I haven’t really kept up, but it looks like Character’s business model was still focused on NSFW content. If so, then this is also a disaster for the industry at large, since NSFW content is one of the few genuine money-printers for chatbot and image-generation services; people really do want thousands of iterations of the same fantasy, slightly changed in order to stay engaging, and Character was highly recommended as a source of affordable slop. (Examples of unaffordability include the rampant free-tier abuse on ChatGPT, Gemini, and Grok; I’m assuming that customers are willing to open their wallets at least a little. In general, the porn industry turns a profit, so I do think that people will pay for porn.)

My prior estimate (previously, on Awful and on Lobsters) was that token generation was subsidized to the tune of an effective 80-90% discount; this looks to be more in the 90-95% range. For contrast, consider chatbot-booster Willison’s belief, in reply above, that Chinese labs will magically produce a five-fold improvement in token production; this still would be between two to four times slower than needed to turn a profit for Character.

While doing research on our friends, I found (a disowned Reddit comment which found) another banger from the guy who pointed out that pinball is too chaotic for superintelligences to predict (discussed previously, on Awful): whole-bird emulation requires quantum mechanics, or “WBE requires QM” for short. Apparently WBE was a minor meme inside the community at some point, but I’m not sure who first noticed the situation. In short, birds with certain chemicals in their eyes may have the ability to sense magnetic fields due to QM effects. (I’d normally recommend WP on this, but the relevant article is bizarrely skeptical about the situation, although it does include a decent diagram.) As such, our friends consider the possibility that their vaunted superintelligence tasks (ancestor simulations, Newcomb predictors, etc.) are harder than WBE, which itself is much harder than e.g. cracking classical cryptography with Shor’s algorithm or some other not-yet-reached quantum-computing task. The author left a great sneer in the comments:

Adding a compass is unlikely to also make the bird disoriented when exposed to a weak magnetic field which oscillates at the right frequency. Which means that the emulated bird will not behave like the real bird in this scenario. You could add this phenomenon in by hand. Attach some detector to your compass and have it turn off the compass when these fields are measured. More generally, adding in these features ad hoc will likely work for the things that you know about ahead of time, but is very unlikely to work like the bird outside of its training distribution. If you have a model of the bird that includes the relevant physics for this phenomenon, it is much more likely to work outside of its training distribution.

Ro-ro-rotate your model of the bird, rotate your model of the bird for science!

I’ve done some of the numbers here, but don’t stand by them enough to share. I do estimate that products like Cursor or Claude are being sold at roughly an 80-90% discount compared to what’s sustainable, which is roughly in line with what Zitron has been saying, but it’s not precise enough for serious predictions.

Your last paragraph makes me think. We often idealize blockchains with VMs, e.g. Ethereum, as a global distributed computer, if the computer were an old Raspberry Pi. But it is Byzantine distributed; the (IMO excessive) cost goes towards establishing a useful property. If I pick another old computer with a useful property, like a radiation-hardened chipset comparable to a Gamecube or G3 Mac, then we have a spectrum of computers to think about. One end of the spectrum is fast, one end is cheap, one end is Byzantine, one end is rad-hardened, etc. Even GPUs are part of this; they’re not that fast, but can act in parallel over very wide data. In remarkably stark contrast, the cost of Transformers on GPUs doesn’t actually go towards any useful property! Anything Transformers can do, a cheaper more specialized algorithm could have also done.