[jmckib]

You bring up good points, but I still think 25% of drivers (or even fewer) would be enough to compete with Uber, and here's why. If you've got a new startup and you want to match Uber's response time, you don't need to have nearly as many drivers as Uber, because you will initially be be serving a small number of customers in a geographically restricted area (if you're smart). Your first customers may actually get better response times than Uber, merely by having a higher initial ratio of drivers to riders. So once you've got that toehold, you can grow your service gradually, while maintaining the same response times that Uber does.

It's true that Uber could try to run promotions to starve competitors, but could Uber keep this up forever when it's so easy to compete? They have deep pockets, but not infinitely deep.

EDIT:

On second thought, your point about Uberpool is more compelling. Whoever has the higher density of riders in a given area will be able to do pooling better (although with diminishing returns as density increases past a certain point), and this could be hard for a newcomer to match. On the other hand, if you can reach that threshold by competing on non-pooled service, you may be able to roll out a competing pooling service later. It seems more surmountable than competing with Airbnb, because you could start by targeting a small geographic area, but I can see this driving some of Uber and Lyft's valuation.

Thankfully, newcomers will always be able to compete on price with regular (non-pooled) taxi service, so we can expect those prices to stay low, effectively setting a ceiling on pooling prices.