I did not find a clear answer on the site wether this project is aiming at independent ridesharing or at being a turn-key solution for more conventional taxi dispatch organisations. There is the idea of "hack and tune it for your city", which is clearly not the global approach, but then they also advertise "1-minute hiring", which is definitely not what I would be looking for if I were to customize it for my hypothetical preexisting organisation.
Something like wordpress or the myriad open source online shop systems, but for taxi dispatch and with a unified smartphone client (or client protocol, actually) that can work with all installations. That might be the best way to avoid a single gobally dominant private network that can set their "tax" on all rides at will. The app/clients might be supported for discovery by a global registry of regional installations and some trust management, but it might actually be not only easier but also more reliable to just leave that to plain old web, with something like libretaxi:-URIs to connect the app to a regional network discovered by your search engine of choice.
Taxi drivers, at least herearound, have mandatory rest periods, which are very strictly enforced. I know that for a fact, since my girlfriend used to drive a cab. Even if you adhere to your rest period (and can prove it), but don't have the paperwork straight this can lead to hefty fines.
There's nothing stopping your friendly Uber driver to drive you around after two back breaking shifts at the factory.
He may even quaff a couple beers, or three, before starting his "Uber shift". Something which is an absolute No!No! for a licensed cabbie. They have very stringent zero tolerance rules when it comes to booze. Your Uber driver does not.
I can't speak for the rest of the world, but imagine that this situation is quite comparable to a lot of places.
1) There's no check to ensure that the vehicle is ensured for commercial rides before they start taking hails.
2) Likewise, no background check.
3) No regulatory body to investigate cases of e.g. scamming passengers out of money or refusal to comply with agreed-upon rates and dropoff. [1]
4) The local PD doesn't stand ready to treat assault on a driver like assault on a police officer, in terms of marshalling resources.
5) No central registry, or even private registry of cars/drivers in operation.
6) Drivers aren't trained in the relevant anti-robbery countermeasures.
7) In NYC, the barrier to prevent the most obvious kinds of robbery.
That makes a big difference in safety of paying cash to a roving operator!
Sadly, I think the "big corp middleman" does add a lot of value that a free app can't replicate without similar investment, scale, and "walled garden"-ness.
[1] I know commenters here like to roll their eyes about the clumsiness and lackadaisical enforcement -- I'm one of them! But it's at least something, which prevents a major scale-up of this behavior.