Nate Silver has had an up and down career since he came to fame forecasting the 2008 election. He went from the New York Times to Disney, which tried to make him a personality at ESPN and then transferred his website fivethirtyeight.com to ABC News, which never wanted it. The site remains a go-to for political sanity. I assume everybody who posts here goes there at least occasionally.
But maybe not for much longer.
Silver plans to leave as soon as his contract is up, apparently really soon, after Disney laid off huge numbers of staff yesterday. Disney is busy cutting 7000 positions. Why, since it has all the money in the world, I don’t know. Hopefully they need more money to fight DeSantis, maybe turn Tallahassee into a theme park.
Anyway the site as we knew it is probably gone despite ABC’s bland words of reassurance. Silver has been talking to people about where he’ll land. Watch this space.
Frankly, FiveThirtyEight has been on a downward slide since ABC specifically targeted Clare Malone to be cut, and I’m kind of relieved to not have to hear any more inane discussion about “Good use of polling or bad use of polling?” I’ve mostly supplanted my commute podcast politics listening to The Lawfare Podcast and Behind the Bastards anyway, and I can only take so much of that before switching to The Joy of Y, Quanta Science or Old Gods of Appalachia anyway.
Streaming services in general seem to be tightening their belts in the aftermath of the pandemic boom. They made shedloads of money from home-viewing subscribers during lockdowns, but not so much nowadays.
I haven’t listened to 538 as much as I used to. Definitely looking for other, better podcasts at this point. Will be interesting to see what Nate does. Too bad ABC owns it.
The “prediction models” (which I believe are just SAS analytics code) have to be rebuilt for every election cycle anyway. Having a bunch of legacy models isn’t of any real benefit except for reviewing the results of past elections.
I think the point is that Disney doesn’t think that FiveThirtyEight has any real value (in a financial sense), and frankly, they’re probably correct that it doesn’t bring in a huge audience or add revenue streams. Next cycle, they’ll probably just prompt ChatGPT for who it thinks will be elected and copypaste the results even if the response is “Scrooge McDiggler”.
ChatGPT’s training data set includes both Disney+ and XXX Porn .com, so of course it produces a syncretism. Now try getting that image out of your head.
Disney gross profit for the twelve months ending December 31, 2022 was $28.195B, a 12.49% increase year-over-year.
Disney has all the money in the world.
After the layoffs it will have even more. Super large companies tend to buy or form fifty small companies in the hopes that 5 or 10 will pan out. In the long term they cut them out and start another fifty. I constantly go back and forth on whether that’s smart business or criminally negligent.
One in twenty times you’re gonna get that wrong. 95% is hardly 100% confidence. If you always get your 95% winners right, your model is dogshit, too. And the final forecast was Trump at 29% of winning the electoral college, which was better than every other polling model. On November 3, they had Trump at 35% (!!!). Actually, I can’t see them predicting Hillary at any more that 90% at any point during the prediction cycle:
So I’m not sure what you’re remembering. My memory is 538 actually made me more nervous about Hillary’s chances than any other news source. And they were right to.
Also, remember Comey on Oct 28 made that announcement about the email probe and Clinton that also sank her chances abruptly.
I thought it through very carefully at the time, the meaning of such a model. And my conclusion was that it wasn’t all that valuable. If a 75% means “tossup,” then call it that and leave off the exact percentage. If 95% means, “Gonna win! But anything could change this so that it’s meaningless,” then maybe don’t ever call anything “95%.” Because that is what it was like: Hillary at this extremely high percentage, whoops something happened so now it’s a 2 in 3 chance, whoops Trump won! Haha, but our model is just the shit–hey, we gave Trump a 1 in 3 chance, so–cool?!
And, quite frankly, Nate comes off as this chirpy, kinda know-it-all twat, so I don’t have too many feels for him as he wanders away from Disney to wherever.
Actual, Silver comes off as an overly qualifying, impossbile-to-pin-down ‘twat’ who hedges his assessments based upon conditional validity of the data. I’ve literally never heard him claim that a result was essential certain because of projections, because like a good statistician he always references the margins even to an absurd degree. But maybe you’ll provide a clip or quote where he makes some unlikely absolute projection about the future that I’ve never, ever heard him make. Yes?
I don’t know. I think a model with numbers helps me understand the nuance: 95% and 75% and 51% all mean “tossup,” but those numbers show me how likely things are to change. It’s just if your tendency is to round 95% to 100% that you go wrong. There’s no certainty in prediction, but their model seems pretty solid.