Using gravitational microlensing to find exoplanets is amazing. I had no idea we had technology sensitive enough for that. That should unlock way more planets, because they won’t have to be aligned so precisely in their orbits relative to us.

FOSS has had a funding problem forever, and AI has made it exponentially worse by increasing the review and maintenance burden on volunteer maintainers and the infrastructure burden with scraping and agentic build workflows. The projects that are surviving best are the ones that have a funding source, which is typically businesses that financially benefit from the software (there are of course exceptions).

When the AI bubble pops, companies are going to be pulling back on all spending, so I expect donations to FOSS projects/foundations to get tight. The ones that will still be spending are the ones that will still have money to spend, and I don’t expect diverse views on AI to be distributed equally across those survivors.

You joke, but they have built experimental Debian builds on at least one non-Linux kernel. I don’t know how far an AI split in the FOSS community will go, but there are more options than you would think. And when the bubble pops, it’s more likely the people that still have money after will be the ones not in the AI camp.

A short read, but I really got a lot out of it. One thing I’ve started losing with the rise of AI and didn’t realize until now is the communication with coworkers about current problems we’re working through. I reach out to people in the git-blame log, and others used to reach out to me. But now everyone asks the AI, so we’re even more siloed than before.

It feels less like we’re working together and more like we’re working next to each other.

I learned about JavaScript’s prototype-based inheritance model a decade ago. It’s different from every other major language, but it’s interesting and useful when used the right way. I’m astonished (1) people are still rediscovering “optimization” techniques like are described in the article when that’s how their constructors should have been designed from day-one, and (2) the JavaScript language still hasn’t added more language features to make the mutated-prototype inheritance pattern less verbose and more ergonomic. It’s a core architecture of the language and hasn’t been improved, it looks like, since the first time I read about it 10 years ago. They added the whole class syntax sugar to hide the differences in the language’s inheritance model, so they can add some improvements to the syntax for the better optimized inheritance pattern.

Except they take all of the land and resources with them. You posit the billionaires have a fully automated robotic industry at their whim and they will leave any scraps for the rest of us? It would be judgment day from the Terminator series, except instead of a super intelligent Skynet AGI mowing down humanity it would be psychopathic pedophiles.

The countries highlighted here have collectively cut gasoline imports by roughly a third so far in 2026 compared to ​the same months last year, while simultaneously boosting imports of Chinese EVs to record highs.

Doesn’t that correspond to the global crunch on oil caused by the war in Iran? Which simultaneously dramatically increased gasoline prices while also accelerating many countries’ transition to cleaner energy sources? It’s not clear their analysis accounts for this common factor.

It’s a web app and you just showed we can do it in text with Unicode and a modern browser. This bug is a level of self-inflicted, shouldn’t-be-in-the-realm-of-possible problem I struggle to comprehend. It’s the software version of not even wrong.

“Today, our long-range sanctions reached the oil refinery in Omsk – nearly 2,500 kilometres from Ukraine,” Ukrainian President Volodymyr Zelenskyy said

The man knows his messaging.

Also of note: According to the Financial Times, some of these SPV deals are being cut up, pooled together and resold to investors as new asset-backed securities.

Oh good. That risk-obfuscating strategy has never caused any economic problems and has only had positive results.

It’s cover. They’re not laying off because of AI. They’re laying off to make line go up.

Violate the law. The US civil rights movement of the 50s and 60s have a lot of success stories, and stories of challenges/failures that careful study of would really equip today’s civil rights advocates well.

The change will de-emphasize the effects of methane, a main component of natural gas that has a potent effect on the climate but breaks down more quickly than carbon dioxide.

I hate language like this. Carbon dioxide doesn’t break down. It’s stable. So methane breaking down in 100 million years would be “more quickly” than carbon dioxide. This turn of phrase puts an inaccurate picture in the layman’s mind of greenhouse gases that’s very avoidable. Just say “but eventually breaks down, unlike carbon dioxide.” There, fixed in the same number of words.

This combined with the OpenAI and Anthropic price increases mean I have hope the bubble will pop by the end of the year. Microsoft has to report their stats as a publicly traded company, and both Anthropic and OpenAI are aiming for IPOs in H2, so they will have to disclose stats for that as well. When all of their “customers” go up in smoke once the real price tag hits, they won’t be able to hide it.

AI in its present form is just too expensive without venture capital footing 90+% of the bill. They burned through that too quickly. They didn’t get the traditional economy addicted to AI enough; most companies can unwind most of the changes they have made to roll AI into their operations.

Edit: GitLab might be screwed, though.

I vaguely recall reading that such child fees would likely be illegal (in the US)? I think it might also be illegal to exclude families with children from being eligible to rent a given location. Don’t quote me on either, though, because I don’t have sources on hand.

I’m significantly allergic to pine. I wonder if this means I would no longer be able to take these medications if industry switched to this method.

The article discusses that IP-based limiting doesn’t work as well as it used to. Because of NATs, proxies, etc., IP addresses are a lot more ephemeral and flexible, so they’ve seen the same big perpetrators adapt and change IPs when rate-limited. I expect we will start to see support for anonymous downloads go away in the next several months in many major OSS registries.

This looks like someone took regular expressions, expanded them to a full programming language, and used Unicode to deal with the explosion of required symbols. I have a hard enough time reading my own regular expressions. I can’t imagine writing full programs like this.

First, don’t need to respond to arguments made in bad faith. There’s no net positive outcome possible for the other person who is coming to the conversation in good faith.

Second, not having their own solution does not invalidate anything critical they say about the one under discussion. There doesn’t have to be a “better” solution to justify not implementing something that has no positive impact on the targeted problem and severe negative impacts elsewhere.

How does the age inference model work?

We leverage an advanced machine learning model developed at Discord to predict whether a user falls into a particular age group based on patterns of user behavior and several other signals associated with their account on Discord. We only use these signals to assign users to an age group when our confidence level is high; when it isn’t, users go through our standard age assurance flow to confirm their age. We do not use your message content in the age estimation model.

Completely opaque explanation of how they use AI to guess your age with a claim that message content is not used. With no independent way to actually verify that claim, I don’t trust them at all.