...about polls.The new YouGov poll of Iowa gives you a sense of how much models & weighting go into polls (& efforts to overcorrect undersampling). It looks like 1233 respondents said Turek and 800 Hinson, unweighted, and then that got weighted to 47/46 Turek top line. https://t.co/JJAz9oWeBE
— Taniel (@Taniel) October 14, 2026
The models are all based on history, but use that, and “science,” to predict the future. As Hume pointed out, that’s not prediction, it’s just expectation. And when the future that was predicted turns into the past without ever meeting expectations, like soothsayers of old, pollsters always a reason the prediction wasn’t wrong, the understanding of it was.
The model cannot fail, it can only be failed. Most of the time the model is right; it’s the people who misunderstood it who are wrong. And oh, how often they misunderstand.
The Delphic Oracle was successful for so long because it understood this. Its schtick was to be gnomic. Or it came out that because the Oracle was stoned out of its mind and didn’t know what it was saying. Opinions differ.
The future remains the place nobody’s ever seen, nor really knows anything about.
No comments:
Post a Comment