the gears to ascenscion's Shortform

the gears to ascension

This is a special post for quick takes by the gears to ascension. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.

I continue to find sources with military background, who present news as relatively emotion-free overviews of strategic situation, are some of the best sources of news I'm aware of. That's not to say they're not biased, only to say they give a higher resolution potentially-biased picture, and that several high resolution pictures from different angles are better than low resolution pictures. I was just listening to one of Perun's presentations again, which had this nice slide:

I also continue to find Belle of the Ranch to be good at finding news that I wouldn't otherwise find and which turns out to be relevant later. There have been a number of noticeable successes where they called something important by tracking moving parts in more detail than other sources I'm aware of.

Would appreciate more, especially ones with different group affiliations than these.

3lilkim20259mo

Methodology: I'm not a fan of political talking heads, so I'm essentially operating by entering the handles of influencers and bloggers I've heard of alongside keywords like "successfully predicted" into search engines. Results: - Michael Knights (a foreign policy writer) predicted the inability of the U.S. campaign in Yemen to subdue the Houthis within a reasonable timeframe, arguing that they could continue to put pressure on the U.S. indefinitely. - George Behizy (a fairly standard Conservative Twitter/X influencer) predicted that Maduro would be abandoned by his political allies and removed from power in a covert operation within a matter of months back in November. - Andreas Krieg (another foreign policy analyst, some appearances on CBC) predicted the lack of major Iranian retaliation after Operation Midnight Hammer, in favor of pursuing a face-saving offramp. I have no special love or dislike for any of the four people whose predictions we've mentioned. Speaking objectively, all of these claims are about on par with each other in terms of impressiveness. They saw a pattern and predicted that the pattern would continue. I should note that the bar is much lower than this, of course. If you randomly guess at geopolitical outcomes for a year, and your target audience knows very well that you share their political tribe, then, by the end of the year, you will have an audience that remembers all of your correct predictions and has forgotten all of your incorrect ones. Qualitatively, I've observed sentiments (and justification) identical to the ones you expressed whenever I have talked to anyone about a political influencer they're a fan of. [...] A google search returned no results for me. What does this mean? ---------------------------------------- As a side note, "Belle of the Ranch", aside from a bunch of Facebook clickbait titles, has quite a few titles that indicate predictions that most non-partisan Americans would say did not come to pass. I'm

2the gears to ascension9mo

Boring levels of predictively calibrated, the quotes were scare quotes. I'll define it as, Alice having well enough calibrated that, in the presence of actually strong forecaster Bob, Alice's words have consistent use as information positively correlated with truth and with few inversions such that Bob finds them useful to be aware of, even if Alice's words are not themselves immediately good calibrated numeric predictions every time. I agree with your proposed heuristic, though it seems too weak to me on the one hand, and on the other, I'd note that technically someone could be more useful than a rock while failing the standard I'm setting out. Might be useful to build a tool that collates and checks all predictions someone makes and then scores them. Would cost a bit but shouldn't be too hard otherwise. Titles are not where you will find the actual predictions I find useful. I still expect that a tool which collects all predictions from someone and rates them on accuracy would show her to be doing reasonably well at avoiding saying "x will happen" and then x doesn't happen; usually she says "the possibilities are x, y, z, but not w, everyone else is saying w, that's impossible because i, j, k". This is much stronger than just "she makes positive predictions", it's lack of failed predictions. Now, I might be misremembering via the sort of process you're describing, and if that were demonstrated, my interest in checking what she's said would suddenly drop to near zero. I opened the video you referenced, here's the transcript; she was talking about it getting a byrd bath, which did occur, it's mentioned in the wiki page. Other sources that are this level of boringly calibrated, doing at least as well as understanding the basic rules at play such that an adversarially selected title reveals an analysis that, by the time events play out, their opponents agree is what happened, would be ones I'm interested in. I am not defending her title as reasonable, I think a re

though it seems too weak to me on the one hand

It is; I prefer a sensible lower bound on the basis that, once the better-than-a-rock bound is satisfied, there's room for complex reasons to like someone who's more accurate on one front than another.

Might be useful to build a tool that collates and checks all predictions someone makes and then scores them. Would cost a bit but shouldn't be too hard otherwise.

I agree, it's definitely tractable to build something like that with a current-gen LLM. Grab future predictions that can be unambiguously verified ("The ... (read more)

the gears to ascenscion's Shortform

10

the gears to ascenscion's Shortform

10