Let’s condense the content of goodness(universe) to the following two principles:

  1. Do good work on cool important problems
  2. Interact meaningfully with other sentient lifeforms
  3. Be healthy (ok this is technically auxiliary, but is worth listing anyways)

Ok, so we know , but how do we do ? Well I guess momentum is supposed to be helpful. How do you build momentum? Well, it could be useful to keep track of things that worked in the past, and then keep doing those. Hence this list.


  • 0711: read a lot of permutation avoidance literature; made a fun “relative error supported on subcube” tester.

  • 0712: read a bunch more about permutations. tried to do better permutations.

  • 0713: a good game of bottichelli with fun “abstract” clues

  • 0714: Made banana bread. Had some vaguely promising ideas for analyzing cancelling scheduler.

  • 0715: Maybe got an analysis of 1.5-cancel? at least drew some good pictures and had some interesting ideas. Great property testing meeting. Tons of promising directions for future work.

  • 0716: I worked really hard on diamond test large epsilon case. it didn’t work out, but was still good effort.

  • 0717: typed up a bunch of stuff for property testing project. was really good to solidify what we know. also permutation party!

  • 0718: null (forgot to record)

  • 0719: nice proof of affinity square in cube.

  • 0720: typed up some cycles.

  • 0721: went to the beach. it was pretty good.

  • 0722: had some fake scheduling idea, but was still fun to think about it and had good conversation about it.

  • 0723: thought about direct product test. understood what is going on a bit better.

  • 0724:

  • 0725: idea to use induction for diamond test

  • 0726:

  • 0727: realized that the induction diamond thing was wrong

  • 0728: “ok lol new idea: analyze a 4-function version of the diamond test”

  • 0729: grinded 4-function hypercontractivity. ~12 hours later it works

  • 0730: grinded diamond writeup

  • 0731: read some DGA

  • 0805: made some good hash browns

  • 0806: made some good pizza.

  • 0807: I think we found a -competitive vanilla scheduler! (edit: it’s not quite this good, but it’s pretty close)

  • 0808: came up with some good notation for the scheduling paper :)

  • 0809 zumba; good progress on scheduling paper.

  • 0810 zumba; good progress on scheduling paper

  • 0811 walked around castle island


I think this was a fun idea from the past. might restart keeping a log of good happy things that happen here.

0216 I came up with some research questions! This has been my main goal for it feels like a while now. I’m very happy to have them written down. I guess I had a lot of ideas scattered around. but some are hard so its hard to commit to them.

Luckily I think i’m actually pretty good at optimizing for deliverables. By which I mean now that I have these ideas written down I can

  • commit to work on one for a bit
  • have a goal of publishing a professional blog post about it
  • go do it
  • potentially find team-mates to join
    • there are so many smart ppl around mit that want to do something to save humanity!
    • the following ppl might be down to join a project based on recent conversations:
      • DL, SZ, B, CZ, D, probably lots of other ppl too, espec lots of empirical dudes.

Anyways, hyped I am!

Okay I mostly resolved one of the research questions. It seems kind of interesting and nice. There are at least 2 major obstacles for it being useful, but I’m glad I wrote about it, and am excited to try to publish a polished version of this in a bit (eg tmw).

I’m pretty excited to think more about myopia. I think starting with some smaller projects that can be completed in like a week seems like an exciting way to gain traction at which point I can start attacking some more tricky questions.