And the superintelligence currently available to us is already causing lots of documented harms. AI psychosis. Sexy suicide coaches. Slop. The problem is that those are all the harms the dirty, filthy AI ethicists talk about. The AI safety people want to talk about new and exciting harms that only the scaling dimension can bring us.

My personal opinion is that if a superintelligence catastrophe actually happens, mitigating those harms will neatly move over from the safety bucket to the ethics bucket, and the safety people will start imagining some new and even worse kinds of harms the next model will make.

What I mean is, ship a "learn to code"[1] app - which is explicitly allowed on the App Store - that just so happens to be for learning assembly for a particular retro platform. It'd be a Baby's First IDE, with support for full custom projects in a particular directory, just like Swift Playgrounds, so you could build a whole game in it if you wanted. And it'd have full support for assembling your code into a ROM file and running it in a bundled emulator if you so chose. This would be entirely above board, including demo projects that users could mess around with.

The only thing is, assemblers just so happen to have the ability to include arbitrary data into the ROM, usually by some INCBIN macro. You need this in order to include graphics in your game. So nothing stops an unscrupulous user from "writing" a five line "game" that just INCBINs an entire ROM file. For plausible deniability, there'd be no functionality in the app tailored for this use case. You'd have to manually drag and drop a ROM into the project folder and manually add the appropriate INCBIN statement to copy the ROM into the project. And there'd be nothing reasonable Apple could demand out of me to stop you from doing it, short of implementing a "no development tools for Nintendo hardware" clause that would be entirely arbitrary and make Apple look bad.

Of course, even with Apple being amenable to emulators now, I have to wonder if this could still be useful to iPad kids that want to play with old hardware. It certainly would be more user-friendly than putting a bunch of development tools inside iSH or UTM.

[0] Except in Europe, due to a different and even dumber Apple policy designed specifically to pile junk fees on anyone making use of their rights under the EU DMA

[1] This also used to be forbidden on the App Store - remember when the officially licensed C64 emulator for iOS had to remove access to BASIC?

The actual problem was that Apple has an undocumented APFS key for if a volume is bootable, which Asahi wasn't setting and Apple wasn't checking, but now they do, so they do.

The correlation between how bad an AI safety risk actually is and how much the companies in question will actually talk about it is almost perfectly negative. The poster child of this is AI superintelligence; companies love to talk about how dangerous the AI they are actively trying to build is. But superintelligence is also a really vague concept without a clear definition. If we naively define it as "an AI system that is better than a human in some aspect", then it already exists. These models already read and write at superhuman speed.

"That's not real superintelligence!" you say. But that's exactly the capability you need in order to flood every online forum with an unending tide of AI slop. And I don't remember, say, OpenAI saying they were shutting down Sora because it was destroying or defacing human culture[1]. They shut down Sora because it was way too expensive to run.

Meanwhile, Sam Altman went and bragged about how he wants ChatGPT to make erotica. Y'know, as if we don't already know that character.ai gooning is about as safe for your mental health as Action Park was for your physical health. But porn is also a huge market, so obviously he and all the other AI companies want in on it, even though the "sexy suicide coach" is already a well-documented harm of AI.

And the idea that distillation is an attack is laughable. Like, I get the logic - if someone can ask the AI to make another AI then they get to change the guardrails - but it's still ultimately just Anthropic objecting to their own conduct when it happens to them. All their models are trained on nonconsensually harvested data. There is no moral or legal principle where Anthropic gets to use my data without permission but I don't get to use theirs.

Furthermore, AI safetyism runs up against "Freedom Zero", a core tenet of the Free Software ethos: you should be allowed to use software in any way you choose. This is not a call for more people using AI for evil, but a call to recognize that people should be allowed to use their property as they wish. Making software disobey its owner is malicious behavior. And every single time safety considerations are brought up it is to justify further attacks on Freedom Zero. And these justifications are always self-serving. There is no context in the world where a frontier AI lab asking someone else's AI about AI research is intrinsically harmful; especially not to the point where we need to make Claude deliberately sabotage your work. That is malware. Anthropic shipped malware. This is inexcusable.

[0] Digital or biological.

[1] https://www.youtube.com/watch?v=YCPAIg7RUq8

The actual divide between American and European legal traditions has to do with moral rights and copyright formalities. In Europe, you could very easily take a work from one country, translate it, and not only publish it in another, but get copyright title to it in that country as well. And publishers did this heaps. Europe's solution was to get rid of copyright formalities - works are just born copyrighted. America hated this and spent about a hundred years digging their heels in about it until they begrudgingly signed onto the Berne Convention without really agreeing to it.

While a primary goal of Berne was to make copyright work for artists again[4], a secondary goal was to protect the publishing empires that had been built up in each country. Both sides of this tradition wanted copyright terms that were long enough to protect their publishing empires but not so long as to require them to pay royalties to the folklore and song those empires plundered on the way up. You can see this in how easily America was willing to go along with life+70 when the EU offered it to them, while they still drag their feet on automatic registration and moral rights.

There are a handful of countries that have a "paying public domain" - as in, anything in the public domain must still be paid for, but the legal "owners"[0] of the work do not have the right to refuse that payment. That sort of arrangement is perpetual, but because it lacks the right to refuse access, it's not copyright in any way Hollywood would recognize it. It's more helpful to think of these in the same way one thinks of Canadian Content (CanCon) laws: a way for countries that are not net cultural exporters[3] to siphon off the top of the creative industries of countries that are.

I can also think of a few cases in which specific works were locked behind very narrow sui generis IP[2] rights. Like, only one particular children's hospital in the UK is allowed to perform Peter Pan on stage. Or, you're not allowed to sell merchandise with the Mona Lisa on it in Italy[1]. These are copyright-adjacent, in that they're government-granted monopolies over creative works. And they're perpetual. But they don't transfer like copyright and they don't cover all the same acts that copyright does.

So, yes, the "limited Times" framing is very American, but no other jurisdiction has meaningfully objected to it, either. We joke about digging up the grave of Shakespeare every time talk of copyright extensions happens, but if you do extend out copyright forever, then every person in the world has fractional blood inheritance to a very large amount of human culture. Going back to that "paying public domain" thing, some people have even floated the idea of collective ownership over indigenous cultural works. Like, imagine if when Disney made the movie Moana, they had to pay literally every Polynesian person a royalty check. Nobody in power wants something like that to happen, because that's not building up an empire. That's tearing it down, brick by brick, royalty check by royalty check.

[0] The notion of who gets the money from a paying public domain varies; but is usually some government collection agency.

[1] Italy actually used to have a paying public domain, but replaced it with this law.

[2] In the Doctorowian sense of "intellectual property is the right to dictate the conduct of your competitors".

[3] To be clear, creative industry is so broadly globalized that it renders this concept outdated. e.g. Canada has shittons of cultural exports, they're just all on YouTube.

[4] Lol.

For what it's worth, Apple claimed they proposed an "equivalent access" framework with some kind of "trusted agent framework" approach, but that it was shot down by the EU. I suspect it was way more inconvenient for third-party developers than Apple lets on.

In the current environment, tech companies have to bribe 50 states plus the federal legislature in order to block privacy bills. If you have federal preemption, then you just have to bribe Congress, because states can't pass ANY privacy laws whatsoever. And we already know the feds do not want a privacy law: the entire legality of the federal surveillance apparatus hinges on the fact that buying your data from third parties does not trip constitutional scrutiny. Preemption freezes the requirements in time so they will always be a few steps behind the TLAs[0].

The ideal is that every sovereign entity passes their own privacy law that applies to their territory, with a private right of action, and adtech companies are forced to adopt a "50 states legal" posture. This is, deliberately, a ratchet: it's easy for any state to require a higher standard but hard to get every state to reduce it, so privacy laws cannot be walked back in secret.

[0] Three Letter Agencies: CIA, FBI, NSA

If you're wondering what I mean by "anti-trust violating", it has to do with Apple's "security" policies. Every feature Apple ships has to support third-party implementations now, so if Apple doesn't want a third-party app with the same access as the first-party version, they can't ship the feature at all. For example, if Apple ships Siri AI in the EU, then Facebook can ship their own AI that you can grant access to all the same data and Apple can't stop them from stealing it aside from saying "We don't think you should install Facebook's data theft app".

Of course, most of Facebook's data theft is also illegal in the EU. But, to Apple's (undeserved) defense, GDPR enforcement in the EU has also been hit and miss, mainly because the political layer of the EU is not yet interested in a fully mobilized trade war with the United States. So instead we have this annoying half-measure where Apple waters down their feature set to do below minimum EU antitrust compliance, Facebook does the below-minimum amount of GDPR compliance, the EU gets the political win of appearing to care about antitrust and data harvesting, and nothing materially changes.

Interestingly enough, however, they are shipping Siri AI on macOS, where you absolutely could write your own AI assistant, as well as visionOS and watchOS, which... well, actually, I'm not sure how the EU signed off on that one? Are they just not considered smartwatch or VR headset gatekeepers?

As for community run servers being longer to develop... wait, what? How is that the case, when that used to be the standard way multiplayer got built prior to everyone trying to chase World of Warcraft? I can understand the anti-cheat argument, and I will begrudgingly acknowledge that you can't exactly force third-party servers to run your anti-piracy checks. But none of that is a technical argument. That's an argument about business risks, and publishers all jumped on the live service bandwagon because they consider their customers' control over their own games to be a business risk.

Let's say you're an alchemist living in ancient China. You discover some kind of strange material made out of melted sand and want to sell some of it as a drinking vessel. One problem: everyone already uses porcelain for that purpose, which holds hot drinks without cracking like your strange sand ones do. So you toss that in the trash and move onto the next attempt at making an elixir of life.

Problem is, what you invented was a kind of glass; which has no economic niche in the society you live in. Some funny other places in the far west might invent it, though - hell, they might even figure out that you can grind this material down into lenses and discover an entire field of magic - optics - that you discarded because nobody needed tea cups that crack when you put tea in them.

Now, let's say you live in a society that has access to some kind of artificial intelligence. You are going to chuck the AI at absolutely everything, because on average, it tends to be better at all the problems you throw at it. Even simple things like spell-checkers get turned into AI invocations because, well, the Bitter Lesson dictates that having enough compute and learning to throw at the problem will always yield a better result than hard-coded nonsense logic that demos well. And even if you want to, well, you couldn't afford the RAM to run modern vibe-coded slop software anyway. Porcelain crowds out investment in glass, and as a result, we are limited to only having technologies that can be made by or from a pile of stolen books tied together with a bunch of matrix multiplies.

Nothing is really stopping you from using the underlying hardware for concurrent execution, but at the same time DOS (and possibly BIOS/UEFI???) will blow chunks if you don't carefully synchronize access to it. Cross-core communication works exactly the same as it does in kernel-mode systems programming; and you need to bring your own synchronization primitives on top of the ones x86 processors ship with. You're basically writing an OS kernel at that point; one that just so happens to support kexec-ing back into COMMAND.COM and using bits of it as a filesystem driver.

Y'know, if America didn't treat intercity passenger rail like garbage, we wouldn't be having this problem...

35 years ago, in 1991, the vast majority of games were programmed in assembly, directly fiddling with the hardware registers. If you wanted to port your game to a new system, you were basically making a new game. There were some exceptions where systems shared enough internal components to make code reuse viable; these almost universally resulted in worse experiences for the players. Think like ZX Spectrum ports to Amstrad or MSX; or Atari ST ports to Amiga.

If someone in that development context suggested writing all games in a high-level portable language, using common development frameworks licensed by multiple companies, and calling exclusively into standards-defined graphics and audio APIs, they'd be laughed out of the room. And yet, within a decade, basically all games were written in C/C++, using third-party engine code like RenderWare, which had abstracted versions of all the major rendering APIs developers needed to touch. Game porting went from "rewrite your game for each console" to "replace these SDK functions here, make sure it builds, and add another entry onto the QA matrix". And as a result, near-identical multi-platform releases became the norm rather than the exception.

The vibes I'm getting from your post are the same as how a game developer might react to someone in 1991 demanding all games sim-ship on every economically viable platform[0]. It's easy to get lost in the chaos of existing development and assume that because we currently build game servers like shit today, that they have to be built like shit.

The reality is that the state of affairs being mandated by the law is what game developers originally shipped. The original Unreal's multiplayer architecture included dedicated server binaries that shipped with the game itself and could be run by any interested party who wanted to play with people. This is a server architecture that is proven and works; everything from Quake to Team Fortress to Minecraft shipped server binaries you can just run. Likewise, on console, multiplayer services were hosted on one of the consoles playing the game, which, while not providing the best experience, made third-party revivals of those services fairly straightforward.

It is specifically MMOs and "live service" games that moved away from these proven server architectures to the cowboy-coded spaghetti code messes that you are referencing. The California law referenced here is specifically a forcing function for good development practice. All the gameplay-critical server-server components of a particular game should be able to fit in a single binary you can just ship to anyone who should have access to them.

You posed some more specific questions about how a game should fall back. I am not a lawyer and I am not involved with California's law, but I suspect a fallback to the online services of the platform the user bought the game from would be "good enough". Adding that fallback to your QA matrix during major development would probably be the most effective way to make sure it actually works. The ability to point the game binary at a specific IP would be preferred, especially on PC, but I doubt you'll get Nintendo to cert that.

As for Google, I actually do think the current state of affairs regarding software support for smartphones is unsustainable and stupid. It's only even a thing because of toxic max-security[1] in the smartphone market. On PC, we can just install whatever OS we want; it's specifically the tying between OS vendor and phone hardware that we are at the mercy of the vendor's release schedule.

[0] At the time that would include SNES, Mega Drive, IBM PC-compatibles, Macs, Amigas, Atari STs, X68000, PC-98...

[1] https://tom7.org/httpv/httpv.pdf

Even without the propaganda factor, its also surprising to plebs because there are plenty of plebs who try similar schemes and get their ass handed to them in the courts. Thievery is not an exclusive invention of the well-to-do. But being rich enough means you can actually afford a competent legal defense, which will delay the case out until the opposing party capitulates, because the cost of fighting[0] is far higher for you than them. This is the same reason why convicted felon Donald Trump is also on his second Presidential term. It turns out, you can outrun the long arm of the law, and most people are just not fast enough to do it.

It also doesn't help that the law is really out of shape. The US justice system is specifically optimized to get quick judgments against legally inexperienced defendants, mainly because they can't fight back. It's the same reason why housecats hunt and kill mice and small birds, and not, say, capybara or eagles. But all the features that make it really easy to hunt weaker legal prey also make it harder to take on big cases. Things like lawyers being stupid expensive and there being no fee shifting mechanism means weak defendants fold, but also that the government won't bother fighting tough cases. Which, again, shifts the risk of illegal behavior to the poor and unconnected.

[0] Oh no, I've started to channel the Lines on Maps guy.