12.07.2015 Views

1 Studies in the History of Statistics and Probability ... - Sheynin, Oscar

1 Studies in the History of Statistics and Probability ... - Sheynin, Oscar

1 Studies in the History of Statistics and Probability ... - Sheynin, Oscar

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

He concludes that s<strong>in</strong>ce <strong>the</strong> action <strong>of</strong> <strong>the</strong> great moral pr<strong>in</strong>ciples isconstant, <strong>and</strong>, as <strong>the</strong> CLT teaches us, <strong>the</strong>y will <strong>in</strong> any case prevail overr<strong>and</strong>omness, it is better to keep to <strong>the</strong>m, o<strong>the</strong>rwise you will experiencebad times. That conclusion is really commendable, but from <strong>the</strong>scientific viewpo<strong>in</strong>t it is obviously not better than convert<strong>in</strong>g <strong>the</strong>Ch<strong>in</strong>ese emperor to Christianity desired by Leibniz. At <strong>the</strong> end <strong>of</strong> <strong>the</strong>Essai (p. 123) we f<strong>in</strong>d <strong>the</strong> celebrated phrase:It is remarkable that a science that began by consider<strong>in</strong>g games <strong>of</strong>chance should itself be raised to <strong>the</strong> rank <strong>of</strong> <strong>the</strong> most importantsubjects <strong>of</strong> human knowledge.He means exactly those political applications <strong>of</strong> <strong>the</strong> <strong>the</strong>ory <strong>of</strong>probability.All <strong>the</strong> strangeness <strong>of</strong> metaphysics <strong>in</strong> <strong>the</strong> philosophical <strong>and</strong>emotional spheres notwithst<strong>and</strong><strong>in</strong>g, Laplace shows an amaz<strong>in</strong>g <strong>in</strong>sightwhen concretely apply<strong>in</strong>g <strong>the</strong> probability <strong>the</strong>ory. I have lookedthrough <strong>the</strong> ATP with a special aim, to f<strong>in</strong>d at least one wrong def<strong>in</strong>itestatement. It seemed that support<strong>in</strong>g myself with a hundred <strong>and</strong> fiftyyears dur<strong>in</strong>g which science has been s<strong>in</strong>ce develop<strong>in</strong>g <strong>and</strong> given suchstrangeness <strong>of</strong> his general philosophical views, it will not be difficultto f<strong>in</strong>d <strong>the</strong>re def<strong>in</strong>ite errors as well. Indeed, he considered somedubious problems on <strong>the</strong> probability <strong>of</strong> judicial decisions etc.It occurred, however, that it was not at all easy to f<strong>in</strong>d at least onewrong statement 9 . A great many applications that he considered can beseparated <strong>in</strong>to three parts:1. Obvious <strong>and</strong> absolutely unquestionable problems such as partialcensuses <strong>of</strong> population or <strong>the</strong> change <strong>of</strong> <strong>the</strong> frequency <strong>of</strong> male births <strong>in</strong>Paris due to foundl<strong>in</strong>gs.2. Treatment <strong>of</strong> <strong>the</strong> results <strong>of</strong> astronomical observations. It isdifficult to discuss those applications s<strong>in</strong>ce vast material ought to bestudied.3. Obviously dubious problems like <strong>the</strong> probabilities <strong>of</strong> judicialdecisions. Here, however, Laplace’s conclusions are so careful thatpurely scientific errors are simply impossible.There is noth<strong>in</strong>g to say here about <strong>the</strong> first group, but someth<strong>in</strong>g<strong>in</strong>structive can be noted concern<strong>in</strong>g <strong>the</strong> second one. There, Laplace (p.46) quotes <strong>the</strong> result <strong>of</strong> <strong>the</strong> treatment <strong>of</strong> observations: <strong>the</strong> ratio <strong>of</strong> <strong>the</strong>masses <strong>of</strong> Jupiter <strong>and</strong> <strong>the</strong> Sun is equal to 1:1071 <strong>and</strong> states that hisprobabilistic method gives odds <strong>of</strong> 1 000 000 to 1 that this result is nota hundredth <strong>in</strong> error 10 . Accord<strong>in</strong>g to modern data, that ratio is a littlemore than 2% larger so that <strong>the</strong> odds are obviously wrong.The great question here is, however, was that occasioned by amistaken treatment <strong>of</strong> <strong>the</strong> observations or by a systematic error <strong>of</strong>those observations impossible to elim<strong>in</strong>ate by any statistical treatment.I was unable to answer that question. In general, it is very easy tocommit such an error, <strong>and</strong> it is relevant to remark that quite recently<strong>the</strong> mass <strong>of</strong> <strong>the</strong> Moon was corrected <strong>in</strong> its third significant digit so that<strong>the</strong> precision <strong>of</strong> modern numbers should be carefully considered. If,however, we tend to believe that <strong>the</strong> observations were treatedcorrectly, <strong>and</strong> modern numbers are also correct, we arrive at an91

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!