- Yglesias conceptualize the idea of uncertainty: challenge rating --- impossible
- We have no idea where Yglesias got his numbers but they’re wrong. This debate has for so long been plagued by numbers basically pulled out of thin air Right pic is from my and @adambonica.bsky.social’s response essay: www.bostonreview.net/forum/how-no...
- I'm not sure you should complain about Yglesias citing over-confident numbers and in the same breath talk about uncertainty that could just as well be explained by the shortcomings of your model
- Matt is mistaking a 1.5-2 _AME_ for the uncertainty in the response/misattributing the estimate to B/G instead of Elliott (+conjecture, I think he's also mistaking the 1.5-2 AME for voteshare, rather than margin). We can/should be honest about forecast uncertainty based on a dataset of <20 cyclesFeb 3, 2026 22:26
- Strongly agree re: forecasts and the conclusions in the screenshot, I'm just not sure what P(50) -> P(55) adds. Is that not just another stat which is uncertain to the point of being meaningless because models can't capture election dynamics?
