Simon @spoltier@qoto.org

#llms and labor:

I think the impact of LLMs on software development labor is, beyond CEO-hype memetics and whatever end-spiral of late stage capitalism we are currently in, real. Machines can to some extent replace a huge part of software engineering labor. It's not just a bubble of hype that will burst (I'm more skeptical on the impact of LLMs in other types of labor).

Not only do I think the impact is real, but I think it is justified. Computer software jobs have been about automating and organizing human labor for pretty much the entirety of its existence. The complexity of the job means that tech jobs had a kind of privileged labor bourgeoisie status, with ridiculous salaries compared to the people whose labor got organized, analyzed and automated. (I've tried to navigate the field by only working jobs where I knew the people my software impacted, meaning I worked either building manufacturing tools for individuals / small labs, artists, and small family businesses, with a stint in working for personal development coaching company. This means that after more than 2 decades in the field, I still make less than an intern at facebook, so be it.)

Am I glad that software developers now end up in the same category as the rest of labor? Yes and no. Noone deserves to have their livelihood threatened, to be cudgeled with technology, to be replaced by a machine.

So why am I talking about LLMs all the time? Because they allow us to finally break software free of the large structures needed to create mass commoditized software targeting the fiction of an average user, to combat the use of "free" (as in beer) software to enshrine surveillance capitalism and mass advertising across society.

They allow us to cut ties with the massive tech conglomerate that even the smallest hairstyling business relies upon to do things like intake, appointment and payment management. Not only is it extracting ever more money of the smallest workers, but it also creates absolutely alienating technological experiences. Ever tried configuring your appointment system in square? Pay $40/month more just to have an even more infuriating intake form builder? For what? Get bombarded with daily ads about the latest templates?

#llms #llm #genai

So I always wondered if the name of "lithium" implies it has some kind of historical connection to "lithuania". But I looked it up and it doesn't. The two aren't even etymologically related. "Lithuania" comes from Latin "litus" for "shore" (think "littoral") and "Lithium" is from greek "lithion" for "stone" ("monolith").

EDIT: Wiktionary disagrees on Litus and claims Lithuania is either from proto-Slavic and cognate with "Latvia"; or named after a river. Still not Greek. https://en.wiktionary.org/wiki/Lithuania

Gestern Abend wurde Cyber Safari lanciert, eine Lern-App, die Zürich zum Lernraum für digitale Kompetenzen macht. Jugendliche erkunden in Teams das Stadtzentrum und lösen Challenges zu Desinformation, Cybermobbing, Datenschutz und digitaler Balance. Die App zeigt, wie digitale Bildung aussehen kann, wenn Forschung, Gestaltung und Stadtraum zusammenspielen.

Kostenlos und jederzeit möglich, ob mit der Familie, Freund:innen oder als Schulklasse: https://cyber-safari.ch/

@algorithmwatch_ch

I have been using email for 40 years. It used to work.

As an (independent) academic researcher, I need to contact new people, primarily in universities, to ask questions.

I refuse to use Google, Microsoft or the other American IT giants.

But they are increasingly preventing refuseniks from sending email at all.

I know what RFC, DNS, MX, SPF and DMARC mean. My email goes through small British companies with intelligent, friendly and helpful staff.

mxtoolbox.com says that I must have DMARC to send email to M$. So I set it up. I now get a dozen copies of the same report from G or M$ for each email that I send out.

They show that my email gets to G and M$ sites, but then it is marked as spam.

The stupid senior management of numerous universities has surrendered their staff email to M$.

Web searches and AIs preach about spam. I don't send spam - I want to contact my colleagues.

Rumour has it that previously unknown senders are treated with suspicion and their emails are sent to spam. In other words, it is impossible to **initiate** communication with someone.

Let's be blunt about this. They are a mafia that is enforcing an **oligopoly**. It's got nothing to do with reducing spam --- I have no doubt that they let through emails from "trusted partners", ie companies that bribe them enough to send their spam.

The result of this is that it will only be possible to send emails by paying M$ to do it, and then it will only be allowed to express "approved" opinions.

What can we do about this?

At the very least, those of you with senior positions in universities can tell your management to revert to competent standards-based email systems hosted on Linux systems.

A good world model, a good theory of crocodilian minds, and a deliberate planning ability on the part of this warthog.

Just FYI.

Alright, it's official! 💰

@matthew_d_green and I bet on what will break first, ML-KEM-768 or X25519. The loser donates to a 501(c)(3) picked by the winner.

If you have an opinion on quantum computers or lattices, you can join with a side bet. Just submit a PR!

https://github.com/FiloSottile/ecc-vs-lattices-long-bet

I wrote something about how I had been the target of an anti-gay thing on a Singaporean subreddit. In September last year a bunch of people (led by a Singaporean school teacher) told everyone to search my social media posts to report anything anti-Republican to ICE

My piece is less about the specifics about that incident but more about how I feel, as a person who has been out online for the last 22 years, that the climate has changed (the homophobia never really went away but was just less public, briefly)

And also about increasing homophobia, transphobia around the world.

Not sure where I’ll publish it. Will welcome ideas for any online or other publication.

a specter is haunting European bathrooms

not sure if religion or art if one turns up in every city, I guess we’ll know it was religion

RE: https://bsky.app/profile/did:plc:kuqaskeln2elxcisr2dthbqm/post/3midn3ddet22q

Vinod Khosla is pushing to end the policy of taxing investment income at a lower rate (capital gains) than salaries then using that money to end taxes on people earning under $100K.

He argues this is needed to address job displacement due to AI.

I think this is the bare minimum and more policy shifts will be needed.

https://www.implicator.ai/openai-backer-khosla-pushes-to-end-income-tax-for-125-million-americans/

I wrote some lines about mitigating vibe-coding risks by adopting a development model inspired by old-school computer breakin folks:

https://addxorrol.blogspot.com/2026/03/slightly-safer-vibecoding-by-adopting.html

Here's an example of the sort of thing I'm talking about (only hearsay, but it serves to illustrate the point).

This is just classic "ML at scale discovering weird causalities" but just think what the system has to work with here.

It can operate in the full high-dimensional space of video and audio and generate any high or low frequency artifact that our brains can perceive, and well as high-level semantic features: sexy people, strange compelling mini stories, weird, hypnotic triggers.

Analysis of network syntax: the skull emoji signifies amusement. Joy is translated into a symbol of cranial decay. This is an efficient data compression strategy. The user is indicating the humor was lethal.

Jensen Huang stated "If your $500K engineer isn’t burning at least $250K in tokens, something is wrong."

Unless token price increases a lot, I can literally work 18 hours a day X 7 days a week and I cannot burn $250K in tokens.

Thoughts on OpenAI acquiring Astral and uv/ruff/ty https://simonwillison.net/2026/Mar/19/openai-acquiring-astral/

new saas idea : pay to see the horizon in the current simulation of reality !

🔗 https://rmendes.net/replies/2026/03/16/0319e

Another interesting development is Intel’s Clearwater Forest server CPU, which was recently introduced.

Intel designed this chip as an all-E-core Xeon, based on the assumption that hyperscalers would want extremely power-efficient CPUs for large-scale data-center workloads. For a while, that strategy looked questionable and product was considered as DOA.

But with the rise of agentic AI, efficient CPU capacity in the data center is becoming important again. Intel just lucked out.

RE: https://www.threads.com/@sung.kim.mw/post/DV8ZyE5kUiU

New chapter for Agentic Engineering Patterns: I tried to distill key details of how coding agents work under the hood that are most useful to understand in order to use them effectively https://simonwillison.net/guides/agentic-engineering-patterns/how-coding-agents-work/

how prevalent are agent organizations *today*? pretty sure it’s non-zero. also i’m pretty sure they exist outside just coding. but i don’t think people talk much publicly about them