Old Fucking Punk @lwriemen@librem.one

@fasterandworse What comes to my mind when I think of Firefox?

I've already made an illustration about that.

I keep seeing people describe the character.ai bankruptcy as the first AI bubble domino, so I’m glad to see some reporting on what’s *actually* going on here: https://arstechnica.com/tech-policy/2026/10/disney-found-a-way-to-buy-character-ai-without-buying-its-lawsuits/

This is a way for Disney to buy the company without really buying the company, so they can avoid antitrust. It also gets them out of inheriting the active lawsuits. Character.ai has been through this once already - what’s left of the company is kind of a shell after Google hired key employees & licensed their tech.

Marc Andressen portrayal in Bloomberg's Business Week magazine shows him to be not just a nasty character but also a dishonest one. https://www.bloomberg.com/news/features/2026-10-15/how-ai-crypto-and-dei-drove-marc-andreessen-to-maga?accessToken=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJzb3VyY2UiOiJTdWJzY3JpYmVyR2lmdGVkQXJ0aWNsZSIsImlhdCI6MTc5MjQzNTg1MiwiZXhwIjoxNzkzMDQwNjUyLCJhcnRpY2xlSWQiOiJUTVhZRElWVFRDWlAwMCIsImJjb25uZWN0SWQiOiJBQjUxRjdCOUYxODk0NjBCOEE2RDU3NUM5QUQzOUUzQyJ9.gu59CBo0KsJrx_HiVBpLojho5OoOzWbDNq8934JPxk8

Excellent reporting...

Who woulda' thought Florida politics could be so swampy?

Gov. Ron DeSantis allegedly diverted $10m in taxpayer money to his wife's charity. It's been headline news for months.

Now the Florida GOP is suing TV stations, claiming it's a crime to report on records about the scandal.

https://floridatrident.org/florida-gop-lawsuit-hope-florida-ad/

This one talks about understanding state is key to making effective decisions and automation abstracts that state away

> The other important aspect of cognitive skills in on-line decision making is that decisions are made within the context of the operator's knowledge of the current state of the process.

Basically, despite it being from 1983 and that a lot has been added to human factors research since then, it's short and well worth at least a skim

Stop trying to upgrade disabled people's body when the problem is your design!
@WeirdWriter argues that the tech industry needs to stop trying to sell useless expensive gadgets no disabled person wants, or can afford, and start fixing actual accessibility barriers, like bad code, and poorly designed infrastructures.
https://sightlessscribbles.com/posts/disability-dongle/

#Accessibility #DisabilityDongle #FixYourHTML

huh, somehow I missed this while I was distracted with netbsd (thank you @techokami)

https://awful.systems/post/10050264/12946932

> Thanks for linking my toot, here’s another update since then: the same person that found the race condition also found that there was a performance regression on non-synthetic benchmarks. It went super fast on his little test but is complete dogshit when used in the real world! https://lwn.net/ml/all/atI0edrmeiRuRtNQ@thinkstation/

the remarkable kernel patch that remarkably convinced linus torvalds to go all-in on LLMs, remarkably running into all of the reasons why this was a horrible fucking idea

anyone with an ounce of familiarity with how LLMs function could’ve seen this coming, but not kernel boy. why wouldn’t you slop up a feature and all of its tests and its documentation and then just go on vacation so someone else can try to clean up your mess?

How the world changed for me with GenAI code generation: from bad to mediocre code piling up slowly, to bad to, at best, mediocre code piling up 10x faster while the team's mental model of how the system they build and own works dissolves.

“Pangram is not an oracle • Buttondown”

https://buttondown.com/maiht3k/archive/pangram-is-not-an-oracle/

> But as always, we have to be wary of any proposed technical solutions and evaluate them with care (including asking whether a technical solution is called for at all). This is again one of those cases where machine learning can easily create something that looks like a solution, but isn't, because it's easy to make a machine that gives “scores” for how likely some text is to be synthetic/authentic.

An exclusive design by HappyToast in tribute to another of the greatest comic creations of all time, The Young Ones, and their classic appearance on University Challenge. Get yours with FREE UK delivery here >>> https://go.newsthump.com/7xYrrT

“The Many Ways Trump Is Kneecapping White-Collar Crime Enforcement – Mother Jones”

https://www.motherjones.com/politics/2026/10/white-collar-crime-enforcement-trump-white-house-sec-doj-cftc/

> Longer term, though, drops in corporate crime enforcement and regulation have often preceded some of the country’s biggest financial crises

The additional context here is that the most serious crimes and fraud in a financial bubble come AFTER the first domino has fallen. Bubbles are full of smaller frauds that everybody excuses because there's always more money later. But when a bubble starts to weaken and the bill comes due, the pressure to ramp up the fraud to perpetuate the bubble becomes impossible to resist

So, the most serious crimes related to the “AI” bubble are likely to take place over the next 12-18 months (at least)

Young People Hate AI CEOs So Passionately That It’s Almost Hard to Believe https://futurism.com/artificial-intelligence/young-people-ai-ceos-executives-poll

I am so glad kids aren’t stupid to buy this AI hype. They understand that future is not bright where 3 or 4 CEOs horde all wealth. Yeah so it is not hard to believe people hate these CEOs. It’s hard to believe anyone actually likes them.

RE: https://mastodon.social/@Zygos/117461869086312391

It strikes me that there’s a real Catch-22 here for AI vendors:

They want to argue that training an LLM on other people’s copyrighted work without permission is OK because the LLM and its output are transformative use.

But if it’s transformative, then the people doing the transforming must be responsible for what they created. No more “we’re just indexing other people’s work!” If it’s misleading or defamatory, that’s their responsibility now.

Can’t have it both ways.

I'm leaving WordPress platform very, very soon.

Mullenweg must give clear, transparent answers on what he wants, because we are thousand of people with disability, left behind.
An Open Letter to Matt Mullenweg on Accessibility in WordPress – Blind Access Journal https://blindaccessjournal.com/2026/10/an-open-letter-to-matt-mullenweg-on-accessibility-in-wordpress/

#wordpress #accessibility #a11y #blind

One of the patterns I’ve seen with people using Claude for programming is:

A programmer in the past looked at a tradeoff space and decided to make a common operation fast and an uncommon operation slow.

Claude lacks proper context of the use of the software but spots a pattern that looks like things it’s training set has flagged as slow and so points to the slow path as a bottleneck.

The user asks for a test case and Claude provides a simple microbenchmark that triggers the slow path.

Now that it’s confirmed as slow, the user asks a ‘agentic’ harness to explore the design space and optimise for the benchmark. The swarm of Claude instances makes plausible changes to the code and reports back with a modified version that passes the test suite and improves the microbenchmark.

The test suite doesn’t exhaustively cover all possible race conditions and so the new version may include subtle concurrency bugs. It also probably makes the common case slower and so is a regression overall.

I don’t know if other LLMs encourage this flow, but I’ve seen it several times from people using Claude.

I suspect Linus’ recent embarrassing interaction with his kernel’s virtual memory code may have been an example of this pattern. It’s very easy to do with an LLM because each step appears to be a demonstrable improvement. It comes with an objective benchmark, it passes all of the tests. Unfortunately, software often depends on a load of extrinsic knowledge that the LLM is trying to infer from a load of likely misleading data points.

(It's the delta that worries me in the code I see—the change in quality, that is. Engineers with a quarter century of experience churning out code by the bucketful that's quite a bit worse than what they used to write. So either they aren't reviewing their own code like they used to or their skills have degraded.)

bonus: since R = U/I, this means that I (current) is always equal to 2 amperes.

this textbook might kill someone

Every time I see a vibe coder say something along the lines of "if you haven't used these tools over the past five minutes, you don't know what they're capable of, because everything's different now!" I get a little bit perplexed.

You do know that we can see the output without using it ourself, right? Some of us are still reviewing pull requests and reading through code repositories, y’know?

Wednesday October 21, 2026, Joe’s Garage , Amsterdam

“Proxycafe – Rise Against Big Tech.

Door opens at 19:00. Presentation, discussion from 20:00

We are living inside a sci-fi book. Which narrative of resistance are we going to write?

In this talk we want to open a discussion, talk about the resistance to big tech.

Which kind of alternatives are we working on?”

https://joesgarage.nl/archives/19168

“Internet of/with Squirrels” :)

https://wiki.techinc.nl/Hackers_tribes#Consider_the_Squirrels.21