The forecast critique is correct in my view, but I think election forecasts problems are deeper than this. Probabilities are calculated on sets, sets have certain rules (zf theory) that make them coherent. It’s not clear to me that “presidential elections” can be meaningfully represented with sets that are coherent enough to do analysis on. Let’s alone machine learning.
The reliability of election forecasting models has less to do with the number of data points, and more to do with a changing environment. Yes, we only have presidential elections every 4 years but we have many polls in those years and many other kinds of elections! And since an election is a binary choice the model can be trained on less data and still be robust.
What's more relevant for election model precision is that the population you are predicting from is constantly changing. The people who respond to polls today are vastly different than even 4 years ago, and different again than 8 years ago. You can't build models using historical data if the history is so different from the present.
Can 2 and 3 be reconciled in practice? If you want public support for universities, encouraging the most politically radical students to shout as loudly as possible is likely not a good idea. Roth may be able to get away with that at Wesleyan (though I will be very curious what happens at Wesleyan when inevitably the angry protesters don't get their way with their Board), but that seems like a very bad idea for the vast majority of schools out there that depend on public support.
It's one thing to support free speech and academic freedom; it's another to pooh-pooh workforce development, which is the base for much of the public support of universities in the first place and seems to have a more direct correlation with national preparedness than most academic research.
The American Political Science Association? Sounds like a real riot. Just kidding I'm sure they are a fine association. Actually it sounds like fun, an insider look at where the political winds are blowing.
"A decade ago, the United States produced by far the most highly cited scientific papers in the world. Today, China does. In 2022, for the first time, China’s contributions surpassed those of the United States in the closely watched Nature Index, which tracks 82 premier science journals"
Since China's population is about four times that of the U.S., it's not terribly surprising that Chinese researchers can turn out more high quality scientific papers than the U.S. can. So many "thoughtful" Americans have the totally unscientific notion that U.S. leadership in the sciences was due to either an innate superiority to other nations or a greater amount of effort. Instead, we were simply larger, richer, and more open. In Europe, everything "important" was, and to a great extent still is, driven through a handful of elite institutions. Since China is so much larger than we are, and since careers in the sciences in China offer both prestige and (usually) security, they are extremely attractive options. People in the U.S. fitfully recognize that, "someday", the weight of world civilization will shift to "the East", but they tacitly assume that won't be until after they're dead and, furthermore, won't have any meaningful impact on world affairs during their lifetime. They're wrong.
Thanks, all three have strong messages.
The forecast critique is correct in my view, but I think election forecasts problems are deeper than this. Probabilities are calculated on sets, sets have certain rules (zf theory) that make them coherent. It’s not clear to me that “presidential elections” can be meaningfully represented with sets that are coherent enough to do analysis on. Let’s alone machine learning.
The reliability of election forecasting models has less to do with the number of data points, and more to do with a changing environment. Yes, we only have presidential elections every 4 years but we have many polls in those years and many other kinds of elections! And since an election is a binary choice the model can be trained on less data and still be robust.
What's more relevant for election model precision is that the population you are predicting from is constantly changing. The people who respond to polls today are vastly different than even 4 years ago, and different again than 8 years ago. You can't build models using historical data if the history is so different from the present.
Can 2 and 3 be reconciled in practice? If you want public support for universities, encouraging the most politically radical students to shout as loudly as possible is likely not a good idea. Roth may be able to get away with that at Wesleyan (though I will be very curious what happens at Wesleyan when inevitably the angry protesters don't get their way with their Board), but that seems like a very bad idea for the vast majority of schools out there that depend on public support.
It's one thing to support free speech and academic freedom; it's another to pooh-pooh workforce development, which is the base for much of the public support of universities in the first place and seems to have a more direct correlation with national preparedness than most academic research.
The American Political Science Association? Sounds like a real riot. Just kidding I'm sure they are a fine association. Actually it sounds like fun, an insider look at where the political winds are blowing.
"A decade ago, the United States produced by far the most highly cited scientific papers in the world. Today, China does. In 2022, for the first time, China’s contributions surpassed those of the United States in the closely watched Nature Index, which tracks 82 premier science journals"
Since China's population is about four times that of the U.S., it's not terribly surprising that Chinese researchers can turn out more high quality scientific papers than the U.S. can. So many "thoughtful" Americans have the totally unscientific notion that U.S. leadership in the sciences was due to either an innate superiority to other nations or a greater amount of effort. Instead, we were simply larger, richer, and more open. In Europe, everything "important" was, and to a great extent still is, driven through a handful of elite institutions. Since China is so much larger than we are, and since careers in the sciences in China offer both prestige and (usually) security, they are extremely attractive options. People in the U.S. fitfully recognize that, "someday", the weight of world civilization will shift to "the East", but they tacitly assume that won't be until after they're dead and, furthermore, won't have any meaningful impact on world affairs during their lifetime. They're wrong.