Kenta Iwasaki (@lithdew) header
Kenta Iwasaki (@lithdew) profile picture
Joined April 2020
Photo by Kenta Iwasaki (@lithdew)Photo by Kenta Iwasaki (@lithdew)Photo by Kenta Iwasaki (@lithdew)Photo by Kenta Iwasaki (@lithdew)Photo by Kenta Iwasaki (@lithdew)Photo by Kenta Iwasaki (@lithdew)Photo by Kenta Iwasaki (@lithdew)Photo by Kenta Iwasaki (@lithdew)Photo by Kenta Iwasaki (@lithdew)Photo by Kenta Iwasaki (@lithdew)
Kenta Iwasaki retweeted
the dumbest smart people you know are debating temperature and rngs and skewing probability distributions over semantically similar "words" because they heard about word2vec once and believe "word choice" has no functional purpose beyond a canonical dictionary definition and a cosine similarity in latent space "isn't hiding a signature in text the same as hiding it in rgb in a pixelated image?", cries the clever-by-half hot-taker
"watermarking with no quality loss" cries the near illiterate tech employee who hasn't read a single literary work for fun as a grown adult aside from the occasional blog post
22
20
4
595
51,966
Kenta Iwasaki retweeted
hello there the jacobian conjecture is false thanx to my close friend akhil for asking about it and my other close friend fable for working during the world cup final ((1+xy)^3 z + y^2 (1+xy) (4+3xy), y + 3 x (1+xy)^2 z + 3 x y^2 (4+3xy), 2 x - 3 x^2 y - x^3 z): \C^3\to \C^3, has jacobian determinant -2, and sends (0, 0, -1/4), (1, -3/2, 13/2), and (-1, 3/2, 13/2) to (-1/4, 0, 0)
1,696
5,373
3,152
43,757
40,107,589
Kenta Iwasaki retweeted
Legends don't guess. They measure. Giannis Antetokounmpo is our first brand ambassador. 6.30.26
43
58
37
619
661,813
Kenta Iwasaki retweeted
I'm usually not one to write thought pieces without much technical depth. But here we go. Slow the fuck down. mariozechner.at/posts/2026-0…
151
496
176
2,868
629,896
Kenta Iwasaki retweeted
Postgres as data platform: make heterogeneity a first-class engine property—SLA-aware planning, pluggable execution, and machine-readable feedback (visibility, degradations, sources). Otherwise it's just storage+glue. noodata.substack.com/p/postg… #PostgreSQL
3
7
1
63
25,551
Kenta Iwasaki retweeted
🥈 ($8,000) Thiel Fellow @lithdew and gigabrain engineer @codycode_0209 launched HealthDB. Their AI agent on @Aptos helps users better manage & enrich their health data via a weighted Monte Carlo tree search 🌳
2
3
18
1,188
ChatGPT exploits a buffer overflow 😳
66
934
247
5,702
Kenta Iwasaki retweeted
Soooo I think the compressor works! Had to fine-tune the options (-2dB threshold, 20:1 ratio, 2ms attack, 1s release), but now I can play a whole bunch of sine waves without my ears bleeding.
1
2
17
3 line bugfix 🤦‍♂️ thanks to @lithdew for pointing out syscall overhead the other day github.com/oven-sh/bun/commi…
1
1
2
30
The craziest thing about formal methods is that after you've used a model checker enough, you suddenly have an intuition for weird concurrency bugs. Concurrency is hard because we don't normally get immediate feedback on our mistakes. FM gives you that, so your brain picks it up.
3
16
165
Kenta Iwasaki retweeted
Ever noticed syscalls are more expensive on threaded applications? Most syscalls need to pin/unpin the file descriptor for a syscall, and if the file table is shared, that's a nasty double atomic. Some io_uring help for that, with numbers: lore.kernel.org/io-uring/d90…
2
30
191
Kenta Iwasaki retweeted
IMO a library should be just a collection of functions that do a certain operation. It should **not** provide some abstract interface that other libraries can plug into. Creating mediums of communication between libraries should be reserved for the programming language itself.
1
6
61
Kenta Iwasaki retweeted
Culturally transmitted skills and values danluu.com/culture/
8
28
1
178
Kenta Iwasaki retweeted
In programming I don't think there is such a thing as a zero cost abstraction. You either pay it at runtime, compile time or developer time.
1
6
1
72
Kenta Iwasaki retweeted
Fuzzing Zig Code Using AFL++ ryanliptak.com/blog/fuzzing-…
1
9
Kenta Iwasaki retweeted
Tips i've learned from writing thread pools (this is easier than writing a blog): - Reduce the amount of state owned & updated by one thread to the minimum - Make all queues lock-free and have a centralized blocking scheme instead of locks on some queues (1/n) I'll elaborate
1
20
106