4 ms·
If you've put enough thought into them to have a confidence level, why not give a specific probability? Then you will be able to see (to some extent) how you ar
by gwern 16y ago
If you've put enough thought into them to have a confidence level, why not give a specific probability? Then you will be able to see (to some extent) how you are poorly calibrated - overconfident, underconfident, etc.
For example, I have a ton of judged predictions on http://predictionbook.com/users/gwern http://predictionbook.com/users/gwern so you can see my problem - low self-esteem. (I expect to take some losses when all my 2010 predictions expire in a few days, but the overall trend of underconfidence will remain.)