I agree on scrollbars and tab design, but not on Thunderbird. Especially after the latest UI changes, and setting the spacing to compact (which is notably not hidden in about:config), it's the best-damn-looking and -usable e-mail client I ever used. And, finally, also the fastest (after many years of issues with large mailboxes).

Apart from that, I think it boils down to this: development needs resources, justifying resources needs market share, and market share builds, among other things, on good branding and marketing. A new mascot and refreshed branding probably isn't a big monetary investment, but I can see it helping a lot in winning undecided and less-technical folks over. If what browser they use doesn't really matter to someone, in addition to their techy friend recommending it, a cute fox mascot is just another little push that might get them to give Firefox a chance.

While it's always been a matter of national security, the big change now is that it's no longer just experts who know that now. People have always raised the issue of how insane it is to have critical services and infrastructure depending on a US service provider. The voices were never heard because nobody would take the worst-case scenarios seriously. In the end, it was still rationally capitalistic enterprise Microsoft/Amazon/Google/Oracle/etc. with money and a reputation to lose, operating in a democratic country with a working legal system.

All of those arguments disappeared very palpably (at least here in Europe) this year. Suddenly, when you say that there's a risk Microsoft could be politically forced to shut everything down overnight, even the random layperson will believe you.

Scams steal migrants? What the heck is that even supposed to mean?

You'll get your wish of bigoted, racist, sexist, ~homoerotic~ manly-homophobic AAA EA games soon enough. They're in the process of selling to the Trump family and Saudi royalty.

That's great. But how about an Outlook rewrite that actually works and doesn't crash five times each workday?

I usually have Outlook permanently on my third screen at the office. My favourite one is when I'm doing absolutely nothing with it, working in another application on another screen, and I notice a flickering on the Outlook screen. I can see the UI gradually disintegrating, toolbars moving around, areas turning black and purple, elements disappearing, before the whole thing crashes and restarts. Happens several times each week. And is just of the fun ways Outlook manages to mess up complex tasks like opening a window, moving an e-mail, or doing nothing.

I expect a vibe-coded, AI-focussed Outlook to surpass it in terms of entertainment value.

Hah, I had a similar experience recently with a library which had changed its API a while back

Claude suggested doing something with the old methods, which no longer compiled. I told it as much, to which it confidently replied that they do and I must be missing an import. I googled the new way on my own and asked Claude to do what I was asking using that new method. Just as confidently, it claimed that I was wrong, this function doesn't exist in the library, but it could help me implement one that does what the library method already was doing. I thought LLMs were supposed to be sycophants, always agreeing with you even when you're full of baloney?

With the flood of AI slop code now over us, this is only going to get worse. LLM coding has been as good as it was going to get, it's only further downhill from here.

It is in my corner of the world! A local restaurant put in two tables, just as a temporary attraction. That was about five years ago now, and they're still there, occasionally swapped out for other tables, as they found that they started making more and more money. Since then, I've discovered three more local places who have set up tables.

Stern is still making some pretty nice new ones. Jaws is my favourite.

I wouldn't say they're stupid. I'd say indignant, rebellious, and not giving a fuck. It was always said that someone should start pushing against the grotesquely unbalanced copyright laws by just breaking stuff and challenging the norms. I'm very glad that with the IA, we finally have someone who does. I'm very unhappy about the fact that it's the same entity that is running the Wayback Machine, therefore risking that mission by pursuing the moral mission of trying to force copyright reform by crowbar. If this stuff ends up dragging the Wayback Machine down with it (to be fair, it could happen for many other reasons, too) humanity ends up instantly losing most of its history of the past 30 years.

Correction: without the publishers' permission. Plenty of authors are on the IA's side in these disputes.

At least my Lego rocket didn't self-detonate on launch -.o.-

You're misunderstanding the "cookie law", which was the goal of the malicious compliance by ad networks (who are why we have annoying cookie banners -- not the law).

Unless you use your cookie to profile users and pass the data on to third parties, you're perfectly compliant with EU law without ever having to ask visitor consent.

There is no cookie banner law. The only ones who are required to place one are those with underhanded business practices.

Not disagreeing with the entirety of the second paragraph, but I'd say "learning to AI" is the polar opposite of learning to code.

In programming, you learn to analyse problems, design solutions, and build them. "Prompt engineering", which I assume is what "learning to AI" boils down to, is about how to avoid doing all of that, and by extension also to avoid any learning. Building software is about leveraging a (relatively) new technology to empower yourself and come up with cool and useful stuff. Using LLMs is about subservience, about developing a complete dependence on some big conglomerate's services. You learn nothing and you create nothing, you're just asking corporate overlords for permission to pay them for getting you out of having to think or work.

Already today I notice some people at my company become utterly helpless and useless if ChatGPT goes down for an hour. Learning to AI is the extreme opposite of an education.

There's no such thing as an AI worker, the term itself is pretty much a contradiction.

Both have their place.

I listen to an insane amount of new music, trying to discover new things and stay on top of new releases by artists or labels I already like. I'm easily listening to 10 new albums a week, and maybe 100 singles. Having most of this available on one platform (Qobuz for me) for a monthly fee is a great deal. It basically replaced freeform radio, podcasts and piracy for me as a way to discover and find out what I like.

What I really like, I still buy. What I really, really like, I buy at my local record shop. But finding out what I want to buy, and listening on the go, has become so much simpler since streaming services, they're more than worth the price for me.

He's still Executive Chairman, I don't think it changes much. He even says himself that it's mostly a title change to reflect what he already was.

I don't know YouTube's terms by heart, but I'd guess by uploading your videos, you gave them something like a "perpetual non-exclusive license" to distribute your work commercially (ads, etc.). In that case, OpenAI would only have to negotiate with YouTube -- you already agreed to whatever YouTube decides to do with your uploads.