The most important quality for a rationalist is to admit that you were wrong and change your mind accordingly: so I will say, as an excercise in strength and calibration, that I was totally wrong.
I thought, with a high degree of probability, that Clinton was going to be the next POTUS. Instead it’s Trump. My model of the world was wrong, and I’ll adjust accordingly.
I, for one, called it months ago. And found myself unable to tell all but 2 people in my social circle my conclusions because if I did they seemingly mistook my assessment of the likely outcome for endorsement and reacted quite unplesantly with incredulity and offense especially whenever I tried to tell them they were living in an echo chamber. Thus I quickly learned not to try.
Interestingly the people who actually listened were an immigrant from Armenia and an expat rather than domestic American citizens, I wonder if that means anything.
(‘Called it’ meaning calling that true support was higher than most thought especially in the greater rust belt plus extremely low enthusiasm for the other party leading to high chances of an electoral victory and that most of what most people around me thought was weakening the campaign was actually strengthening it or completely neutral.)
Eh, elections seem hard to update on though. Before the election, I thought Clinton was 70% likely to win or so, because that’s what Nate Silver said. Then Trump won. Was I wrong? Maybe, but it’s not statistically significant at even p = 0.05.
So just looking at U.S. presidential elections, you’ll never have enough data to see if you’re calibrated or not. I guess you can seriously geek out on politics, and follow and make predictions for lots of local and foreign elections also. At that point, it’s a serious hobby though, I’m much more of a casual.
No intentions on leaning heavily on US politics, since I’ve already hairy Italian politics that is more relevant to me... I’ll just change a couple of parameters in my model of US, as per answer to Gunnar.
Personally, I think the update most people should be making is the one getting the least attention. That even a 30% chance means 3 out of 10 times. Things far more unlikely than 3 out of 10 happen every day. But because we assign such importance to the election, we assign a much greater confidence to our predictions, even when we know we’re not that confident.
Except that the chances weren’t 30%. That was a number generated by Nate Silver based on polling methodology that was not calibrated to the reality on the ground. I think you can find much deeper lessons here than that, especially given it seems to be a repeat of the Brexit phenomenon. Fool me once, fool me twice...
In two directions: on one side, I thought that the proportion of progressives in US was larger than it has shown to be; on the other, I had a tenet that “things that menace status quo don’t happen”, so I’ll lower the probability on that and change what the current status quo is.
I literally wrote a post to my buds holding forth on why the race went as it did. It included what we (conservatives) had gained, and what they (liberals) had done, and what that implied. Long and detailed.
But to look like a smarty pants I wrote it yesterday. And I wrote it assuming that Clinton had won.
You are far from alone, MrMind.
Looking back, there were things that should have clued me in. Most notably, when people were scoffing at a “phantom” Trump vote I was shifting uneasily in my chair, because three of my liberal buds had confided in me that they’d be voting Trump this year, but couldn’t admit it because their social circle viewed that as a mortal sin.
I told myself that anecdote != data, that websites can’t be wrong.
I too was wrong. I gave him a 45% chance on this site several months ago and my estimate had hardly changed by yesterday (in fact my estimate got slightly worse, down to 40%.)
Part of the reason I estimated the chance being that high was because I thought (at that time) we were fairly likely to have a recession or major terror attack, which would swing the election to Trump. Neither of those happened, but Trump still won. More recently, II did think the big media company polls were systemically biased by at least a few points in Clinton’s favor, so I give myself some credit for that.
Are you really that wrong though, if you gave him a 40-45% chance? Am I making an error to say that based on the real results, someone who was 60%+ sure of a Trump win is more wrong than you are?
If I truly believed 538, and that is what I told myself, I shouldn’t have been surprised – and yet I was. So what is happening here? I’m not usually surprised by anything that I assign a probability of greater than 10% of happening, why do I feel the way I feel? Perhaps my true probability was <10%.
Are you tracking your calibration with something like prediction book? You may be generally calibrated And this could have just been an instance of a low probability event happening
Also next time I’ll use log-odds instead of percentage: having 538 giving 60% to Hillary is psychologically different from saying 0.4 evidence for her.
The most important quality for a rationalist is to admit that you were wrong and change your mind accordingly: so I will say, as an excercise in strength and calibration, that I was totally wrong.
I thought, with a high degree of probability, that Clinton was going to be the next POTUS. Instead it’s Trump. My model of the world was wrong, and I’ll adjust accordingly.
I, for one, called it months ago. And found myself unable to tell all but 2 people in my social circle my conclusions because if I did they seemingly mistook my assessment of the likely outcome for endorsement and reacted quite unplesantly with incredulity and offense especially whenever I tried to tell them they were living in an echo chamber. Thus I quickly learned not to try.
Interestingly the people who actually listened were an immigrant from Armenia and an expat rather than domestic American citizens, I wonder if that means anything.
(‘Called it’ meaning calling that true support was higher than most thought especially in the greater rust belt plus extremely low enthusiasm for the other party leading to high chances of an electoral victory and that most of what most people around me thought was weakening the campaign was actually strengthening it or completely neutral.)
Eh, elections seem hard to update on though. Before the election, I thought Clinton was 70% likely to win or so, because that’s what Nate Silver said. Then Trump won. Was I wrong? Maybe, but it’s not statistically significant at even p = 0.05.
So just looking at U.S. presidential elections, you’ll never have enough data to see if you’re calibrated or not. I guess you can seriously geek out on politics, and follow and make predictions for lots of local and foreign elections also. At that point, it’s a serious hobby though, I’m much more of a casual.
No intentions on leaning heavily on US politics, since I’ve already hairy Italian politics that is more relevant to me...
I’ll just change a couple of parameters in my model of US, as per answer to Gunnar.
I’m curious how in particular you want to update.
Personally, I think the update most people should be making is the one getting the least attention. That even a 30% chance means 3 out of 10 times. Things far more unlikely than 3 out of 10 happen every day. But because we assign such importance to the election, we assign a much greater confidence to our predictions, even when we know we’re not that confident.
Except that the chances weren’t 30%. That was a number generated by Nate Silver based on polling methodology that was not calibrated to the reality on the ground. I think you can find much deeper lessons here than that, especially given it seems to be a repeat of the Brexit phenomenon. Fool me once, fool me twice...
In two directions: on one side, I thought that the proportion of progressives in US was larger than it has shown to be; on the other, I had a tenet that “things that menace status quo don’t happen”, so I’ll lower the probability on that and change what the current status quo is.
Not all progressives would vote for Hillary (even over Trump).
I literally wrote a post to my buds holding forth on why the race went as it did. It included what we (conservatives) had gained, and what they (liberals) had done, and what that implied. Long and detailed.
But to look like a smarty pants I wrote it yesterday. And I wrote it assuming that Clinton had won.
You are far from alone, MrMind.
Looking back, there were things that should have clued me in. Most notably, when people were scoffing at a “phantom” Trump vote I was shifting uneasily in my chair, because three of my liberal buds had confided in me that they’d be voting Trump this year, but couldn’t admit it because their social circle viewed that as a mortal sin.
I told myself that anecdote != data, that websites can’t be wrong.
Brexit was the biggest clue. The Trumpocalypse is basically, just a repeat of the Brexit voting patterns.
I too was wrong. I gave him a 45% chance on this site several months ago and my estimate had hardly changed by yesterday (in fact my estimate got slightly worse, down to 40%.)
That’s actually better than most (all?) pollsters including Nate Sliver.
Part of the reason I estimated the chance being that high was because I thought (at that time) we were fairly likely to have a recession or major terror attack, which would swing the election to Trump. Neither of those happened, but Trump still won. More recently, II did think the big media company polls were systemically biased by at least a few points in Clinton’s favor, so I give myself some credit for that.
Are you really that wrong though, if you gave him a 40-45% chance? Am I making an error to say that based on the real results, someone who was 60%+ sure of a Trump win is more wrong than you are?
If I truly believed 538, and that is what I told myself, I shouldn’t have been surprised – and yet I was. So what is happening here? I’m not usually surprised by anything that I assign a probability of greater than 10% of happening, why do I feel the way I feel? Perhaps my true probability was <10%.
Maybe the surprise is more related to your emotions regarding the outcome than to the objective numbers you assigned.
Are you tracking your calibration with something like prediction book? You may be generally calibrated And this could have just been an instance of a low probability event happening
Normally I don’t, but that’s a good idea.
Also next time I’ll use log-odds instead of percentage: having 538 giving 60% to Hillary is psychologically different from saying 0.4 evidence for her.