Same goes for startups. Given the barriers to entry in the software business, which are low but still very much nonzero (hence the need for VC), and given that VC can't fund all the possible ideas that want to enter, what's in the market is what non-consumer participants have decided in their wisdom that consumers will probably like. Which may well be very profitable, but is unlikely to maximize profit.
Which leads to a simple reply to PG's claim, "When you accuse Silicon Valley of x, you're implicitly saying x works well, which doesn't seem smart if you're against x." No, I'm implicitly saying that x works rather poorly, because Silicon Valley works rather poorly and is still profitable. It could be much more profitable.