Could you tell us what that app is?

And yes, the fact that they blast out amber alerts at the same level as “imminent nuclear attack” is unbelievably stupid. There’s no justification for it whatsoever.

Ah but consider… It’s a game you know, all your friends are there, and the rules are being explained very, very badly. And you’re not allowed to say anything because everyone at the table glares at you if you try to interrupt.

Satan is now six minutes deep into explaining how the corn exchange phase works, but he’s not even told anyone what the objective is, the basic actions, the flow of play or anything else. He’s also gotten half the rules wrong and keeps contradicting himself. Everyone’s eyes are glazing over. The corn exchange phase isn’t even used unless a particular faction is in play…

Pity the vortex only exists in the same way that Armata does; yes, a few people have seen one, but they’re not actually being deployed anywhere in meaningful numbers. Even the units issued the M7 still don’t have them.

And their actual practical benefit is… Questionable. I regularly interact with Canadian combat arms, and it’s notable that the people who actually do frontline stuff think the Vortex sounds like ass. None of them have the slightest interest in a scope that claims to aim for you.

Then again, those same people have been deployed alongside US military, and been OPFOR in training exercises against them, they’ve seen them shoot, and they were not impressed. Consensus in the CAF is that US marksmanship is dogshit. So I guess I can see the appeal of a device that attempts to replace training with technology, provided that the training you’re replacing is really, really bad.

they can postpone actually building this thing until he’s out.

I cannot even begin to sufficiently emphasize how wrong this is.

First off, to be clear, these are design changes to carriers that are already in production. These aren’t future projects, these are decades in the making projects that are finally coming to fruition. This is literally Trump asking them to stop work on a ship that is half built so that they can redesign it. And we’re talking major redesigns here. The kind that involve basically scrapping everything you’ve built and starting again.

What’s worse is that these are ships that are already seriously delayed, and very seriously needed. Carriers are the primary mode of power projection and the core of every naval battlegroup. A carrier is basically a floating forward operating base. They are vital and essential, and most of the 11 in service today were built before the turn of the century. Some are fifty years old.

The original plan for the Gerald Ford class called for one ship to be built every five years. That began in earnest in 2009. Seventeen years later they have… One.

That’s really bad. A ship that is supposed to take five years to build, on average, is currently sitting at over fifteen. Now, to be fair, production speeds on anything start slow and improve with time as you get the process down, but even allowing that caveat, the Gerald Ford class is very, very badly behind schedule. The original projections were for the Nimitz class to be fully retired by 2058. The US is about a third of the way towards that deadline and 1/11th of the way towards that production goal.

So what Trump wants to do now is take three currently in production carriers, all of them desperately needed, all of them seriously delayed, and throw out all the work that’s been done because he doesn’t like their aesthetics.

This is a disaster for the US Navy, one that compounds upon decades of disasters. The last fifty years of US shipbuilding have been failure upon failure upon failure. The Zumwalt and the Litoral Combat Ship, both intended to revolutionize the Navy, both reduced to production runs of two or three ships, left hanging on to the fleet like useless vestigial appendages. The new frigate, cancelled because the Navy couldn’t stop over-designing the damn thing, leaving the US with an up-gunned coastguard ship as their potential new frigate instead. The new destroyer, a desperately needed replacement for the ancient Arleigh-Burke, left on life support after its budget was pilfered so that Trump can have a battleship.

Other than the Gerald Ford, the US Navy basically hasn’t successfully produced a new design of ship in decades. Meanwhile China is pumping out Type 55 frigates by the dozens. it will not be long before the US is thoroughly outgunned at sea. And worse still, those creaking old Arleigh-Burkes have hit the absolute limits of their hull and power plant. Future upgrades? Forget it. You won’t be mounting lasers or railguns on an Arleigh-Burke. Don’t get me wrong, it’s a very good ship, but it’s at the limit of its design. The new Chinese ships are not.

Even Trump’s battleship is a monumental problem, because it’s pulling vital resources away from all of the projects that actually need it. And because of its immense size and the recently added requirement for a nuclear power plant, actually building it will require the same facilities that are currently building the Gerald Fords. So you’ve got the vast majority of all shipbuilding resources now being focused on two ships that literally cannot be built simultaneously. Meanwhile shipyards that could be putting out new destroyers or frigates are doing nothing because there are no new designs to build, and the expertise and money needed to make those new designs has been co-opted.

The clock is ticking, and the US Navy was already running out of time a decade ago. They are in serious danger of no longer being able to pose a credible threat to China at all. The fact that the US currently does not have a carrier in the South China Sea, for the first time in decades, is absolute proof of that. There has never been a time in history when the Navy could less afford to let an idiot child play ship designer.

That last part is key. Data centres use up a huge amount of real estate and produce incredible strain on local infrastructure, but they produce an incredibly small amount of employment. It’s mostly just a few low-wage security guards. You could fit dozens of other businesses, employing thousands of people, into the space taken up by one data center.

While I agree that “settler” is too soft a term, “terrorist” is a completely nonsensical substitution in this context.

What I think you’re reaching for is more like “invader” or “occupier.” This is why many people refer to the Israeli Defense Force (IDF) as the Israeli Occupation Force (IOF).

When this is all over the world owes Ed Zitron a massive fucking apology.

They’re not even better optimized. Hallucination rates are up. Inference costs are up. There’s only an AI industry at all because they’re selling a highly subsidized product, but when they try to raise prices even a little the market collapses. Companies that were encouraging employees to up their AI use are now rationing tokens like chocolate in wartime. This isn’t like Uber where they can push out the old providers and then obtain market capture on something everyone needs. AI is not, and cannot be, essential, because you can always just get a human to do it.

I’ve been using it for years and this has never been an issue for me.

Yes, Riven really improved on that aspect, by a lot. The puzzles feel incredibly natural and purposeful. Everything in Riven serves as world-building. Everything has a reason to exist. And the puzzles flow naturally together. It’s a game of discovery that rewards exploration. I’d absolutely recommend the recent remake.

Very cool. I’ll need to sit down and read the full spec, but I’ve got some FastAPI apps where I could see this being useful.

Yeah, hi, um… Fuckface McParasite… You have literally zero fucking say over what people do in their home you fucking moron. If that means your precious fucking lightswitches wear out faster, too bad, so sad.

More circular financing. Definitely a sign that everything is going great.

Well, yes it does, but I’m not sure what story exactly. It’s certainly not the easy “Trump did it” that we would - with good reason and precedent - usually assume. Seems like he actually inherited this one.

deleted by creator

The really obvious question here is “Why should the president be the final arbiter of what is or isn’t accurate history?”

Absolutely there’s a difference. LLMs, when it comes to this specific task, are better. That’s why they’re being used here. It’s a job they are uniquely well suited to. They do indeed come with high hardware requirements, which is why you’re not forced to use them, and why they provide the option to offload the work to a cloud service.

Personally I would absolutely not want to ever feed my documents into an off-device model, but the point of self-hosted software is that it does what you tell it to and they absolutely should include letting you make bad decisions.

I think you’re under the impression that the difference between those things is far greater than it actually is. Large Language Models work by developing statistical maps of associations. That’s discrimination. They’re a direct evolution of categorization models. The ability to associate a hash of a JPEG with “cat” is the same as the ability to associate “How are you?” with “Great, how about you?” It’s all associative mapping. LLMs are just the current leading edge of that technology. If you want to, for example, generate a list of tags that describe a document, an LLM is the best tool we currently have for doing that.

To put it another way, what you term “discriminative AI” is “generative AI.” It’s generating a category or list of categories in response to an input. That’s not functionally different than generating a sentence in response to a sentence, it’s just an order of magnitude less complex. You can argue terminology but the technology exists on an evolutionary curve, with no real hard boundaries.

Paperless and Papermerge have always done a lot more than just OCR. If that’s all they were, most people can do that already on the software that comes with their scanner. The core selling point of these applications is automatic categorization, sorting and tagging, and those have always relied on machine learning tools. Literally the first thing you do after setting Paperless up is start training the AI.

what’s the best alternative?

You’re asking for suggestions for an automatic document categorization tool that doesn’t use any automatic document categorization?