On Doubling

When I was nine years old, I was terrified at how big and scary the world was. I saw my elders and their advanced degrees, their impressive contributions and lofty friends, and I could hardly imagine contributing so much to society—moreover, I saw how big the world was, millions of children the world over, millions of other minds, and I wondered how I was ever supposed to compete. I sort of consider this the start of my life, because at this moment I decided there were two ways for my life to go: either to be the best darn housewife out there (embarrassingly) or to be the best programmer in the world. And so I set down my video games and I got to work.

I threw myself at the things I thought would matter. Some of them did—deep learning, web development, systems knowledge. Others... did not. (I memorized the dictionary because I thought it would make me smarter. In fact, it just ruined my writing for the years it took to realize that biggerer word does not always mean betterer sentence. I optimized for the wrong things plenty of times, and it worked out alright for me — the important thing was that I cared.) I read Japan 1941 and as much as I enjoyed the history, I looked up to the kamikaze pilots, willing to die for a lost cause. ("death before dishonor!" I said, before another late night of half-awake reading.) I still want to be the best, but I think there's a better way to frame these impulses that's more productive.

I just want to double every two years. Moore or less.

I want to look back at myself two years prior and say confidently that my life is twice as good in the metrics that matter—that I am twice as smart, twice as capable, twice as excited to be alive. I want my friends to be two times as close and as cool, my projects to be twice as renowned, my expected value to follow not some sorry sigmoid but an ebullient exponential (that dictionary reading hasn't quite left me yet, has it?).

This is a bit corny, but I think it's a really useful bias to have, because it means you have to avoid marginal goals. To climb one rung of the corporate ladder is a great thing, but it is hardly twice as good on a two year timescale. You want to be not just to become better, but to become unrecognizably better. You want past you — who you respect a lot today — to seem naïve and silly. And you need a concrete vision for a future self you'd admire twice as much. For me, that was shipping software to millions at chess, building brain chips at Neuralink, flying coast to coast as I managed school/work full time, doing machine learning research, leading neural speech, and now... well, my next adventure must be twice as impactful. And so I'm going to keep reaching with that same stubborn vigor, damnit.