I'm drafting this comment before reading the whole article based only on the quote from St. Augustine. This is one of my absolute favorites.
From his exegesis of Genesis:
Quapropter bono christiano, sive mathematici, sive quilibet impie divinantium, maxime dicentes vera, cavendi sunt, ne consortio daemoniorum animam deceptam, pacto quodam societatis irretiant.
A fairer translation of "mathematici" here would be "astrologer" or "numerologist." But keeping it as "mathematician" is so much more fun. I can attest that it makes for a rousing introduction to a Sunday school lesson.
I don't think I am following the central point of these series of posts. How can defensive forecasting simply do some bookkeeping to correct past mistakes thereby making one's forecasts over time look like good forecasts by any pre-chosen metric. What if the thing being forecasted is hurricane intensity, and the metric being minimized is the average number of windows damaged once the forecast is used to decide whether windows need to be boarded up. Is all this simply a way to get "calibration" without any skill? Any clarifications or intuitions will be much appreciated.
I'm drafting this comment before reading the whole article based only on the quote from St. Augustine. This is one of my absolute favorites.
From his exegesis of Genesis:
Quapropter bono christiano, sive mathematici, sive quilibet impie divinantium, maxime dicentes vera, cavendi sunt, ne consortio daemoniorum animam deceptam, pacto quodam societatis irretiant.
A fairer translation of "mathematici" here would be "astrologer" or "numerologist." But keeping it as "mathematician" is so much more fun. I can attest that it makes for a rousing introduction to a Sunday school lesson.
wait so being calibrated doesn't mean you understood anything
"The philosophers have only interpreted the world, in various ways; the point is to change it."
You might like the book by Romain Brette "The Brain, In Theory". He makes a lot of good points about this idea and predictive coding in general.
I don't think I am following the central point of these series of posts. How can defensive forecasting simply do some bookkeeping to correct past mistakes thereby making one's forecasts over time look like good forecasts by any pre-chosen metric. What if the thing being forecasted is hurricane intensity, and the metric being minimized is the average number of windows damaged once the forecast is used to decide whether windows need to be boarded up. Is all this simply a way to get "calibration" without any skill? Any clarifications or intuitions will be much appreciated.
a quick read but a great one .... !!!!