German Tank Problem

en.wikipedia.org

39 points·stefankuehnel··7 comments

Interesting that they discuss how frequentist and Bayesian methods give (slightly) different answers, but not which one was closer to the historical example. Although, either one would be so much closer to the truth than the intelligence community's estimate that perhaps it doesn't matter.

Why does Frequentism and Bayesianism give slightly different answers? Could someone look through every logistics involving tanks in history and see whether the Bayesian or Frequentist approach is closer, thereby gaining evidence for which interpretation is the correct one?

Why do they disagree? Because they are estimates. Which one is correct? Neither and both: they are estimates.

Estimate presupposes estimating something in objective reality. Because the verb "to estimate" takes a direct object. Therefore if there are two estimates, and they differ, one is closer to objective reality than the other.

Your comment is asking which of two methods of estimating is correct.

Neither is correct, in that neither gives the actual objective truth. Both are correct, in that they both give you estimates that can only be incorrect if they give a zero probability to the actual objective truth.

Even statistically, assume the true value is 0. Is "2 ±5" or "1 ±7" the better estimate? Assume the methods used to derive them consistently yield similar estimates. Which one is the correct method?

> Neither is correct, in that neither gives the actual objective truth.

Then why call them estimations?

During the early years of Tesla, TSLA followers would post the VIN numbers of delivered cars to estimate the rate of production. They were accurate for a few quarters. After some time Tesla changed the practice of sequential VINs.

Reply on news.ycombinator.com