Unexplained deletions: continuing what appears to be a common theme
Wow, these are some interesting examples you've dug up. An answer to our querent, I would mention that, when I was at a conference of Indo-European scholars last month, one of the speakers brought up the concept of Nostratic in his talk, eliciting a hearty chuckle from those assembled. More generally, the field of long-range comparative linguistics (or "paleolinguistics" for those that believe it) normally does not use the same methods of rigorous historical phonological and morphological reconstruction as the historical linguistic mainstream, instead relying heavily on word lists often containing accidental resemblances. To be fair, I know that the adherents of Altaic are growing steadily, and that particularly proposal may bear further consideration soon, but Indo-Uralic and Nostratic, no way.
I downloaded a copy of Dolgopolsky's Nostratic Dictionary (https://www.repository.cam.ac.uk/bitstream/1810/196512/49/00ND_ALL.pdf) and searched for "OJ " (note the single space afterwards). Within 13 entries that included a purported Old Japanese cognate, I found the above four errors. I stopped searching at that point; some day I may dig deeper if a further analysis is called for. FWIW, scanning quickly through the next 15 just now suggests a similar error rate.