I read some speculation today that British election law might be interpreted to mean that if Jon Harvey wins the election as Count Binface, then he would not only be permitted to wear the Binface costume and persona while sitting in the house but he would be required to. Since it's Count Binface that the people chose and gave the mandate to, not some regular old human schlub named Jon Harvey.

Presumably there'd be nothing wrong with him taking it off when not on official business and having a secret identity that way, but for votes and whatnot he'd have to don the cape and bin.

Russia's getting bombed pretty regularly deep in its heartland these days, and there's no gasoline anywhere. It's a bit more than just the fringes now.

Meanwhile the AI trainers get all the deleted content by buying direct from Reddit Inc., so this kind of thing only hinders human readers.

Even the worst Kessler syndrome scenarios don't stop new satellites from being launched through the debris-containing altitudes to reach clear orbits.

As a Canadian, an "apology tour" would feel like this to me.

If the next administration wants forgiveness it'll have to show real progress on fixing America's broken system of government and its broken culture. That'll take longer than four years so the next administration should not expect forgiveness. The best it can hope for is to get started on eventual forgiveness.

In the meantime we need to keep diversifying away from America and building defenses against them because there's no reason to extend the benefit of the doubt. They elected not-Trump after the previous time Trump was elected and where did that lead them? Back to Trump.

A common cause of this kind of problem is lots of people simply default to whatever the "best" model is and throw every problem at it. The best model can handle those problems, sure, because it's the best model. But it's also the most expensive model.

A better strategy is a multi-model agent that breaks tasks down into smaller sub-tasks and then uses the minimal model that can manage each of those tasks. The high-level program architecture can be figured out by Fable 5 or whatever, but then each of the functions and classes can be written by a cheap local model like one of the Qwens, for example. An AI that's been told to find correlations in a large corpus of documents could make use of a smaller model to analyze each of the documents to filter out the ones least likely to be useful and only actually "read" the most promising ones. And so forth.

This is something that's an area of very active development. The harness is going to be just as important as the model, if not moreso.

Ah, I see where my misunderstanding lies, I had the wrong "they're."

That would make them useless for cheating.

If that was the case, why is it useful for cheating?

The word "slop" has lost all meaning.

The older the jpeg the more likely it is that he's moved on to another job by now.

They passed a law saying that, but the Kentucky constitution still says otherwise. So we'll see what happens.