Lessons of 2019:
1) Correlation isn't close to what you think it actually is.
2) Most people don't know statistics let alone probability.
3) It's not that field is lacking earnest students. It's that the field is taught wrong all around.
4) A second rate statistician is deadly
1) Correlation isn't close to what you think it actually is.
2) Most people don't know statistics let alone probability.
3) It's not that field is lacking earnest students. It's that the field is taught wrong all around.
4) A second rate statistician is deadly
5) If you don't know stats, it's best to use your emotional intuition to disbelieve whatever the social scientist is pushing.
6) They take nature and compartmentalize it into mini models. The models are wrong.
7) They project the model back on to reality with confidence.
6) They take nature and compartmentalize it into mini models. The models are wrong.
7) They project the model back on to reality with confidence.
8) Reality laughs. The modellers keep on modelling.
9) Science is hard because it won't conform to your wishes, desires, equations, data sets, etc.
10) Data collected is a snapshot of the past. Not a predictor of the future. It almost always excludes quiet but important info.
9) Science is hard because it won't conform to your wishes, desires, equations, data sets, etc.
10) Data collected is a snapshot of the past. Not a predictor of the future. It almost always excludes quiet but important info.
11) Fat tail events start with single data points. Recall how Steve Jobs went to Intel and asked for a chip for the new iPhone at the time. They laughed because their MBAs said there was no data to back the potential market for it.
12) Since data can't be used directly to make calls one way or another, what should one do? Experiment and iterate with tight feedback loops tied to reality.
13) As time passes, assumptions get burned.
14) Don't buy into things that will have to change in time.
13) As time passes, assumptions get burned.
14) Don't buy into things that will have to change in time.
15) Buy into things that have been time tested with very few assumptions built-in.
16) If something seems stupid, it's because your assumptions about its fundamentals are wrong.
17) Correlation in fields that are normally distributed is spurious.
16) If something seems stupid, it's because your assumptions about its fundamentals are wrong.
17) Correlation in fields that are normally distributed is spurious.
18) If the field is fat tailed, correlation is insanely out of whack. It will cause you to misjudge the problem by a factor of 100-1000. Sometimes even more.
