This is an interesting review or set of observations about the NA28, largely positive, and full of information. Among other points, he does suggest that the basis of the new parts of the text on the so-called coherence based genealogical method should have been explained better within the edition.
Without a proper understanding of the Coherence-Based Genealogical Method, the 28th revised edition of Nestle-Aland will remain enigmatic and even problematic. (p. 273)He discusses James 1.20; 1 John 3.7 and Jude 5, comparing the text and apparatus of NA27 with that of NA28. This is a frustrating operation since there is nothing in the evidence cited in the apparatus which would explain the changes, it is all about a different method and a different way to evaluate witnesses: 'The textual choices of the editors can only be understood in light of the Coherence-Based Genealogical Method.' (p. 276) Indeed he suggests that the apparatus should include the abbreviation TP (for Transcriptional Probability) 'to indicate those instances where the lectio difficilior, although supported by fewer witnesses, is probably the original reading' (p. 278 n 15).
He shows how the fuller apparatus of the NA28 is generally much more easily understood than the more compact apparatus of NA27. This expansion comes at a cost though, in a hundred more pages (p. 282), and a much higher proportion of each page given over to the apparatus and less to the actual text (I have noticed this a lot recently).
He closes with three suggestions for future revisions: i) discontinue single square brackets (p. 288). [I doubt whether this will happen immediately, I think they will be discontinued as the ECM is produced, uncertainties in ECM sections can be signalled with black diamonds.]; ii) use VL instead of Itala; and iii) include the umlaut/distigmai from Vaticanus in the apparatus. [This would be radical, especially in the supposed [but nonsensical] compound form of the distigme-obelos previously known as the bar-umlaut. But it is interesting that he thinks there is 'a consensus among textual critics and codicologists that text-critical signs can be dated to the time of the writing of Codex Vaticanus' (p. 289). I'm not sure whether it is a consensus or not; I would regard the suggestion of Payne and Canart as interesting and flawed. They are most likely from the sixteenth century.]