Doubt is Possible
This is an interesting little case study in the use and abuse of evidence, investigative techniques, language and rhetoric, inference and conclusion. One of those (all too familiar) occasions when attention-seeking and self-aggrandizement dress themselves up in scientific (or pseudo-scientific) vocabulary and give the whole enterprise a bad name.
Dominique Labbé, a specialist in what is known as lexical statistics, claims that he has solved a “fascinating scientific enigma” by determining that all of Molière’s masterpieces…were in fact the work of Pierre Corneille…”There is such a powerful convergence of clues that no doubt is possible,” Mr. Labbé said. The centerpiece of his supposed discovery is that the vocabularies used in the greatest plays of Molière and two comedies of Corneille bear an uncanny similarity…Mr. Labbé contends he has infallible statistical evidence of Corneille’s “fingerprints” all over Molière’s greatest works.
Well that’s a bad sign right there, a ‘scientist’ saying that no doubt is possible. That’s a pretty dodgy thing to say even with overwhelming evidence – in fact one could say it’s simply nonsensical, because doubt is always possible. And in real science (which avoids words like ‘infallible’) it’s also essential. A putative scientist who tries to rule it out in advance, even (or no, perhaps especially) when talking to journalists, is letting the side down. And Zanganeh’s article does an excellent job of collecting quotes from other scholars (scientists, honest inquirers) in other fields that point out what a lot of doubt is possible and why.
“Lexical statistics can be useful as an exploratory tool with a descriptive and investigative goal…In no way can it be used as a proof.” In a nutshell, attribution of authorship necessitates a convergence of presumptions. Joseph Rudman, a professor of applied statistics at Carnegie Mellon, agrees that even the best authorship-attribution studies could yield only probabilities. “You can never say definitely, just like in a DNA result,” he said…Indeed, at the heart of this debate lies a more fundamental question about the use and abuse of scientific tools in the field of letters.
But it could be argued that the way Labbé has used it, his method isn’t really a ‘scientific tool’ at all. Obviously these matters are always contested and debated and argued over, but surely one hallmark of good science is not producing evidence of one narrow fact (that Molière and Corneille used very similar vocabularies in their plays) and extrapolating from that to much broader claims. And surely another is not relying on only one narrow piece of evidence to support (let alone ‘prove’) a claim which would require evdience from a number of other fields – a ‘convergence of presumptions’. Ignoring, apparently not even noticing, what kind of evidence one needs to make a case, is hardly good science.