> Uber/Lyft did outcompete taxis on merit, and to suggest otherwise is revisionist history!
Although your point is poignant, it would be more valid if their price structure was (cough) sustainable.
What if Lyft and Uber were 25% cheaper than taxis; primarily because they didn't need to pay a dispatchers' salaries? This would be the classic case of automation lowering prices.
IMO, they could have out-competed traditional taxis without needing to price below cost; which would make your observation 100% true.
Instead, Uber and Lyft merely demonstrated that there is a market for the self-driving taxi business. Unfortunately, their relentless expansion shows the danger of letting ones' imagination run away and overestimating how quickly tech can be built. Self-driving cars is a hard problem; and building a business assuming that "AI can be delivered on a date" is a huge, huge, risk.
Clearly both companies lost the bet (that self-driving cars would work in time for their price model to work). Perhaps the Uber and Lyft stories are bad examples of the overall thesis of the article?
Taxis in many areas dramatically increased costs by artificially limiting supply to sometimes absurd levels. The obvious example is NYC where taxi medallions (a transferrable license to operate a taxi) were going for $1.4 million not long ago [1], and rapidly trending upwards. $1.4 million for a license to operate a cab is no less brokenly unsustainable than businesses running off funny money from investors, owing to "perpetually" low interests rates motivating that very behavior.
I expect some of Uber's appeal, from the perspective of investors, was predicated on a belief in the arrival of widespread full vehicle automation. If that happened, they'd have had the infrastructure, systems, and brand to immediately swap over, on a global level, and become an unbelievably profitable (and massive) company. Add in some massive scale vertical integration of maintenance and other costs, and it's hard to overstate how big they could have become. Of course it didn't happen, so instead we just got a bunch of cheap taxi rides, and that's cool. Feel bad for the medallion guys who got left holding the bag (those licenses now go for < $100k, which is a vastly more sustainable equilibrium) but at some point you've gotta trust your tulip detector.
NYC had those medallion limits on purpose though, because the primary goal was limiting congestion. Manhattan is a tiny island with a small amount of road space, and taxis or Ubers roving around contribute significantly more congestion than a commuter parking their car for most of the day. And yellow cabs in effect mostly travel in Manhattan.
As with any sufficiently complex system, simple solutions fail to take into account knock on effects. As such, I'm curious about what impact that decreased traffic speed had. Did it reduce traffic deaths? Did it increase deaths in ambulances? Was it uniform or were/are there areas that are at a 50% decrease and more at 1%?
One issue I think we are seeing is that we, as tech workers, are super excited to fix the immediate problems. We have the skillset and it's really engaging! But we will often miss the forest for the trees and then double down on our solution being the solution instead of one part of the solution.
All this to say, thank you for discussing the traffic slowdown caused by increased cars. It's a side effect I hadn't been considering before.
Ironically traffic speeds in New York are measured using yellow cabs since they drive around constantly and are required to have GPS.
I think one of the biggest impacts has been on the sustainability of the transit services. For all road users, slower speeds require you to maintain more labor and vehicles out to maintain SLAs/frequencies, or to increase waits. For bus users in particular, it starts a vicious cycle because of several additional factors
* some baseline amount of people will just switch to rideshare instead of buses, reducing revenues and decreasing traffic speeds since buses are more space-efficient than cars
* with less revenues and less ridership, there is pressure to cut service to be more "efficient" at spending money
* with poorer services, more people switch to rideshare, further reducing revenues and decreasing traffic speeds
Now what level of congestion is 'acceptable' is mostly based on comparing to other cities, and how the electorate is feeling. That being said, the political pressure was so strong that NYC had already implemented congestion charges on rideshare and taxis, and is currently implementing tolling in Manhattan to try and reduce car numbers. Uber actually came out in support of it. https://www.uber.com/blog/new-york/uber-supports-congestion-...
> NYC had those medallion limits on purpose though, because the primary goal was limiting congestion.
Medallion limits exist primarily to maintain and/or increase the value of existing medallions. Other reasons given were just propaganda by existing medallion owners.
The question is, do cities have an obligation to meet all supply if their residents don’t want to? People want all kinds of things that often contradict.
Now NYC is introducing congestion charging on all vehicle trips in Manhattan to try and restore traffic.
Uber and Lyft could win--depending on location--based on product. But they also deeply undercut taxis on the basis of price until people got used to using them.
>What if Lyft and Uber were 25% cheaper than taxis; primarily because they didn't need to pay a dispatchers' salaries?
I'm pretty sure dispatchers are cheaper than expensive SV engineers.
> >What if Lyft and Uber were 25% cheaper than taxis; primarily because they didn't need to pay a dispatchers' salaries?
>I'm pretty sure dispatchers are cheaper than expensive SV engineers.
That’s like saying paying human computer wage is less than electrical engineers who develop electronic computers.
Non-recurring engineering cost is higher BUT IT’S NON-RECURRING and largely independent of the number of users, unlike dispatchers.
Per SEC filings Uber has close to 4 million drivers worldwide. Even if you divide by 4 to account for part time drivers, you'd need 10k dispatchers. If their engineers make 5x as much as the dispatchers than you could have 2k engineers solely dedicated to automating dispatch and still break even.
I think they roughly have 2k engineers total and I have no idea what the breakdown is of what they do.
One thing I didn't see discussed anywhere in this entire thread is how impossible it was to get a cab at night (at least in Boston). On a Friday or Saturday night it was absolutely impossible to get a cab past 11:30pm. When I would rarely find one, they would ask for $50 cash to go to the destination, which would have been a $10 ride if metered. I frequently walked 5 miles home from a party at 1am. I have even heard stories of girls sleeping over at random guys apartments since it was likely safer than walking home alone. All of this completely disappeared when ride sharing appeared.
Which reinforces a point that I couldn't make above (because I had to go to a meeting)
I also wonder if Uber/Lyft pricing below cost had to do with competing with each other? IE, if all they could do was compete on price, would that (cough) force them to undercut each other? If one decided to price below cost, would the other fear that they would loose too much market share too quickly if they kept their prices sustainable?
Although your point is poignant, it would be more valid if their price structure was (cough) sustainable.
What if Lyft and Uber were 25% cheaper than taxis; primarily because they didn't need to pay a dispatchers' salaries? This would be the classic case of automation lowering prices.
IMO, they could have out-competed traditional taxis without needing to price below cost; which would make your observation 100% true.
Instead, Uber and Lyft merely demonstrated that there is a market for the self-driving taxi business. Unfortunately, their relentless expansion shows the danger of letting ones' imagination run away and overestimating how quickly tech can be built. Self-driving cars is a hard problem; and building a business assuming that "AI can be delivered on a date" is a huge, huge, risk.
Clearly both companies lost the bet (that self-driving cars would work in time for their price model to work). Perhaps the Uber and Lyft stories are bad examples of the overall thesis of the article?