Daniel Bernoulli, Leonhard Euler, and maximum likelihood (Q4355995)

The short St. Petersburg paper of 1778 of Daniel Bernoulli is often cited as the first to formulate the method of maximum likelihood. The present paper sheds light on the understanding of the topic by Bernoulli, and by Leonhard Euler who appended his own commentary to Bernoulli's paper. This is done by reexamining the 1778 publication using an unpublished manuscript treatment by Daniel Bernoulli of the topic, dated 1769, and appended to the present paper (as pp. 358-367) in English translation. Two essentially distinct ideas are found to be contained in Bernoulli's attempts to form an estimate by combining observations, by two differing uses of a semicircular probability density. One use emphasises application as weight/value and results in a weighted average -- an early version of an \(M\)-estimator -- and in a commensurate early version of an algorithm to compute an \(M\)-estimator. The second use multiplies values of the density corresponding to observations and corresponds to the method of maximum likelihood for obtaining an estimate of true point. For the entire collection see [Zbl 0861.00032].