I'm pretty sure that a betting market would give >1/1000 odds to visiting aliens explaining some UFOs. And at those odds the world should be spending much more than $22M/yr investigating UFO sightings.https://www.nytimes.com/2017/12/16/us/politics/unidentified-flying-object-navy.html …
-
-
One obvious approach is to wait 1000 years before setting the bets. I expect that at the actual market odds, I'd bet against it too. Even so, it seems a good general policy for policy to follow the market odds.
-
Can't argue with that.
-
could someone explain that reasoning? also how quantify the perceived sunk losses of science that turns up negative findings, especially in light of the serious problems with publication bias? do prediction markets potentially improve research funding decisions?
-
Yes prediction markets could be used to improve research funding decisions.
-
what kind of bets would be made? i.e. how to tackle the question of what research is valuable with respect to the beliefs/values lens? at least in my experience it's seems like there's a lot of conflation of the two in academia
-
Distinguish between values and beliefs. Read Robin's work on futarchy; http://mason.gmu.edu/~rhanson/futarchy.html …
-
I have, but I'm not sure what kind of betting would aggregate good information to steer science... if there's a market for whether or not an experiment will give a negative result, is it scientific to use that market to aggregate information for funding policy?
-
For any given proposed project, markets could tell you the chances of various outcomes, define in terms of specific results, or in terms of prestige of its products. Yes of course it is "scientific" to use best available info.
- 6 more replies
New conversation -
-
-
The idea that they're here is less preposterous than the idea that they're here *and random people sometimes see them but otherwise there is no evidence*
-
Okay, but events with odds over 1/1000 are often called "preposterous".
-
Equally, betting markets don't (in my experience) make accurate assessments of probability for these low-frequency events at the tails of the probability distribution. "Odds" are usually far, far likelier than the event. Seldom do you see anything worse than 1000/1
-
That actually depends on the taxes/fees and market trading mechanism. With low fees & good mechanism, you can get odds much lower than 1/1000.
-
@ctwardy Do you know what was the lowest probability for any SciCast event (no doubt a conjunction)? That is, a prob the system could and did represent? -
It's not hard to program SciCast or any other PM to allow arbitrarily many decimals. But it's very hard to establish a strong track record of accuracy with probabilities so small.
-
Saying that prediction markets have been proven to give bad values for low prob events is very different from saying that, by the nature of such probs, it is very hard to prove that prediction markets give good probs.
-
I don't actually know the track record that well...my point is just that you'd need enormous sample size to establish reliability at very low probabilities
- 2 more replies
New conversation -
-
-
It’s like you haven’t even seen Star Trek
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Well said: “You don't cross interstellar distances, hide, and then fly giant non-nanotechnological aircraft with their exterior lights lit where only sporadic witnesses can see you.”
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.