[aetherson]

An interesting problem of working with "real" cabs is that you have to compete with street hails (that is, people doing the traditional stand-on-the-curb-and-wave-their-arms deal). This is a surprisingly big issue.

During slow times, it's not a big deal. But during slow times, you also aren't providing very many rides to the cabbies. And also pricing, which I'll come back to.

During peak times, it's very likely in a dense urban environment that a taxi driver will see a prospective customer and want to bail on your ride.

It's hard to make drivers not do this. You can easily get bogged down in a war of you trying to prevent them from bailing on people during peak hours (but remain hailable) and them trying to game you back.

Passengers will only use you if you can reliably get them a ride during peak hours, quickly. Also, taxis have mandated rate structures, and they will likely be more expensive than UberX or Lyft during non-peak hours, so the only time that you'll be able to really get a lot of passengers is during peak hours, and that's the time when all your drivers are at their most flakey.

Source: I spent two years as the lead engineer of Flywheel's server team. Flywheel being an Uber competitor that works with real cabs. Disclaimer: I've been away from Flywheel for more than a year, so my experiences are not current.

[greggman]

I'm curious how this is any different than Uber other than presentation? During peak hours there's peak pricing and not enough cars to go around so I have to wait.

Also, if it was legal for drivers to pick up hails I'm sure many drivers would just turn off Uber during peak and then turn it back on after, no "flakiness". Besides, their rep will go down if they bail on pickups so it seems like they'd just turn if off during peak to avoid that.

[aetherson]

Obviously in the ideal world for passengers, they'd get prompt pickups at any time for low prices. But given that that's not realistic, our experience is that they prefer moderate surge pricing plus prompt reliable pickups to "your cab is on its way oh wait no it's not," or just "we can't get you a ride right now."

Turning off during peak is bad. Again, your service will only be used if you can remotely get people rides during peak times. They're peak times because that's when passengers want rides.

[danielnaab]

Interesting observations on the dynamic with "real cabbies."

I would envision a decentralized system including both "real cabs", whose parent company vouches for a driver's licensing, as well as independent drivers, whose insurance company does. But if any driver participated in a transparent network, their reputation would be important to them, and I would assume that would be a deterrent to taking fares you don't actually intend to pick up.

[aetherson]

Well... I mean, it depends on how you do reputation.

Passengers do not by-and-large want to page through a list of cabbies and try to trade off the reputation of the person versus how far away they are or whatever.

Uber gives its reputation system some bite by just delisting drivers who fall below some rating system (and probably by threatening to do so considerably before they actually do so), and that's more-or-less what we did at Flywheel as well. But how does a decentralized system do that? For that matter, how does a decentralized system prevent the driver from just giving themselves a good rating by creating some dummy passenger accounts and upvoting themselves?

The reputation system also sort of implicitly depends on you having a working system that gets a substantial volume of rides. But the most vulnerable time for a TNC is right when you're starting out. You need to gain your users' trust, and that means getting them rides promptly the first few times they use you.

I'm not saying that it's impossible, and for the record I don't think that Uber has a "natural monopoly." I actually wouldn't invest in Uber precisely because I don't think they have any way to protect any profit margin they have. But mostly because I kind of expect that if Uber ever really demonstrates that they can actually make a decent margin, Amazon will come in and clone the service, not because you can do it with OSS.

[danielnaab]

The software agent that consumers interact with would have to be responsible for determining the accuracy of ratings.

I suspect that the most important aspect of such a system would be trusted participants: eg, insurance companies, city licensing agencies, etc. If you get one hundred great, but fake, reviews from shell accounts, it would be rather meaningless if your city didn't issue you a license on the blockchain, or you claimed false insurance which was somehow disputed on the same blockchain by the legitimate insurer.

[bfung]

What's in it for the insurance company to do this blockchain activity? What's in it for the city? They're not going to do this for nothing. And whatever entity that will help get this to work - wouldn't they be the next "uber"?

If the city is "issuing" licenses on a blockchain, how is this different than the medallion system already in use?