I think you misunderstand how network effects work. What you're talking about is too localized and transient to really be considered network effects -- I can also share rides in a normal taxi if I hail a cab with 2 friends. A key component of a true network effect is that it enables exponential scalability: Facebook is the textbook example here in that it becomes more useful the more people use it. Another key component of network effects is that they enable platform lock-in by creating enormous barriers to entry for potential competition.
Uber enjoys a small amount of the second component through their large driver network; but taxi drivers are more or less a commodity so there aren't really barriers to entry for a competitor to go hire a bunch of drivers. With Facebook, there's more lock-in: even if lots of people are on a competing service, if YOUR friends aren't, it has no increased value. There is no such personal component with Uber.
Eventually, what I see happening are a number of competing services popping up that drive down the payments that drivers receive. This will eventually bottom out the market for taxi drivers to a supply/demand equilibrium point, which is exactly what's supposed to happen in a commodity market.
My guess is that Uber's long-term plan is a brand play: everyone with even a little bit of future vision knows that self-driving cars are right around the corner, and Uber is trying to establish itself as the premier on-demand transportation service before that happens. Because once self-driving cars ARE available, anyone with deep pockets will just be able to buy a fleet and put them on the road. If Uber can establish itself before that happens, they'll have differentiated themselves with the strength of their brand, which again is the optimal strategy in a commoditized market. But I see Uber being the #1 company in a crowded market with a ~20-30% market share; not 70-90% as is often seen in markets with strong network effects.