I think you're imagining "communication" to mean inspirational TED talks or something, which is really just one very specialized case. I've spent quite a bit of time talking to Larry, Sergey, and Zuck, and in my experience they are all very direct and effective in their communication.
It might be a completely different type of communicator you need, depending on the direction of the status hierarchy you're addressing.
Talking to employees, or even partners, where the other party is responsible for understanding the general mental framework and needs a precise answer to a precise question, is very different to talking up the status hierarchy.
When you speak to an investor who expects a "crisp" response to an arbitrary knee-jerk question they just thought up, which buries in it several wrong assumptions, the correct response is to reverse engineer that, and address the assumptions themselves, while conveying a whole host of other attributes at the same time. Essentially you need to decode the other person's mental framework and encode your message on their terms. If the inferential distance is significant, there is no way this can be done in real-time. But what I worry about here is that the more off-center the entrepreneur's mental model, (which is what you need for innovation), and the more complex the execution, the harder it is to cross the chasm to communicating with an up-the-status-chain interlocutor that expects everything to be in bite-sized insights they can digest.
An observation here is that, in one case a correct answer is a direct answer, and in the other case a correct answer is an indirect answer.
the YC interview is particularly error-prone in that regard, as you don't know who you are going to be speaking to in advance, so you have little means for preparing. That setup (plus the 10 minutes limit) is a recipe for randomness. Essentialy you get good interviews if the two parties happen to have similar mental contexts, but it doesn't seem that YC takes that into account at all. On the other hand, it also favours mainstream mental models and simple-to-explain ideas, which seems directly contradictory to what YC wants to achieve.
However I don't see how convincing a non-motivated arbitrary person of a fairly complicated idea in 10 minutes is highly correlated to all the other contexts a founder has to deal with. Sure it's a great attribute if you can have it, but it's one in a long list.
I think the idea is that the founder's mental context should be a superset of the investor's mental context, so that he can quickly identify and address the particular issue the investor raises, having already thought about it.
Big ideas are usually really simple to explain to users. Google Search is basically "Type anything into a text box, and we'll give you answers." Ideas that are complicated rarely become big (as in widely adopted by many people).
