Ideally, pay cold hard cash to opt in.

Second choice would be to provide some real add-on value to opt in.

What's really going to happen if opt-out becomes too toxic for twitch: They will go opt-in but enshittify the product for those who don't opt-in, touting a "better" (meaning maybe as good as August 2026) experience for those who opt in.

Sort of like the "opt in" toll roads in places where the free roads went from being decent to dangerous due to "lack of funding" to pay for maintenance.

The parallels between this and letters of marque are scary.

I'm reasonably confident there were other, previous known-at-the-time deaths from trebuchets but the records of their deaths or the reasons for the deaths have been lost.

https://www.youtube.com/supported_browsers?next_url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DCrbp8Y_yCIs

is not what you want

https://www.youtube.com/watch?v=Crbp8Y_yCIs

may be what you want.

The problem with the above is that you are mistaking "yourself" for "the market."

"Everyone isn't doing what I want them to, obviously 'the market' is broken."

You can bet that both black and white hats are poking around any popular tool that allows collaboration.

Teams, Discord, and others likely have high-impact bugs yet to be found.

If it has a high false-error rate the city may miss out on revenue and someone with a warrant may not be arrested, but any false-positive is an injustice to the person pulled over.

I'm willing to be the one-in-10,000 false-positive from a license-plate misread but not the one-in-100 (1% false positive) or 1-in-4 (25% false positive).

On the other hand I can tolerate my city missing out on ticket revenue or a missed warrant-arrest because only half of its license plates can be confirmed as "almost certainly correctly read."

If all characters had the same failure rates, that would be a plate-failure rate of about 25% per plate. Yes, I know it's not the same for each character, so the per-plate accuracy may be higher.

Until computer-only recognition is about as good as either human-only or human+computer-only recognition, you need a human in the loop.

You fundamentally misunderstand the purpose of my proposal. The purpose of my proposal is for AI agents to tell web sites "you are gaming the system, we are going to do whatever it takes, no matter how inefficient, to make sure you don't."

This kills me. Instead of a "simple" (I realize that e.g. a google search is not all that simple) database query, we have an LLM that does the query for you, then analyzes it, and redisplays the output (making it thousands of times more computationally expensive) but now you are suggesting that the chatbot control a VM or similar and perform the steps as if it were a human, adding millions of times more overhead. It's mind boggling.

Conspiracy theory: the "anyone" doing the promoting is already affected by the product they're hawking and it's trying to spread.

I assure you that the existence of these things does not mean you want to have to deal with actually exercising your rights under them, and that you're stupid if you don't go over the exact vehicle you're buying before signing anything. And that going-over includes "driving" the thing.

You have to be talking about Europe or something, because most people buying new cars in the US are buying off the lot. Dealers order vehicles with option packages most likely to appeal to a broad set of customers or designed to maximize their sales. Some people do order the exact vehicle they want, of course, but they're the minority.

A correctly written, correctly-implemented "human filter" would render this invisible to the AI.

It's principle, not principal. AIs are not my pals.