http://www.weather.com/weather/tenday/UKXX0085 See for yourself. Regardless, these aren’t particularly open questions. I would trust those people who have spent years developing fairly good models to have a better answer than I do.
My point is that while this is non-mindkilling and an open question, there isn’t much productive discussion you can have about it. Forecasters can slightly beat the averages when talking 6 months out, but they do that through the use of sophisticated computer models, not rational discussion. Without spending months or years learning how exactly those models worked and developing improvements, I don’t expect to be able to come up with a better answer than the current accepted one. So this really isn’t an open question of the class we’re interested in.
I’m not sure whether you and Yvain are using “open question” the same way here. I think Yvain is just using it to mean “we don’t know either way”, not in the “not figured out yet but we want a real solution” sense of open problem.
So what are the important questions that average (or somewhat-above-average) people will likely agree are complicated open questions where both sides have good points?
I don’t really see sides here. It’s more “the forecast says x%”. So while reasonable people will admit to probability estimates other than 0 or 100%, that’s because of the format the information is presented in.
So, you very much expect it to rain some time next week, but specifically not next thursday. Is that because you have storm clouds coming in? ’cause if they’re independent, that math doesn’t work out at all.
Will it rain next Thursday?
Where I am, I would feel comfortable assigning that <10% probability.
Will it rain in London next Thursday?
(11-15 rainy days per month)
http://www.weather.com/weather/tenday/UKXX0085 See for yourself. Regardless, these aren’t particularly open questions. I would trust those people who have spent years developing fairly good models to have a better answer than I do.
Will it rain in London six months from today? (Perhaps I’m planning an outdoor wedding.)
I don’t believe anyone has a way to do better predicting six months out than just looking at historical rain/shine rates.
My point is that while this is non-mindkilling and an open question, there isn’t much productive discussion you can have about it. Forecasters can slightly beat the averages when talking 6 months out, but they do that through the use of sophisticated computer models, not rational discussion. Without spending months or years learning how exactly those models worked and developing improvements, I don’t expect to be able to come up with a better answer than the current accepted one. So this really isn’t an open question of the class we’re interested in.
I’m not sure whether you and Yvain are using “open question” the same way here. I think Yvain is just using it to mean “we don’t know either way”, not in the “not figured out yet but we want a real solution” sense of open problem.
Yes, but open problems that we won’t come up with better answers than the answers already out there are not terribly useful to discuss.
But the point is not to discuss them, it’s to show people that they should not assert 100% probabilities.
I don’t really see sides here. It’s more “the forecast says x%”. So while reasonable people will admit to probability estimates other than 0 or 100%, that’s because of the format the information is presented in.
Substitute “next week” or “next month” as appropriate.
Next week: 90%+ Next month: See next week.
So, you very much expect it to rain some time next week, but specifically not next thursday. Is that because you have storm clouds coming in? ’cause if they’re independent, that math doesn’t work out at all.
probability of 7 days of not-rain
0.9 ^ 7 > 0.1
Indeed, am am expecting no rain on Thursday, but it’s raining now. Specifically on Thursday, we will have a dry air mass moving in.
That’s a good one—it also allows for very easy demonstration of updating ones belief on new information (ie., was the day comes closer).