ktrst (@ktrst) header
ktrst (@ktrst) profile picture

誰のせい?それはあれだ夏のせい/附高(下馬の47D)/とんぺー/cmpss/物性物理/磁石/世田谷→仙台→Orsay→つくば→Brugg→文京/起業しちゃった側

右耳と左耳の間
Joined May 2009
Photo by ktrst (@ktrst)Photo by ktrst (@ktrst)Photo by ktrst (@ktrst)Photo by ktrst (@ktrst)Photo by ktrst (@ktrst)Photo by ktrst (@ktrst)Photo by ktrst (@ktrst)Photo by ktrst (@ktrst)Photo by ktrst (@ktrst)Photo by ktrst (@ktrst)
誰かと思ったら『非線形ダイナミクスとカオス』のスティーヴン・ストロガッツだ。
コーネル大の数学者が時にマジ泣きしながらながら語る、AIの台頭と数学者としての人生の目的の喪失の話。本人の実直さが伝わってきて、涙なしには聞けない。他人事ではない。 podcasts.apple.com/jp/podcas…
1
107
12
532
53,246
ひええ > 「OpenAI の論文は正しい.もっと短く,人間にわかりやすくできるので今書いている.このアイディアでほかの関連問題も解ける.我々は今まで30年間間違った方向を向いていた.」 ms.u-tokyo.ac.jp/~yasuyuki/n…
10
448
47
2,774
606,755
AI has sped up research so much. One can ask and answer questions on the same day on arXiv :P
9
36
8
397
40,059
ktrst retweeted
A sign is a sign of bad design
3
3
40
5,038
New Claude code update is crazy
104
957
85
15,399
1,171,770
SPReADニュース 文科省謹製のpythonコードで申請書の書式セルフチェックができるぞ!
72
11
329
54,135
板橋区の小学校、教室の増やし方すげーな
384
2,206
678
61,319
4,506,387
ktrst retweeted
when poor claude gets to this point u just know it's utterly and hopelessly lost
39
71
4
3,695
157,374
今年の夏も暑くなりそうですので、我が国のサイバー防衛の現状を知って背筋を冷やしていただければ幸いです
71
1
299
25,125
自衛官がサイバー訓練用に模擬ウイルスをつくったところ、不正指令電磁記録として摘発されてしまった事例について講演する予定です
⚡Hack Fes. 2026 🚀 📢プログラム第一弾 発表! & 🎫チケット販売開始 !! 🌍テーマ「ハッカーとサイバー防衛」 🔥『イラン戦争に学ぶサイバー空間での攻防の現状と今後』 🛰️『宇宙システムの攻防:事例分析とシミュレーションで学ぶ宇宙サイバーセキュリティ』 🤖『AIエクスプロイト時代のサイバー防衛:現役ハッカーが語る、AIによる0-day発見の最前線と防衛側のSurvival of the Fittest』 🔑『量子技術による暗号への脅威と量子耐性をもつ暗号システムの可能性』 ❌『フェイクニュースの拡散現象とその対抗策』 ⚖️『「サイバー防衛」に内在するリスク:模擬検体の作成をめぐる刑事弁護の実例』 📕『徳丸本アップデート2026』 🐛『AI時代のバグハンティング入門 〜AIエージェントと歩むバグハンターへの道〜』 今後もアップデート続々! 乞うご期待!! #hackfes
9
1,331
40
4,540
538,799
2026
501
15,029
716
234,975
9,930,283
I don't work on reliability & scaling at GitHub, but the people who do aren't bad at their jobs. They're dealing with unprecedented scale from agents. It's easy to shit on GitHub from the outside if you're not in charge of 30X-ing capacity within a few months. Have some grace.
Being the foundation for millions of developers means our bar must be higher for availability, reliability, and security. I’m sorry it’s been a rocky stretch at GitHub. We know we need to do better. Today we published an update on two recent incidents: one on April 23 involving merge queue behavior, and one on April 27 affecting pull requests, issues, projects, and search-backed experiences. We’re taking this seriously. We’re listening, and you have my commitment that we’ll communicate more frequently about the work underway to improve reliability and scale GitHub for what comes next. github.blog/news-insights/co…
84
102
40
1,182
194,813
ktrst retweeted
my linkedin feed
19
297
26
3,158
184,972
anytime i finish a blog post, i feed it to an LLM asking it to produce 20-40 HN or Reddit comments. immensely effective. stole that idea from @mitsuhiko
- Drafted a blog post - Used an LLM to meticulously improve the argument over 4 hours. - Wow, feeling great, it’s so convincing! - Fun idea let’s ask it to argue the opposite. - LLM demolishes the entire argument and convinces me that the opposite is in fact true. - lol The LLMs may elicit an opinion when asked but are extremely competent in arguing almost any direction. This is actually super useful as a tool for forming your own opinions, just make sure to ask different directions and be careful with the sycophancy.
57
112
19
2,901
288,588
BREAKING: Alibaba tested 18 AI coding agents on 100 real codebases, spanning 233 days each. they failed spectacularly. turns out passing tests once is easy. maintaining code for 8 months without breaking everything is where AI completely collapses. SWE-CI is the first benchmark that measures long-term code maintenance instead of one-shot bug fixes. each task tracks 71 consecutive commits of real evolution. 75% of models break previously working code during maintenance. only Claude Opus 4.5 and 4.6 stay above 50% zero-regression rate. every other model accumulates technical debt that compounds with every single iteration. here's the brutal part: - HumanEval and SWE-bench measure "does it work right now" - SWE-CI measures "does it still work after 8 months of changes" agents optimized for snapshot testing write brittle code that passes tests today but becomes completely unmaintainable tomorrow. they built EvoScore to weight later iterations heavier than early ones. agents that sacrifice code quality for quick wins get punished when the consequences compound. the AI coding narrative just got more honest. most models can write code. almost none can maintain it.
80
299
70
1,489
526,811
中身をろくに確認せずに承認のハンコを押す管理職の気持ちがClaude Codeを使う今ならわかる
3
256
16
1,827
145,264
ktrst retweeted
Prof. Donald Knuth opened his new paper with "Shock! Shock!" Claude Opus 4.6 had just solved an open problem he'd been working on for weeks — a graph decomposition conjecture from The Art of Computer Programming. He named the paper "Claude's Cycles." 31 explorations. ~1 hour. Knuth read the output, wrote the formal proof, and closed with: "It seems I'll have to revise my opinions about generative AI one of these days." The man who wrote the bible of computer science just said that. In a paper named after an AI. Paper: cs.stanford.edu/~knuth/paper…
143
1,836
311
8,977
1,419,773
Need another season of Silicon Valley so badly.
With the takeoff of OpenClaw and MoltBook, a new agent-driven economy is taking shape. On the @LightconePod, we took a look at the explosive growth of AI dev tools and whether the time has come for builders to make something agents want. 00:00 - Intro 02:12 - No human involvement is changing the experience 04:55 - Does YC need to change its motto? 07:48 - Email tools and agent infrastructure 09:36 - Agent-driven documentation 13:00 - Swarm intelligence 15:36 - Content generation and dead Internet theory 18:12 - Growth, rules, and founder insights
79
231
10
5,060
453,091
いま本当に40年以上やってる中でも最高にプログラミング楽しいんだけど、疑問に思ったことになんでも答えてくれる教えたがりの先輩と、簡単だけど面倒くさい実装を厭わずやってくれる後輩とが両方揃ってるんだからそりゃ楽しいに決まってるよな。ひとりだけど、四畳半スタートアップみたいな雰囲気ある
8
450
92
3,007
515,992
Geminiに自分が物理学科ということをインプットしてるせいで何に関しても「物理的な側面から」説明してくるのうざいな 部屋のレイアウト聞いただけで拡散がどうのとかエントロピーがどうのとかうるせえよ
64
1,282
469
30,724
1,292,295