That said I question both your and grandparents choice to look at only people who chose to vote. 68% of people who could have voted against this did not - either actively supporting it with a vote or deciding that they didn't give a fuck if it happened by not voting. Every potential voter either knew or ought to have known the result of this election result...

Destroying sources of truth is a long standing goal for those who want people to act against their own interest, and scientists are a source of truth. It's shrinking the pie so that the republican leadership can capture a much bigger slice of it.

Nine PBS might well have non-digital data, but it's clearly not the data this article is about.

By non-deterministic I think people really mean "chaotic" in the chaos theory sense. Small perturbations in the input lead to wild and unpredictable changes in the output. Even with temperature parameters a fixed PRNG seed could mean an LLM was just chaotic and not technically non-deterministic.

But more literally while LLMs are in theory deterministic (though perhaps not inference providers implementations if there's anything like a race condition affecting how things are rounded when added together) - we use the LLMs in harnesses that aren't. There are very likely races in the terminal outputs, dates both intentionally put in the context and accidentally leaked to the context, things like that.

It's close enough for most purposes... but it isn't.

https://en.wikipedia.org/wiki/Foreshock

But getting that data seems... practically impossible at our current tech level.

It was done on people suffering from asthma - with all the reason in the world to improve air quality. Replacing gas stoves with electric stoves a huge effect, on the other hand "installation of a ventilation hood over an existing gas stove in 15 other homes did not significantly change NO2 levels".

Article: https://theconversation.com/replacing-gas-stoves-with-electr...

Underling Study: https://www.jaci-inpractice.org/article/S2213-2198(26)00537-...

Admittedly I have other complaints about the microwave controls (why is it so damn many button presses to vary the power output) but it's not because the buttons are hard to use.

I can imagine schemes for instance where context is compressed into chunks and then chunks that are ranked highly relevant for the token are decompressed. Which would sort of be between a long context and a memory retrieval scheme...

Moreover you're picking a date hundreds of years into the eastern empire's existence and declaring "because they switched official languages hundreds of years later, they were always Greek". Time doesn't flow like that.

Moreover the roman empire, pre-split, spoke quite a bit of this language (well, Koine Greek, the predecessor to Romaic). Including in official communications (particularly in the east), and at least one emperor's private diaries [1].

[1] https://en.wikipedia.org/wiki/Meditations

Eh, inheritance implies the predecessor willed it to you upon death. In the case of the "Holy Roman Empire" the predecessor died 334 years earlier with no successor. This isn't a case of inheritance, it's a case of just making shit up.

The "byzantine" empire did inherit the name though. They continuously used and claimed the name "roman" from when the roman empire split (and called both halves roman) right up until 1453, even their language was called romaic. That we don't refer to them as roman to this day is really a strange success of propaganda.

> And the Byzantine Empire shrunk greatly after the first few hundred years.

Which by itself sort of contradicts the claim that empires historically fall quickly? The roman empire began in 27 and in 565 it controlled almost the entire Mediterranean. Sure a few hundred years after the split (which itself is a few hundred years after the beginning of being an empire) it shrank but that's a long time.