Published on April 10, 2025 5:39 PM GMT
When I was a really small kid, one of my favorite activities was to try and dam up the creek in my backyard. I would carefully move rocks into high walls, pile up leaves, or try patching the holes with sand. The goal was just to see how high I could get the lake, knowing that if I plugged every hole, eventually the water would always rise and defeat my efforts. Beaver behaviour.
One day, I had the realization that there was a simpler approach. I could just go get a big 5 foot long shovel, and instead of intricately locking together rocks and leaves and sticks, I could collapse the sides of the riverbank down and really build a proper big dam. I went to ask my dad for the shovel to try this out, and he told me, very heavily paraphrasing, 'Congratulations. You've cracked the river damming puzzle. And unfortunately, this means you no longer can try as hard as you can to dam up the creek." The price of victory is that the space of games I could play was permanently smaller, and checkmate was not played on the board. This is what growing up looks like.
I experienced many variations. After I put a hole in the ceiling, building K'Nex catapults switched from a careless activity to a careful activity. On the more supervised side, when I was curious how big of an explosion I could make by taping together sparklers, I was not allowed to find the limit. After I cracked the trick of tillering, I could no longer try as hard as I could to build bows and arrows for the backyard battles my friends and I engaged in. I was and am particularly proud of earning this ban.
A couple years back, I got a job offer from an investment bank to help them win zero sum games against people who didn't necessarily deserve to lose. I had tried very hard to get that offer: leetcoding, studying build systems, crafting my resume. It was only once I got it that I realized I no longer could play the game "make as much money as I can." I had found the equivalent of smashing the river banks with a shovel and needed a new game.
There's an enormous amount of nuance here. If I find something that's sufficiently stronger than me, I can play hardball for a long time. My favorite game at the beach is playing in the outflow of a tidal pool at low tide. The river is pouring across the sand, but I, a child again, am stronger than river. My hands can dam it, make lakes, deltas, and canyons. The incoming tide is so vastly stronger than my hands that any consequences from me trying as hard as I can are erased.
Here still the ratchet of my understanding keeps closing off avenues. I could move past pushing sand with my hands or a shovel and start lobbying city council members to put in a groin or seawall, and seriously move that beach sand. And so I can't (and wouldn't really want to) actually try with my full effort. And even digging with my hands, I see the signs of innumerable coquina clams burrowing back down after every wave, signs that were invisible to a 4 year old, and I wonder: Can they really get back to the surface after being accidentally and swiftly buried in 2 feet of miniature civil engineering? I suspect that most of them can, and I continue.
This is, of course, about artificial intelligence development. Trying as hard as we can has been a fascinating and genuinely rewarding adventure. Personally, succeeding and blowing off my hand with the metaphorical sparkler bomb, would be less rewarding. This would be true even if I was a race with my neighbor to see who could blow off their hand first, or if, stretching anatomy and the metaphor, we all shared one hand.
On the positive side, Anthropic appears, in their latest report on usage in education, to be at least noticing the cocquinas.
Discuss
