CS · HCI · Econ · Widely interested · Pathologically curious
Joined January 2009
- Tweets 20,122
- Following 2,864
- Followers 2,551
- Likes 127,101
3,658 Photos and videos
made a custom Garmin watch face somewhat inspired by the LED clocks you sometimes see in broadcast studios
1
6
183
the big font is Nose Transport (designed for transit displays by Zurich design agency Nose; nose-design.github.io/Nose-T…), the small-digits font is DIY'd in Glyphs (still needs minor tweaks, a 5x3 dot matrix font is probably as far as my patience will ever take me in type design lol)
1
3
124
might publish/put source code on GH once I've cleaned things up a little
2
40
Robert Caro on meeting his editor Bob Gottlieb (from "Working")
3
1
14
517
on finding community (see also x.com/bschne/status/19215445…)
11 May 2025
a big hairy motivational issue when you're working on tricky stuff in isolation is that it is very hard to develop a sense of what the normal amount of effort for something to take is, so when things get hard or your fifth attempt isn't amazing either you conclude you can't do it
4
156
I can‘t babe, I have to re-run my model so that when I inevitably lose the office world cup pool I can say it‘s fine because I made maximum EV bets and that‘s what really counts
1
122
me finally getting my sleep back on track vs. random presumably drunk person repeatedly ringing my doorbell at 3.30am who would win
2
121
quick messy attempt at summarizing my current mental model of great teaching in one tweet:
1. start from existing want/goal/interest/desire
2. show/outline/give a taste of how it could be taken further
3. help scaffolding/project-managing/breaking down to get there
4. repeat
I think there's a whole philosophy of education in here that I've been circling since I was 14 and annoyed/bored in school
1
5
545
3. bridges the skill gap to a point where it's hard enough that you learn something, but not so daunting you give up or can only take steps so small there's no reward at the other end. this is partially about providing psychological safety for even trying.
1
1
42
4. a new project maybe, but I think iterating on what you achieved in 1-3 to take the end result as the starting point is even better because most things worthwhile in real life aren't done in two weeks, and insights along the way usually surface new stuff to tackle
1
37
agi will have been achieved when you tell Claude „you were right to call me on that“ more often than the other way around
2
100
if you overengineer something and it doesn't work great, at least you will have had fun the fun of overengineering it
107
offer: if you let me into your life, I will fix one (1) broken thing for you, for the low price of starting to take apart but never quite getting around to putting back together twenty-three (23) broken things
2
1
13
486
Benjamin (offline for a bit) retweeted
Jun 11
‘yeah but I have access to you’ 🥺
26
16
1,784
65,273