Carlson refers to Klaus Wachtel's example in Jude 5 (see Carlson's bibliography) where the ECM adopts Ιησους. Since I have dealt a lot with Jude over the years I am of course very interested in this crux. First, it is important to note that the evidence is not clear cut in this passage even with the CBGM (as far as I know), although the coherence might be better for Ιησους than κυριος. Therefore, in the installment κυριος was printed with bold dot below the main line reading.
Further, I note that a minuscule witness like 1501 which is basically a Byzantine witness yet attests to Ιησους, and none of its 10 closest potential ancestors have this reading, which is also an indication that this reading could also emerge independently, i.e., the adopted reading in the ECM does not have perfect coherence, so I suppose that this particular criterion is not entirely decisive. (Minuscule 93 is another witness which four closest potential ancestors have κυριος.) The program "Genealogical queries" on the website of the INTF can generate a textual flow diagram for Jude 5/20 involving the subject of the verb which looks like this. Just click on the image to get a larger version in your browser. One can note the line up to 1501 from 424 attesting to reading n (with κυριος). Unfortunately, I am not sure which all of the readings are in this stemma (although it can be inferred from the attesting witnesses), because the variation unit (Jude 5/20) does not correspond with the ECM apparatus that gives the larger stretch of text Jude 5/12-20, involving several distinct textual problems.
In any case, Carlson concludes in regard to the CBGM in general:
To its credit, the CGBM appears to be more rigorous and precise about evaluating external evidence than the heuristic rules of thumb currently employed in reasoned eclecticism. Its main drawback is that this promised precision may be more apparent than real. The theoretical basis for the CGBM is the work of one man, and the technique has been applied only to the Catholic Epistles. It has not been tested for other texts and its operation is still poorly understood outside of Muenster. Indeed, it has not borrowed its concepts from classical stemmatics or even from phylogenetics in biology, which have been tested and found to be robust over many different applications. It is still very much an unproven method; time will have to tell as more people gain experience with it.