There is zero friction to install a new app and accept fares via that app, except perhaps how many ridesharing apps you can run at once on your device for accepting passengers.
The linked article says Lyft now have an average pickup time in SF of 2.5 minutes. Good luck bootstrapping to that kind of performance with a brand new ride share company starting from scratch.
Based on what I've seen, Hailo tried to buy their way into this market, albeit with the twist of using licensed cabs. Seems like they had little trouble signing up drivers (who were predisposed to distrust the UberTAXI) but my assumption is that they gave up in North America because they couldn't afford to keep giving away the free rides and discounts needed to build up a customer base.
Drivers using both Lyft and Uber is a perfect demonstration that they care most about the size and liquidity of the demand side of the market. They are using both in order to increase demand side liquidity from their perspective.
And BTW, according to some uncofirmed twit[1], more than 50% of lyft rides in sf are done through lyft-line, which is a pretty new service.
To a lesser extent, user density is also related(in theory) to driver ride time, both in reducing wait time, and reducing time from call to passenger.
Also there might be a possibility they can subtly incentivize drivers to only drive for them by giving complying drivers a bit more work, or better work.
And as for your risks lists,you can paraphrase it as: the hand of god(government), a darpa scale technology(self driving cars), very well resourced competitor.
But every company is under risk from the first 2, even for businesses with a strong competitive advantage.
Anyway! I see Uber going the way of webvan. Great proof of concept that VC money is going to burn through, followed up by some combination competitors, self-driving cars, and local/state government transportation agencies putting together federated/open APIs to perform the same services but in a regulated manner (or governments contracting with smaller players to perform the same services).
No amount of VC is going to remove a government's ability to regulate transportation (Montreal, Canada impounded 40 Uber drivers' vehicles the other day [2]). Then again, it shouldn't. Transportation should be regulated, but not for monopolistic reasons.
Anyone want to make a cheap long bet for funsies?
[1] http://www.andresmh.com/nyctaxitrips/
[2] http://www.cbc.ca/news/canada/montreal/montreal-taxi-bureau-...