Celebratio Mathematica

David H. Blackwell

Statistics  ·  UC Berkeley

A Tribute to David Blackwell

by Juliet Shaffer

With the death of Dav­id Black­well, fol­low­ing the death last year of Erich Lehmann, both at ninety-one years old, the seni­or level of stat­ist­i­cians at U.C. Berke­ley is gone.

Erich and I were reas­on­ably close to Dav­id. For many years we drove him to the joint Berke­ley–Stan­ford col­loquia when they were held at Stan­ford, giv­ing us time for much con­ver­sa­tion.

Erich told me that, when Dav­id first came to Berke­ley, his large fam­ily could not find any place that would ac­com­mod­ate them, and they lived for some months camp­ing out in a park. In Dav­id’s in­ter­view in Stat­ist­ic­al Sci­ence in 1986, he stated that the early dis­crim­in­a­tion he faced (be­fore Berke­ley) “nev­er bothered me”. Since the early Berke­ley ex­per­i­ence wasn’t men­tioned, it’s not clear that he was sim­il­arly un­bothered by this early Berke­ley dis­crim­in­a­tion. I know that he was keenly aware of such is­sues later.

I talked with him from time to time about dis­crim­in­a­tion, men­tion­ing things that had happened to me both as a wo­man and a Jew. I had the im­pres­sion that he thought a lot about dis­crim­in­a­tion in gen­er­al (not ne­ces­sar­ily against him per­son­ally) in later years. I un­der­stand his ini­tial un­con­cern very well. When I was first look­ing for jobs there were many ads, at least half, that stated “men only”. It didn’t both­er me then — it was the way things were. Only later with the rise of fem­in­ism did I be­gin to see things dif­fer­ently. Dav­id was usu­ally very un­ruffled, but I saw a rare strong re­ac­tion when I told him how the Geor­gia flag, which re­sembles the Con­fed­er­ate flag, bothered me as I saw it fly­ing over a hotel in which I had just stayed in At­lanta. He men­tioned that a white beg­gar on Tele­graph Av­en­ue had ap­proached him for money while wear­ing a Con­fed­er­ate flag cos­tume and how angry he had felt about that.

It must have been some­what dif­fi­cult be­ing a Bayesian in the strongly-non-Bayesian Berke­ley Stat­ist­ics De­part­ment. I once men­tioned to Dav­id that I was not very sym­path­et­ic to the Bayes ap­proach but did have some in­terest in em­pir­ic­al Bayes. He noted that he didn’t be­lieve in em­pir­ic­al Bayes and showed that it didn’t make sense when ap­plied to a single in­fer­ence. I re­marked that it made sense in the con­text of a large num­ber of sim­il­ar in­fer­ences, to which there was no reply. I in­ter­preted his re­ac­tion as il­lus­trat­ive of an as­pect of his ap­proach to stat­ist­ics. He liked el­eg­ance and sim­pli­city. Is­sues had to be clear in the very simplest of situ­ations. Em­pir­ic­al Bayes didn’t meet this test. He felt a Bayesian pri­or was ne­ces­sary. His abil­ity to cla­ri­fy is­sues in simple and el­eg­ant ways was pre­sum­ably what made him such an out­stand­ing lec­turer and teach­er.

Dav­id was a kind and won­der­ful per­son, but he was also a very private per­son, and there was al­ways the feel­ing of an in­ner core that couldn’t be pen­et­rated. I urged him many times to write his mem­oirs, but he nev­er did.