That's right, a peer to peer system that derives its strength from that network model doesn't need more peers!

You actually have been focusing on a single, simplistic dynamic (as well as several over-simplified arguments, such as your nodes don't matter argument), the blocksize <> tx/s argument. It's one of the simplest of bitcoin dynamics to understand, and you seem to believe that it exists independently of all other factors. Consistent with the over-simplistic moreblocksize == moretxs that Hearn and Andresen were pushing. Fortunately, people that have real investments in the system decided that they wouldn't allow that to happen.