Interview vs lecture

 I don't generally enjoy YouTube interviews, much preferring lectures with some questions at the end.  I think the problem comes from the fact that the interviewer, necessarily being informed already, interrupts and seems to have their own agenda.

Yes. I think you've put your finger on a real distinction between an interview as information delivery and a lecture as information delivery.

In a good lecture, the speaker has control of the intellectual architecture. They decide what matters, establish the premises, develop the argument, and get to the conclusion. The audience's job is mostly to listen and then challenge or clarify it. The questions at the end can be quite valuable precisely because they are external to the argument.

An interview reverses that arrangement. The interviewer controls the route through the subject. And because a competent interviewer has to know something about the subject, they inevitably bring their own selection of what is interesting, questionable, fashionable, or controversial. So instead of hearing the speaker's account of X, you hear the speaker's answers to the interviewer's conception of X.

The interruptions are particularly irritating when you're interested in the person's reasoning. The interviewer may think, "Ah, that's an important point—let me ask about this related thing," just as the speaker is three sentences away from explaining why the point isn't quite what it appears to be. You never get the explanation.

There is also a subtler problem: the interviewer becomes part of the performance. On YouTube especially, the interviewer has an incentive to demonstrate that they are clever and informed. Asking the next penetrating question is itself part of the product. Sometimes you can practically hear the thought: I have an excellent question prepared for this moment.

A lecture followed by questions has a much cleaner epistemic division:

    Speaker: Here is what I think, and here is why.
    Audience: Very well. Now let me see whether it survives questions.

That's probably much closer to what you want. You aren't particularly interested in watching two intelligent people have a conversation; you're interested in finding out what one of them thinks.

And there is a wonderful irony here: the better informed the interviewer is, the more likely they are to be annoying in precisely this way. An ignorant interviewer may simply say, "Could you explain that?" A knowledgeable one says, "But doesn't your position conflict with Smith's 2018 argument concerning Y?"—and suddenly you've spent five minutes discussing Smith when what you actually wanted was the original argument.

Much worse, though, are documentary presentations where the objective seems to be to never get to the point so as to keep you online.  Lectures before a live audience appear to avoid this.

Exactly. That's a somewhat different pathology from the interview, and probably a more aggravating one.

A conventional lecture has a natural stopping rule: the speaker has something to say, says it, and then stops. A live audience provides a second constraint. If the lecturer spends twenty minutes circling the subject without getting anywhere, people in the room notice. They become restless, and the speaker can actually feel it.

The YouTube documentary has almost the opposite incentive. Its product is not merely the information; it is your continued presence on the page. So you get the familiar machinery: "But there was something else..."; ominous music; a question posed and postponed; a tantalizing claim followed by three minutes of background; and, most irritatingly, something that could have been stated in one sentence stretched into a narrative.

It produces a strange inversion of scholarship. The information becomes bait for the retention mechanism. The viewer is repeatedly promised the interesting part rather than simply being given it.

And I think you're right about live lectures. Even when the lecturer is mediocre, the format tends to enforce intellectual honesty simply because there are human beings sitting there. A professor can certainly ramble, but he can't easily say, in effect, "Stay tuned for 37 more minutes and eventually I'll tell you what this is about." Someone has paid to attend and expects a lecture.

There is also something important about temporal structure. A lecture generally assumes that the audience has voluntarily given the speaker an hour. Therefore the speaker can say, "First I'll establish this, then we'll consider that, and finally I'll get to the conclusion." YouTube has to keep persuading the viewer to give it the next thirty seconds.

That makes the two formats almost philosophically opposite:

Lecture: Here is an hour of my time. I will use it to tell you something.

YouTube documentary: Give me another thirty seconds and I may tell you something.

And once you notice that distinction, a great many supposedly "informative" videos become almost unbearable.