[SEQ RERUN] Futuristic Predictions as Consumable Goods
Today’s post, Futuristic Predictions as Consumable Goods was originally published on April 10, 2007. A summary (from the LW wiki):
The Friedman Unit is named after Thomas Friedman who called “the next six months” the critical period in Iraq eight times between 2003 and 2007. This is because future predictions are created and consumed in the now; they are used to create feelings of delicious goodness or delicious horror now, not provide useful future advice.
Discuss the post here (rather than in the comments of the original post).
This post is part of a series rerunning Eliezer Yudkowsky’s old posts so those interested can (re-)read and discuss them. The previous post was Inductive Bias, and you can use the sequence_reruns tag or rss feed to follow the rest of the series.
Sequence reruns are a community-driven effort. You can participate by re-reading the sequence post, discussing it, posting the next day’s sequence reruns post, summarizing forthcoming articles on the wiki, or creating exercises. Go here for more details, or to discuss the Sequence Reruns.
Judging both from experience and after reading the original comments for this post, it seems that many people share the misconception that betting—be it on ideas or sports, etc—is always done at odds of 1:1, ie I bet $100 that my prediction y will occur by date z and if i am wrong I lose $100 and if I am right I win $100 (plus my original wager). In fact, astute gambling is about predicting the likelihood of an event and then finding someone to bet with whom you think has done a worse job than you at predicting the likelihood of the event.
For instance, when I watch golf and a player has a long and difficult putt, I frequently hear the commentators say “I wouldn’t bet on this putt going in.” If betting was restricted to wagers at 1 to 1 odds then they would be correct not to bet on the putt dropping, as it is less than 50% likely that it will. But what if I gave the commentator 1000 to 1 odds on the putt going in—would they take that bet? They should. Similarly, I wouldn’t place a bet at 1 to 1 odds that there will be superhuman intelligence by the end of the year. But I’d certainly place this bet if someone would give me 100000 to 1 odds, and I’d consider placing this bet at much lower odds.
On a related note, if you ever see or hear someone make an outrageous prediction of certainty which they are willing to bet on, you are making a mistake by wagering with them at 1 to 1 odds on their prediction. You are allowing them to place a bet where they break even if their prediction is only 50% likely to occur, yet they are claiming 100% likelihood. You should push them to give you better on odds on the bet—as good as you can get from them since theoretically they should think they will never lose this bet. But be careful pushing too hard because the mark will realize that they aren’t actually certain about their prediction.
The key to take advantage of suckers of this kind (that only bet on 50⁄50 odds) is not to make them realize their mistake by changing the odds. This is precisely when they start to realize something is up. I’ve won two bets against friends in the past six months doing this (though in the first bet the friend would have probably done something at uneven odds).
On the other hand, I have another friend that makes lots of overconfident assertions and, when possible, I try to make even-odds bets with him, but he usually refuses with some excuse about why it wouldn’t be appropriate to bet. Even when I tell him to take advantage of my ignorance. I’m unsure how to crack nuts like this.
(For those of you who find the idea of taking advantage of your friends offensive, this practice also raises the sanity waterline—both your own sanity and your friends. So long as the stakes aren’t too high, I think it’s a great practice)
I am surprised that this post appeared on OB, and even had (independently) a comment about staking money on predictions and a comment written by Robin Hanson, without eliciting a mention of ideas futures. So I’ll go ahead: if you actually care about predicting the future—or just getting people not to listen to silly predictions—I suggest creating and consulting a market. If social and institutional barriers make this proposal infeasible or unhelpful, I suggest chipping away at them (or at least thinking about how effectively you could chip away at them).
I would suggest just plain writing down your predictions and keep track, or even just predicting on an existing corpus. Waiting for prediction markets is letting the perfect be the enemy of the better.
I believe it was Cheerios.
I have the same thing for breakfast every day. I just have to know what I had for breakfast in 2006.
But yeah, that’s completely missing the point.