those who had it validated through status as an adult. There is a gulf in organic confidence in one’s abilities, probably entirely due to the fact that it’s wired in as a kid when you’re truly cracked at a skill game.
This is why I think the concept of “ego death” is particularly dangerous in the Western construct, either through psychedelics or exploring Eastern philosophy. As we’ve with “ayahuasca oneshots”
or within postmodern dance, ego death has a decent chance at completely scrambling your brains: you’re not supposed to mess with your base programming on that level and start writing internal Assembly all over again.
The chess world has become particularly evil, because the culture is wantonly inducing this toxic recursion in people. The crux of a black box is that it’s unexplainable — so when someone is accused of cheating, it’s essentially an unfalsifiable allegation. An accused individual is faced with the kafkaesque problem of explaining how their intuition works, while morons use “statistical analysis” to try and assign causality to something that doesn’t fit within the realm of statistics. Even if your supporters believe you, there’s no way to completely erase doubt, and bad actors manipulate the circumstances to create psychological dissonance, either for their own amusement or to rule the meaningless anthill this world has created. The accused has the options thrust upon them, through no fault of their own, of walking away from the game they’ve devoted their life to, or endlessly explaining that which cannot be explained to those incapable of properly listening and understanding what you have to say. There isn’t really a more cruel fate for individuals with ultra-high levels of cognition to deal with.
The core of LLM psychosis is the inability to properly communicate and be understood by another human. Eventually, the siren beckons, or the internal loop of explaining yourself exhausts itself. The worst part is, we impose it on one another with every accusation that something is “AI generated” or treating “statistical evidence” as an immutable truth, as if statistics aren’t approximations themselves. If chess is so indicative as to the structure of tech, what does the trajectory of the chess world imply about networked reality?
Increasingly I fear that intuition is dead, and we killed him.