Sparked by my recent interested in PredictionBook.com, I went back to take a look at Wrong Tomorrow, a prediction registry for pundits—but it’s down. And doesn’t seem to have been active recently.
I’ve emailed the address listed on the original OB ANN for WT, but while I’m waiting on that, does anyone know what happened to it?
I got a reply from Maciej Ceglowski today; apparently WT was taken down to free resources for another site. It’s back up, for now.
(I have to say, seriously going through prediction sites is kind of discouraging. The free ones all seem to be marginal and very unpopular, while the commercial ones aren’t usable in the long run and are too fragmented.)
In relation to these sorts of sites, what’s a normal level of success on this sort of thing for LW readers? If people chose ten things now that they thought were fifty percent likely to occur by the end of next week, would exactly five of them end up happening?
I don’t know of any LWers who have used PB enough to really have a solid level of normal. My own PB stats are badly distorted by all my janitorial work.
So, I couldn’t say. It would be nice if we were all calibrated. (But incidentally you can be perfectly calibrated and not have 5⁄10 of 50% items happen; it could just be a bad week for you.)
Sparked by my recent interested in PredictionBook.com, I went back to take a look at Wrong Tomorrow, a prediction registry for pundits—but it’s down. And doesn’t seem to have been active recently.
I’ve emailed the address listed on the original OB ANN for WT, but while I’m waiting on that, does anyone know what happened to it?
I got a reply from Maciej Ceglowski today; apparently WT was taken down to free resources for another site. It’s back up, for now.
(I have to say, seriously going through prediction sites is kind of discouraging. The free ones all seem to be marginal and very unpopular, while the commercial ones aren’t usable in the long run and are too fragmented.)
In relation to these sorts of sites, what’s a normal level of success on this sort of thing for LW readers? If people chose ten things now that they thought were fifty percent likely to occur by the end of next week, would exactly five of them end up happening?
I don’t know of any LWers who have used PB enough to really have a solid level of normal. My own PB stats are badly distorted by all my janitorial work.
I suspect not many LWers have put in the work for calibration; at least, I see very few scores posted at http://lesswrong.com/lw/1f8/test_your_calibration/
So, I couldn’t say. It would be nice if we were all calibrated. (But incidentally you can be perfectly calibrated and not have 5⁄10 of 50% items happen; it could just be a bad week for you.)