Richard Marmorstein

@evilsocket Is there something specific you wish you could do with Claude Code but can't because of the memory footprint?

A Claude Code instance in a terminal, consumes more RAM and CPU than a freaking web browser. How is this possible? Under the hood, Claude is a nodejs React app using some monstrosity called Ink for "TUI". How does this make any sense? Make software great (& optimized) again!

A driver behind me in the turn lane honked because I wasn't getting into the intersection aggressively enough for his liking. "Beep beep!" says my one-year-old, delightedly. Joke's on him, lol.

Ironically, hypermedia (HATEOAS) has accidentally become a plausible API design scheme again. LLMs will robustly follow API links just like its designers hoped.

bikesheds for the mind

@kayintveen @dhh It can just build the feature and then you can see it is a waste of time and delete it.

@dhh 18 years in and the 10x myth always annoyed me. the real difference was never raw coding speed - it was knowing what NOT to build AI makes everyone type faster but still cant tell you which feature is a waste of time

Hilarious how we used to earnestly discuss whether the 10x programmer was real.

@thdxr @grok what would the framework in courses.cit.cornell.edu/econ352jpw/rea… have to say about this

why do most businesses stay so narrow once you have a machine going and a team in place why limit yourself why isn't spotify building an ai coding agent before you say focus remember how aws came to exist

Be kind, for everyone you meet is solving an incredibly complex optimization problem.

@DGannonTN Star Sith? Then they could rip off Star Wars too!

Okay, hot take: I've never liked DS9. It was a rip-off of B5 that merely managed to go to air first. It was preachy, directionless, and *boring.* They set the show on a space station and stayed on the station for *years!* "Hey, gang! Let's rest on our duffs and see what comes our way this week!" That's not "Star Trek." It's "Star Sit." Hm. Perhaps that could have used an H...

Finally cracked

A few random notes from claude coding quite a bit last few weeks. Coding workflow. Given the latest lift in LLM coding capability, like many others I rapidly went from about 80% manual+autocomplete coding and 20% agents in November to 80% agent coding and 20% edits+touchups in December. i.e. I really am mostly programming in English now, a bit sheepishly telling the LLM what code to write... in words. It hurts the ego a bit but the power to operate over software in large "code actions" is just too net useful, especially once you adapt to it, configure it, learn to use it, and wrap your head around what it can and cannot do. This is easily the biggest change to my basic coding workflow in ~2 decades of programming and it happened over the course of a few weeks. I'd expect something similar to be happening to well into double digit percent of engineers out there, while the awareness of it in the general population feels well into low single digit percent. IDEs/agent swarms/fallability. Both the "no need for IDE anymore" hype and the "agent swarm" hype is imo too much for right now. The models definitely still make mistakes and if you have any code you actually care about I would watch them like a hawk, in a nice large IDE on the side. The mistakes have changed a lot - they are not simple syntax errors anymore, they are subtle conceptual errors that a slightly sloppy, hasty junior dev might do. The most common category is that the models make wrong assumptions on your behalf and just run along with them without checking. They also don't manage their confusion, they don't seek clarifications, they don't surface inconsistencies, they don't present tradeoffs, they don't push back when they should, and they are still a little too sycophantic. Things get better in plan mode, but there is some need for a lightweight inline plan mode. They also really like to overcomplicate code and APIs, they bloat abstractions, they don't clean up dead code after themselves, etc. They will implement an inefficient, bloated, brittle construction over 1000 lines of code and it's up to you to be like "umm couldn't you just do this instead?" and they will be like "of course!" and immediately cut it down to 100 lines. They still sometimes change/remove comments and code they don't like or don't sufficiently understand as side effects, even if it is orthogonal to the task at hand. All of this happens despite a few simple attempts to fix it via instructions in CLAUDE . md. Despite all these issues, it is still a net huge improvement and it's very difficult to imagine going back to manual coding. TLDR everyone has their developing flow, my current is a small few CC sessions on the left in ghostty windows/tabs and an IDE on the right for viewing the code + manual edits. Tenacity. It's so interesting to watch an agent relentlessly work at something. They never get tired, they never get demoralized, they just keep going and trying things where a person would have given up long ago to fight another day. It's a "feel the AGI" moment to watch it struggle with something for a long time just to come out victorious 30 minutes later. You realize that stamina is a core bottleneck to work and that with LLMs in hand it has been dramatically increased. Speedups. It's not clear how to measure the "speedup" of LLM assistance. Certainly I feel net way faster at what I was going to do, but the main effect is that I do a lot more than I was going to do because 1) I can code up all kinds of things that just wouldn't have been worth coding before and 2) I can approach code that I couldn't work on before because of knowledge/skill issue. So certainly it's speedup, but it's possibly a lot more an expansion. Leverage. LLMs are exceptionally good at looping until they meet specific goals and this is where most of the "feel the AGI" magic is to be found. Don't tell it what to do, give it success criteria and watch it go. Get it to write tests first and then pass them. Put it in the loop with a browser MCP. Write the naive algorithm that is very likely correct first, then ask it to optimize it while preserving correctness. Change your approach from imperative to declarative to get the agents looping longer and gain leverage. Fun. I didn't anticipate that with agents programming feels *more* fun because a lot of the fill in the blanks drudgery is removed and what remains is the creative part. I also feel less blocked/stuck (which is not fun) and I experience a lot more courage because there's almost always a way to work hand in hand with it to make some positive progress. I have seen the opposite sentiment from other people too; LLM coding will split up engineers based on those who primarily liked coding and those who primarily liked building. Atrophy. I've already noticed that I am slowly starting to atrophy my ability to write code manually. Generation (writing code) and discrimination (reading code) are different capabilities in the brain. Largely due to all the little mostly syntactic details involved in programming, you can review code just fine even if you struggle to write it. Slopacolypse. I am bracing for 2026 as the year of the slopacolypse across all of github, substack, arxiv, X/instagram, and generally all digital media. We're also going to see a lot more AI hype productivity theater (is that even possible?), on the side of actual, real improvements. Questions. A few of the questions on my mind: - What happens to the "10X engineer" - the ratio of productivity between the mean and the max engineer? It's quite possible that this grows *a lot*. - Armed with LLMs, do generalists increasingly outperform specialists? LLMs are a lot better at fill in the blanks (the micro) than grand strategy (the macro). - What does LLM coding feel like in the future? Is it like playing StarCraft? Playing Factorio? Playing music? - How much of society is bottlenecked by digital knowledge work? TLDR Where does this leave us? LLM agent capabilities (Claude & Codex especially) have crossed some kind of threshold of coherence around December 2025 and caused a phase shift in software engineering and closely related. The intelligence part suddenly feels quite a bit ahead of all the rest of it - integrations (tools, knowledge), the necessity for new organizational workflows, processes, diffusion more generally. 2026 is going to be a high energy year as the industry metabolizes the new capability.

@ptrschmdtnlsn I think the point is the parable is that preference utilitarianism is absurd or even repugnant. It is not morally neutral.

Every time this comes up, a bunch of people miss the point. Each one clearly values watching someone eat shit at >$100, and disvalues eating shit at <$100. This is two people with (utterly bizarre) preferences, getting gains from trade that they wouldn't have gotten were it not for dollars as an intermediary. They both *want* to trade eating shit for watching someone eat shit, but needed dollars to coordinate that positive-sum trade across time! Had they come across the piles simultaneously, they could have negotiated eating simultaneously, but they can't. This is *the* huge advantage of money, as a universal IOU; it enables "virtual trades" between parties separated in time and space.

Why this took so long We started with almost no tests to verify rendering. It was humbling: fix something, break something else. What unblocked us was property-based testing. We wrote tests that rendered components through both the old & new systems and diffed them.

@devongovett @inimino Mathematicians use English mixed with equations. Have you read a math paper?

@inimino but then... why? just write code. English is so much more verbose for the same information if you're being exact. There's a reason mathematicians use equations instead of English.

I don't understand takes like this. Every time I try AI, it generates code that doesn't work, makes up fake methods, uses outdated libraries, etc. It's certainly useful as a better google/stackoverflow, but nowhere even close to replacing humans.

@M_P_Hazell We did!

Well, I'm sure lots of people got excellent homilies on the 2nd reading this Sunday at the Novus Ordo – exactly the sort of reading that encourages preaching on St Paul! 😅

@ezyang here's an example of polish: "Cześć! Miłego dnia i dużo uśmiechu!"

Here's an example of polish. In PT2 we have a graph partitioner that uses min-cut to find an optimal set of values to save for backwards. Sometimes we generate illegal constraints which result in an "infinite flow" NetworkX error. I finally got fed up with this...

@zetalyrae User's choice. It's the same principle as a "memory palace", putting things you wish to remember in physical locations doesn't encode any information about what you wish to memorize but it allows you to use your brain's powerful spatial reasoning to aid your memory

My main problem with the whole “RTS interface for agents” meme cluster is what information is the spatial dimension meant to encode?

@hunterhammonds Ive used it for one off reorganizations and summarizations. Now I mostly use filesystem MCP that I can trigger from Claude mobile.

Who is using Obsidian + Claude Code? I want to talk to you.

@mMoriLiu @nateberkopec twitchard.github.io

@twitchard @nateberkopec is there a link to this

one thing that users really love about UI is when it looks just very slightly different every time they use it

@nateberkopec You kid, but it is for this purpose I designed my personal website to apply a gradually increasing "wobble" parameter to all the elements.