Claude Mythos and the tower of holes

“This is the EXPLOIT CHAIN!” –Claude Mythos, while devising a privilege escalation attack on a configuration file I previously wrote about how our civilization looks a lot like a tower of holes, constructed almost entirely out of weird edge cases we’ve discovered in Nature and exploited. In turn, society itself is often deeply vulnerable to … Read more

On getting unstuck

After more than a year of trials and new models, Anthropic’s Claude AI has finally managed to beat Pokémon Red. The writeup that clued me in to this is worth a read; the story of Claude’s many failures leading up to its success are frankly hilarious. There’s even a catchy song. There was no clear … Read more

Civilization as a tower of holes

Epistemic note: This essay conflates two related concepts, the tendency of humans to build civilization by exploiting Nature’s rules and the vulnerability of that same civilization to exploits in turn, under one name. I’d like a better name for these two concepts and am open to suggestions. There is a kind of mindset that pays … Read more

Book Review: The Phantom Tollbooth

To avoid spoilers, skip the first section. Twenty Chapters in Twenty Sentences A boy named Milo feels lost and listless, until one day he opens a mysterious package containing a cardboard tollbooth and a one-sentence note. He drives his toy automobile through and finds himself in another world. He meets a curious dog named Tock, … Read more