RexGoliath

@UrbanCourtyard Very cool. Just a citizen but would be great to see the proposal. Sent a LinkedIn request.

Hi a group of leading local architects and I are working on the "Make Big Plans" for Chicago 2050. Message me if you'd like to get involved, sign on to the submission proposal, etc.

@UrbanCourtyard My hot take that might be modernist is that the Haussmann's of America should have rooftop decks for residents. Never understood why this isn't an obvious automatic sign off by a zoning commission if designed properly.

there's a "trad versus modern" architecture debate raging in the replies to this post. I had to mute it a long time ago because it's so annoying, but feel free to jump in if that would make your Friday better the modernist guy needs some back up bad

@PaulSkallas This seems to validate the trends around snowbirds I’ve been seeing. Upper Wisconsin and Michigan are unbeatable combos with Florida or SC.

People talk about retirees going to Florida. But for lots of midwesterns from Chicago, Minneapolis, and Wisconsin they retire to their lake house in the upper midwest.

@1llegalEngineer Mostly true. Lidl also good. Small regional stores that are nice as well.

I'm sorry Americans but Aldi just mogs every single one of your supermarkets Notable exception being Costco

@wmataGM I don’t know how y’all are doing it, but I have to admit as a Chicagoan, it was very nice living near DC and basically having RFR / Tokyo metro co quality service in terms of frequency and timeliness. Great job so far! (Spread the wisdom to metra/CTA haha)

Thx for recognizing the quality service the Metro team is working hard to deliver. Never going to be perfect but we continue to see safety, security & reliability improvements. On the weekend we had about 1.2m 🚇🚌 trips w/ 🚇 OTP over 90% and bus ~ 77%. Thanks for riding.

@Yuchenj_UW Maybe this is naïve, but it’s good to know that code was produced by an agent. I assume this is at least part of why Anthropic does this.

I noticed something interesting: Claude Code auto-adds itself as a co-author on every git commit. Codex doesn’t. That’s why you see Claude everywhere on GitHub, but not Codex. I wonder why OpenAI is not doing that. Feels like an obvious branding strategy OpenAI is skipping.

@fchollet What’s the solution you think? Exploratory work that somehow allows language creation and discovery? Some like making a new language out of assembly and compiler design without the reliance on gcc memorization?

This is more evidence that current frontier models remain completely reliant on content-level memorization, as opposed to higher-level generalizable knowledge (such as metalearning knowledge, problem-solving strategies...)

@midego1 @cursor_ai @davidgomes Yeah this makes sense, seems like fast is on a reasonable but less impressive part of the curve, guessing they’re still dialing it in. Great IDE but seems like their marketing team likes their ambiguous xy plots like other labs do. Keen to try though.

@cursor_ai @davidgomes But why you make fast (3x more expensive) the default? This feels like a criminal move.

Composer 2 is now available in Cursor.

@joelgrus Only the FTC commissioner, knower of the “Embrace, Extend, Extinguish” model, could stop them. But when the world needed her most, she vanished.

no way this goes through, it would give openai a monopoly on sane python tooling

@imaurer @TheAhmadOsman Can you delete this? I really don’t want zig to somehow be bought out by a lab and have to learn another language, lol.

Wrong, it's actually Zig versus Rust. That's the true epic battle of the ages. Unfortunately I prefer Python and Zig so I'm a man without a country. But really rust is fine too. And Pi made me not think badly of TypeScript anymore. Actually who cares? No one reads the code anymore anyways.

you can see how OpenAI is building around Python + UV while Anthropic is building around TS + Bun

@JoshuaKDominic Is PTC built out on the lines ya’ll are servicing? Cool to have European style private operators, but I’ve heard freight rail line owners are not always keen to share their operations and forecasting data (ex: Amtrak, metra).

He isn't wrong. Freight is very important. But it's time we had both!

@pmarca If anybody shall reprove me, and shall make it apparent unto me, that in any either opinion or action of mine I do err, I will most gladly retract. For it is the truth that I seek after, by which I am sure that never any man was hurt.

Marcus Aurelius, Meditations, X.16: “To stop talking about what the good man is like, and just be one.”

@___4o____ I’ll take the permanent underclass over moron-core any day, this shit’s great.

Marc Andreessen believes in introspection and he doesn’t have ai psychosis. Garry Tan doesn’t actually vibe code 10K LOC a day. It’s all a rhetorical performance to normalize their dystopia. I took the bait.

@UrbanCourtyard @shycollie Cool, paseo’s are sick. Barcelona has the right approach given the climate. I would have thought this would be harder to sell to parents though, but I guess that’s why you have these concept spreads with levels of privacy. Just signed up for the letter, good stuff.

@shycollie @RexGoliath3 We have $ and are building prototypes

A question I hear a lot: “If we legalize courtyard housing, how do we actually build it in the U.S.?” Some thoughts in a🧵

@sporadica Why? It’s fantastic seeing the narcissistic doubling down in full swing, now it’s undeniable why his firm funds con artists intermixed with great companies. There’s no need for introspection, will make cluely and others far funnier to watch.

it’s getting sad how much he can’t stop

@shycollie @UrbanCourtyard Honestly think that’s unfair to the work she’s done on the awareness front. I wouldn’t be urging it if I hadn’t seen her posts, and she seems to work with a lot of cities on making this happen. If you’ve worked startups before you know dreams are one of the primary fuel systems.

@RexGoliath3 @UrbanCourtyard Totally agree, I’ve chided Alicia Courtyard the Romantic Dreamer before; the first step is to actually build one that fills itself with upper middle class young families and successful retail so that people can see the dream, just one, others will follow if it works

@krakek1 @Howaboua @pvergadia Good lord, I guess every article written on ai (but especially Chinese ones) has to be through the lens of “is this just generated slop?”. Agreed on early adopters though, best I’ve found is asking continuous questions and building up system prompts and thorough review processes.

@RexGoliath3 @Howaboua @pvergadia I mean, it is also probably written by AI. Still, the paper is real and I think that early adopters of AI for serious use cases (ironically, Anthropic with CC) are going to feel the pain, if they don’t already.

🤯BREAKING: Alibaba just proved that AI Coding isn't taking your job, it's just writing the legacy code that will keep you employed fixing it for the next decade. 🤣 Passing a coding test once is easy. Maintaining that code for 8 months without it exploding? Apparently, it’s nearly impossible for AI. Alibaba tested 18 AI agents on 100 real codebases over 233-day cycles. They didn't just look for "quick fixes"—they looked for long-term survival. The results were a bloodbath: 75% of models broke previously working code during maintenance. Only Claude Opus 4.5/4.6 maintained a >50% zero-regression rate. Every other model accumulated technical debt that compounded until the codebase collapsed. We’ve been using "snapshot" benchmarks like HumanEval that only ask "Does it work right now?" The new SWE-CI benchmark asks: "Does it still work after 8 months of evolution?" Most AI agents are "Quick-Fix Artists." They write brittle code that passes tests today but becomes a maintenance nightmare tomorrow. They aren't building software; they're building a house of cards. The narrative just got honest: Most models can write code. Almost none can maintain it.

@macjshiggins lol, @FFmpeg I’d watch out. That post reads like slop, but it’s this: loc.gov/item/global-le… with lovely clear clauses like “immunity from liability for unintended consequences or errors in experimental AI activities conducted in noncommercial environments”

What in the world does “first country to give open-source AI developers full legal protection” mean? Is America putting them in camps?

@Howaboua @pvergadia Yeah, reinforces that scalability concerns should be high with more operational codebases, haha. Many are saying that because the mandated internal AWS model was poor, it caused the large outages. Paper might not be repeatable, but its likely that 4.6 or 5.4 have same issues.

I am "just a guy", but... This is all GenAI operating on closed source models (mostly). None of these runs are deterministic. So with all due respect to the titanic work that the researchers are doing, there are way too many moving parts for it to be the be-all-end-all benchmark. We all wish it wasn't like that. But this is the reality, we're all spraying-and-praying a little bit, and throwing our markdown files and custom CLIs & system at the wall to see if the stick. Fun, but this all changes too often and papers like these don't really get updated, do they?...