Tommy Wasserman

Consider this review to be a double feature.

That's because anyone who writes about this book really has to do two things. One is to review the book, the other is to explain the Coherence-Based Genealogical Method (CBGM).

The CBGM, the subject of the book, is a computer-based system designed by Gerd Mink which takes vast amounts of data about New Testament manuscripts and... produces a bunch of other data. What it gives us is a sort of measure of relatedness ("coherence") -- but only a sort of measure. It does not tell us if the relationship is genetic, or how it came about, or exactly what it means. And it is based on a strong measure of subjectivity (the direction of local stemma). With all that, it does not give us a stemma (or what the authors obfuscatingly call a stemma codicum, because they want to reserve the term "stemma," which has always been used in textual criticism for a family tree of manuscripts, for something else -- a sort of hand-wavy set of links between texts).

The key advantage of the CBGM is the huge amount of data it processes. The key disadvantage is that it requires all that data to produce its psychedelic daydreams of not-really-stemma. In other words, it requires a tremendous commitment of effort and resources to produce its results, and there is no reason to believe that this is the best use of all that effort. Give me all that data and I -- an actual mathematician, if a rather low-level one -- and I'll give you a lot more results than this foggy haze; give it to Stephen C. Carlson and he will produce an actual stemma -- we know that he can do so; he's done it. No one has even shown that the CBGM results mean anything, and yet the new Editio Critica Maior is based on this thing because, well, it's better than the acknowledged alternative, which is no mass comparison at all. (Carlson's work has been entirely ignored by the ECM editors, and while this book mentions Carlson, it buries him under the rug because, well, who wants to admit that somebody else has done something much more useful with your data?)

OK, so that's the CBGM. We might say that it represents a lot of progress in a somewhat wrong direction: It has genuinely taken us closer to the goal of a better edition of the New Testament, but we could have gotten us a lot closer still if all that effort had been pointed in the right direction. But what about the book? Does it explain the CBGM?

Yes and no. No, as in, there is no usable explanation of the algorithm, let alone sample code. In other words, you can't, based on this book, take the CBGM and hack at it for your own purposes. If you're a programmer, forget this book.

It does offer some examples of how to use the online CBGM software to draw your own conclusions about the data. The examples are, of course, cooked; they don't really justify the use of the CBGM. But at least they give you an idea of how to do your own tests.

I'd feel better about that, though, if there were some real explanation of why this is supposed to work. There isn't; it's all handwaved. Part of the problem is language. I'm told by someone who knows him well that Tommy Wasserman is a very intelligent man -- but I know, having exchanged e-mail with him, that that doesn't come through when he is writing in English. I don't know anyone who knows Peter J. Gurry, but he is an English-speaker who had to learn about a method invented by Germans. And neither one of them is a mathematician or a programmer. So the explanations are vague and not very clear. You obviously can tell that I have strong reservations about the CBGM. I don't know that I'm right, because I haven't seen the algorithm; I just know that nothing said about the method gives me reason for confidence. And this book does nothing to change that.… (mehr)