Not that i disagree with you, but coherence is one of those things that highly subjective and context dependent.

A non-science inclined person reading most scientific papers would think they were incoherent.

Not because they couldn’t be written in a way more comprehensible to the non-science person, but because that’s not the target audience.

The audience that is targeted will have a lot of the same shared context/knowledge and thus would be able to decipher the content.

It could well be that he’s talking using context, knowledge and language/phrasing that’s not in the normal lexicon.

I don’t think that’s what’s happening here, but it’s not impossible.