a ton of crates
tonofcrates.bsky.social
a ton of crates
@tonofcrates.bsky.social
contrast
November 15, 2025 at 10:55 PM
actually really nice to open an epub and not have a ton of ceremony or latency
November 15, 2025 at 10:44 PM
oh if you're looking for the dictionary it's right next to my 300-page tome of Clip Art
November 15, 2025 at 10:09 PM
November 15, 2025 at 10:05 PM
ya our garbage collection is actually web scale
October 28, 2025 at 9:51 PM
?????
October 18, 2025 at 5:37 PM
i regret to inform you this means mojo is praxis
October 5, 2025 at 2:14 PM
here's a carousel of baggage
October 4, 2025 at 7:04 PM
✅ ᴛᴇꜱᴛꜱ ᴘᴀꜱꜱᴇᴅ
✅ ʙᴜɢ ꜰɪxᴇᴅ
✅ ᴜɪ ᴛᴡᴇᴀᴋᴇᴅ
✅ ᴘʀᴏᴅ ʀᴇᴀᴅʏ
September 28, 2025 at 10:26 PM
there's a kind of brain damage that turns you into an LLM. cool cool cool
September 23, 2025 at 1:12 PM
epub to kindle converter producing some solid poetry
September 1, 2025 at 1:26 PM
man was never meant to explore beyond ℚ
August 17, 2025 at 3:46 PM
you: wow symbolic execution is the wave of the future

me:
August 15, 2025 at 11:11 PM
evergreen
August 13, 2025 at 6:55 PM
i owe you an apology Queen Belessunu of Babylonia, i wasn't really familiar with your game
August 13, 2025 at 6:43 PM
and after all, lean is not designed for computing :P
July 18, 2025 at 11:55 PM
very bullish
July 18, 2025 at 11:29 PM
ousterhout solved this problem three decades ago by bravely declaring java a systems programming language
July 15, 2025 at 11:16 PM
afaik the closest you get in agda is either superscripts or individual modifiers like circumflex. anything more would require going beyond clever unicode hacks and doing proper math rendering in the IDE
July 14, 2025 at 11:56 PM
programming languages will never progress until we can \overset little symbols onto our function arrows. no -{α}-> is not good enough
July 14, 2025 at 11:08 PM
helloooooo boston
July 3, 2025 at 8:58 PM
gonna miss you
June 30, 2025 at 11:12 PM
goodbye california
June 30, 2025 at 10:42 PM
I think the issue is theoretically `z` could be a ref and `g` could mutate `z`, and even their shared modality isn't designed to solve this as per the drfcaml paper
June 20, 2025 at 12:26 AM
a causal model helps generate ideas for intervention, as opposed to merely "think harder" (roughly what the blog post proposes). satisficing tends to look like spindly depth-first explorations through a big problem space. LLM-generated code is essentially plopping a person into a deep node
June 19, 2025 at 5:16 AM