Balinares

Source?

10 bucks say it's just a high-demand region and GCP's hardware supply chains are at capacity. So you see stockouts.

I do like Flatpak, for the most part, though I wish that the distribution and the sandboxing parts of it were not tied at the hip.

The sandboxing would be an easier sell if it allowed interconnections between apps as appropriate. Yes, I do want KeepassXC to be exposed to the KeepassXC extension in Firefox. Yes, I do want yabridge to expose VST instruments to BitWig. We don't want to be hardcoding support for individual pairs of apps either, so the mechanism would have to be generic.

Abstracting away filesystems also only goes so far. Running Filelight sandboxed for instance is just too slow to be useful.

And the distribution would be an easier sell if it allowed for standalone app downloads without going through a centralized hub with dedicated download tools.

Hopefully these are things we'll eventually see solved. Until then, it's still pretty damn good to be able to install many common apps regardless of distro specifics. An easy thing to take for granted, that. Used to be the answer to installation woes was "what's so hard about tracking down every dev dependency and compiling it yourself?" and it's nice that we're firmly moving on from that.

Possibly because the Gnome project has announced a Flatpak-based distribution (if I got that right anyway), which would have brought Flatpak prominently to people's minds again.

Strata is intriguing and it definitely looks like there's something useful happening there with weight streaming at usable speeds, but I've not managed to have it serve a full answer without inference crashing before long. Pity.

I'm pretty sure it isn't. Or at least I should clarify: Anthropic got some orgs (Honeycomb people mentioned it for instance) on committed spend contracts, which de facto makes using non-/A models unworthwhile since the /A spend is already committed anyway. Which I'll grant you is only functionally the same as lock-in if not quite definitionally.

Friend, it makes me so sad that you can't when imagine how much better off your life could be if you managed to fight for and secure the same rights others did. I sincerely hope you'll get to find out for yourself.

I'm afraid it isn't much, and getting worse fast. You can still just turn up if you have EU citizenship, and the place legitimately isn't without merits, but the personal rights acquired over centuries of social combat and the odd monarchial beheadings are vanishing fast, as far as I can tell. There's a reason we fucked off outta there.

If you have the immense luxury of picking your destination, the Netherlands or much of Scandinavia seem to remain the best options.

Actually a good question, and in practice: no it doesn't, not significantly. Aside from 2008-style systemic failures across the board, the risk is non-correlated across borrowers, and hedged through the collateral of the property itself anyway, whose deed remains with the lender until the mortgage is paid off. In practice, due diligence regarding the borrower's present finances seems plenty. This seems especially true where there's a decent social safety net that reduces the occurrences of individual bankruptcies.

And anecdotally, despite the aforementioned letter, we got top market rates on our mortgage.

Worth noting that 2008 occurred in part because mortgages were seen as entirely risk-free in aggregate, even as credit reports were already a thing in the US at the time.

As far as I can tell, credit reports are more a tool that lets lenders wring more money out of borrowers in confidence that their competitors will not undercut them, since the borrowers carry their credit reports with them.

I use Docker Compose so that I don't have to care if the daemon is a single binary or a whole stack: I can use the same setup for everything, and deploying an entire new thing in a maintainable way is quick and easy. Each app's startup/shutdown is managed using the same templated systemd unit, and Traefik handles the routing and Let's Encrypt cert creation and renewal.

Basically I want things reliable and low-maintenance and this setup gives me that.

The capability ceiling of models in general has risen massively in not even one year. Back in 2025, I also found them to generally be a net-negative expense of time and effort. I wouldn't universally make the same claim today.

I'm told that Anthropic has been locking enterprise customers into contracts with a mandatory amount of spend in exchange for a minor rebate. I don't know how widespread those contracts are, but either way, my guess is that when they eventually expire, those customers will switch to e.g. DeepSeek for the 80% of workloads that don't need the extra oomph of Opus. I don't know if the numbers at that point will add up to Anthropic surviving. They do seem to be concerned, judging by how hard they seem to be pushing for regulatory capture.

Well, there are multiple approaches, but here's one that worked well for me: when I purchased a home in the country where I now live, and the mortgage lender asked for a credit report from the (high-privacy) country where I lived the longest, what they got was a letter from that country's central bank listing the penalties for requesting that information. Up to and including jail time.

Suddenly the credit report was no longer a prerequisite to the mortgage.

Some never left. Notoriously, some bigtech companies have an internal IRC network for tech discussions because that's one communication system that you can rely on to stay up even during catastrophic outages.

Because it succinctly but correctly spotlights the motivation for the release and the fact that motivation is self-serving and not necessarily a positive for the field.

Duchamp's Fountain.

Anything can be art.

3.8 is absolutely quite a bit better than 3.5. My go-to quick benchmark is a bit like yours, implementing a simple but complete game in pygame. Most small models output code that's too buggy to be fixable. Qwen 3.8 wrote buggy code too, but the bugs were minor, and the game legitimately playable and fun.

That said, I've not found it very reliable, especially at the low quant required to run on my hardware. It's very capable for its size, enough to be legitimately useful, but it's never certain whether it will hit its capability peak on a given attempt, and sometimes you have to give it multiple tries.

100% wealth tax rate past e.g. $999M -- capping wealth like a Zelda wallet -- would address a certain number of issues around the many externalities of unbounded, bottomless greed.

Mind you, I don't think that's ever happening, considering that one such issue is there's currently nothing to keep a dishonest actor who ended up beyond that threshold from spending mountains of cash (still a small fraction of the total there) toward convincing those susceptible to such that a system which permits such dishonest actors is good akshyually.

Your example supports the thesis that R&D tax credits are in fact decorrelated from R&D successes so I'm confused what point you are trying to make.

The Macron government went out of its way to cut taxes for the wealthiest percentiles. Debt exploded, costs were cut, infrastructure went unmaintained (which only ended up adding to structural costs), but the GDP growth remained anemic.

However the wealth of the wealthiest increased substantially, so there's that, I guess.

This is a good question. It's worth noting that the US forces did not in fact uniformly liberate every concentration camps prisoner: they left those incarcerated for being homosexual. So even back then the lines were blurrier than commonly taught.

I mean, there is such a thing as reality, and it exists regardless of your observing it; we can agree on this much, I'm sure?

Also super fast loading times. Playing e.g. RDR2 on a physical console, by comparison, is painful.

Stadia worked absurdly well. One of the beta games was Thumper, a demanding rhythm game (I might be the reason why it was in the beta lineup actually). How well did that work out? Well, the latency made it hard to play... until I switched my TV to so-called "gaming mode". And then it was fine.

No kidding. The TV was introducing more latency than Stadia itself.

I also once tested it on hotel wifi. It worked great.

(Mind you, this was in Europe. YMMV.)

The business model however was terrible and killed the product IMO. Which, honestly, good: Stadia would have killed video game archiving efforts. If a publisher folds and Stadia and similar platforms stop streaming its streaming-exclusives, then those games are gone. Not worth it, no matter how well it worked.

You'd think reducing crime and saving money in the process would be enough incentive, but it turns out that can be a hard sell on a public that's been trained to relish in seeing violence applied to people presented as bad. I've found that there are people so keenly keyed to said relish that they'll just dead refuse to believe that supportive approaches to drug addiction can be cheaper than coercive approaches with better results to boot, and that's kind of tragic honestly.

Unless they already have something similar of their own, which is always possible, they'd be stupid not to. I don't suppose we'll ever know, though. It would not be a good look if after the trillions of dollars that have been thrown at US labs, investors found out that they're down to copying Chinese tech.

I don't understand on what you are basing this reasoning? If a well educated workforce can produce Fable then why couldn't a well educated workforce produce MiMo?

Besides, the latter actually published and open-sourced its RL stack to make it reproducible, which would in fact make it more trustworthy than the models you are speculating were distilled.

I wonder how well it performs in practice, because I can't help seriously doubting those benchmarks. That would put this 6GB model in Opus 4.6+ ballpark. Granted, that's mostly to Qwen 3.8's credit, but it's hard to believe that Qwen's already unbelievable capability density can still be compressed this much more.

The number of people abject enough to end up with transphobic views is thankfully nowhere near that. Most people, as it turns out, are not creepily obsessed with other people's genitals. Most people also aren't driven by a horrible urge to find minorities to torment. Though those who are would sure claim otherwise, of course; trying to normalize their awfulness is part of the playbook.

Because then they'll move on to the next victims and in having surrendered without even a fight you've granted them momentum for no good reason whatsoever, and you'll have fewer allies left too.

You are however under no obligation to grant your instincts unrestricted access to the pilot seat.