Technically speaking, the two questions about Cromwell birth are not equivalent : if you have a 90% lower bound and a 90% upper bound it does give you a 80% confidence interval, but not all 80% confidence intervals will have a 90% lower bound and a 90% upper bound. For example you give “now” for upper bound to your interval, and 80% lower bound, that’ll still give a 80% interval. That may be part of the reason behind higher error rate for 80% interval : there are many ways to build such an interval, so it’s easier to get mixed up.
Else, interesting article, but for some parts it underestimates social complexity of the task : for the planning fallacy, using the “outside view” is usually a good thing, but it’s very hard in a professional context to make use of it (neither my boss nor my customers are usually receptive to it, they ask “how long for that feature and that one and that one” and don’t care much about “last time we did a project of that complexity, we took twice as long as initially planned”). De-biasing yourself is very important, but sometimes it’s not enough, you’ve to debias others too, and that’s even harder...
In my experience debiasing others who have strongly held opinions is far more effort than it’s worth, a better road seems to be to facilitate them debiasing themselves.
Plant the seed and move on, coming back to assess and perhaps water it later on. I don’t try to cut down their tree… as it were.
Technically speaking, the two questions about Cromwell birth are not equivalent : if you have a 90% lower bound and a 90% upper bound it does give you a 80% confidence interval, but not all 80% confidence intervals will have a 90% lower bound and a 90% upper bound. For example you give “now” for upper bound to your interval, and 80% lower bound, that’ll still give a 80% interval. That may be part of the reason behind higher error rate for 80% interval : there are many ways to build such an interval, so it’s easier to get mixed up.
Else, interesting article, but for some parts it underestimates social complexity of the task : for the planning fallacy, using the “outside view” is usually a good thing, but it’s very hard in a professional context to make use of it (neither my boss nor my customers are usually receptive to it, they ask “how long for that feature and that one and that one” and don’t care much about “last time we did a project of that complexity, we took twice as long as initially planned”). De-biasing yourself is very important, but sometimes it’s not enough, you’ve to debias others too, and that’s even harder...
In my experience debiasing others who have strongly held opinions is far more effort than it’s worth, a better road seems to be to facilitate them debiasing themselves. Plant the seed and move on, coming back to assess and perhaps water it later on. I don’t try to cut down their tree… as it were.