Faulty generalization in uno, faulty generalization in omnibus

I know I just gave one of these, but listen, it's warranted (and besides, it's my photoshop and I can do what I want). Check out Respectful Insolence for an incisive unmasking of the fallacious reasoning behind a popular gambit: the inappropriate application of the legal concept "falsus in uno, falsus in omnibus." This is such an over-the-top crackpottish move, and yet it actually makes use of a pretty deft shift in scope. Orac, the brains behind Respectful Insolence, gives it the treatment it deserves.

Steel your stomach, because here's an example of the tactic that Orac so handily demolishes:

For half a century now historians have told us that during World War II the Nazis had a policy to exterminate the Jews of Europe, along with homosexuals and Gypsies. We are told that millions were "gassed" at German camps such as Auschwitz and Treblinka.

We have been told that the ghastly process of mass murder was also carried out in Belzec, Buchenwald and Sobibor. And aren't there thousands of survivors who "escaped the gas ovens" and swear that all this is true?

And didn't the Nazis make lamp shades from human skin, and manufacture soap from the fat of exterminated Jews? Of course, you may answer, everyone knows it. After all, aren't such bars of "Jewish soap" on display in museums in Israel and other countries? How can there be any doubt?

"Falsus in uno, falsus in omnibus," or "false in one thing, false in everything," was a Roman legal principle. If a witness may not be believed in one thing, he should not be believed in anything. This principle is as valid today as it was two thousand years ago.

Well, of course it is, my horrible little friend. But did you catch the switcheroo here? It's an almost admirable feat of sleight-of-logic -- a little Latin amphibology, and "false in one thing, false in everything" becomes "false in one part, false in all." A rule used to cast doubt on a single witness (if she has lied once, all her statements should be under suspicion) is suddenly and absurdly expanded to apply to historical accounts or scientific theories (if any part has ever been shown to be false, the whole thing must therefore be false). Simultaneously, inaccuracy in fact or scope becomes equivalent to lying. Here's a sample syllogism: Perjury invalidates a testimony. Scientists once thought that heat was a fluid. Ergo, thermodynamics is a lie. But of course, the sleight-of-hand involved makes it hard, at least for the credulous, to perceive the stark absurdity of the argument.

Orac offers examples of this fallacy at work in our mutual bete noir, "intelligent design" theory. Here's his take on why the tactic, inappropriate when discussing history, is even more ridiculous when attacking science:

The problem is, this principle doesn't work in science. Why? The scientific literature literature is littered with papers whose results were later shown to be either incorrect or only partially correct. In most cases, being incorrect doesn't mean the scientists were lying, and it is the totality of the evidence that must be weighed. Moreover, it is not valid to treat all of science as a single source. Science is not a single witness that can be interrogated. Well-accepted scientific theories (like evolution, for example) are supported by many interweaving lines of evidence from many different sources. If you impeach one minor source or piece of data, that does not invalidate the rest of the supporting data.

Couldn't have said it better myself. Orac, sir, a logic award to you.