Hacker News Reader

Comment by lukifer | Hacker News Reader

The P2P transportation market is an ideal one for a workers' cooperative. The fact that Uber and Lyft are running at a loss (...for now) does not make them any less rent-seeking in their business model.

https://drivers.coop/

https://ridefair.io/

View on HN

From the jobs page of the first link for a software engineer position: "The position is salaried up to $72,000 a year depending on geography. Fixed-term contracts are available."

Sorry, but you're not buying competitive talent with 72k/year salary. Uber can squeeze out tons of marginal efficiencies via better routing/matching, price discrimination, and surge algorithms. Implementing those algorithms means hiring good talent, especially in ML. Having a fancy frontend is only 10% of the picture if you want drivers to have good utilization.

> which will never scale past the size of a single US state

People live in a city and take cabs in that city. Build a city-scale app. For extra credit, add federation APIs to interoperate with other city-scale worker-owned co-op apps.

Only a tiny 0.1% of people (including Uber's dumb investors) think that being able to use the same app in every city on earth is a killer feature.

It kinda is a big feature though. I'd imagine people who need a rideshare service are over-represented in people who travel a lot for either business or pleasure, and likely do so to multiple random cities.

I remember traveling to a city in Canada 4-5 years ago when they didn't have Uber or Lyft, and there was a local app that was vexing to try to use if you didn't already live there, and assumed you had a local address. There were a lot of places I just didn't end up going to because I couldn't work out the app, and couldn't find a normal taxi on the street that didn't require cash that I didn't have.

For a lot of people, the times you're calling a cab are a stressful experience already - you're in a new place, maybe with a new language, often under time pressure to get somewhere. There's definite advantage to using a familiar app in that use case.

I guess you're right that having Uber is a big help for all the people who fit the center of the Venn diagram for:

- Unwilling to use public transport

- Unfamiliar with the city

- Don't carry any cash

- Don't speak the local language

- Unable to install/set up a new app

- Travelling alone, without a local who can help them

Are there enough of those people to justify Uber as a $100B business? I am very sceptical.

What you've described is most business travelers, which is essentially what keeps most of the airline and hotel industry alive, as I understand it.

You don't even need multiple apps. Even if a Postgres DB only scaled to a single city, which is wrong to begin with, you could shard it per city/state/country pretty easily.

All postgres needs to handle is a large metro area and then is trivial to shard and scale out.

> edit: and the second link advertises using Postgres, which will never scale past the size of a single US state

to be fair depending on the exact implementation details you probably could do everything you need, including management using Spanner and BigTable.

I think most of those efficiencies can be done better by a human dispatcher. The key thing which made Uber take off was a combination of convenience and better cars. Solve the convenience and that can take you a far way for many people.

I don't really think humans are better equipped to route 10-15 cars in a region factoring in drivers' preferences for direction, pickup and dropoff zone attractiveness to drivers, and traffic. See this research on human solutions to the vehicle routing problem: "When comparing the human performance with the optimal solution and classical heuristics (nearest neighbor, savings, and sweep), we see that participants typically perform better than the worst heuristic and worse than the best heuristic" [1]. All of these algorithms are pretty naive baselines and you can do a lot better with actual routing software. And this being a low margin winner take all business, Uber being slightly better with algorithms makes it substantially better as an option compared to the coop.

Also I think you're underestimating the difficulty of consistent hiring. You can definitely find a good dispatcher in NYC, but can you scale that level of skill to the entire US?

[1] https://www.sciencedirect.com/science/article/abs/pii/S03050...

this assumes Uber can only hire silicon valley engineers. $72k is plenty money in a cheap gas/food/rent no state tax city like Dallas

$72k for Texas lol. I don’t know a single Silicon Valley quality engineer working for that anywhere in the US.

> ideal one for a workers' cooperative

I am sorry, but there is not way a workers' cooperative can build a safe product for passenger and drivers. All of the leading rideshare companies have invested 10s if not 100s of millions into safety platforms that leverage the latest face recognition, fraud algorithms, and more to ensure that everyone can have a safe ride.

Both passengers and drivers robbed and murdered each before these investments were made to keep the bad guys off and the good guys safe.

...you know why they need to do that? Because they have an incentive to let basically anyone join and have an adversarial relationship. A workers cooperative can be a faceless organisation, but unlike a company it is also an actual community, so things like vetting who let in I'd much more feasible to do you'd do in a community than in a "startup family" trying to squeeze their drivers. It could be as simple as requiring somebody to vouch for you and being responsible for whatever fuckups you do.

How are cooperatives not incentivized to "let basically anyone join"?

The more drivers a co-op has means shorter wait times and higher geographic coverage for the passengers. If passengers have to wait too long, they will lose faith in the reliability of the co-op and be forced to use other forms of transit.

The more passengers a co-op has means drivers have to idle less or shorter (unpaid) distances to travel to pickup a passenger.

It is always better for the co-op to have more active drivers and more active passengers. The growth comes in the form of "let basically anyone join".

Why do you think a cooperative will be conscious about reputational awareness of reliability, but not about safety?

I never said a co-op would not care about safety. I am saying that they lack the resources to implement low-cost safety solutions.

None of the co-ops listed have industry standard safety systems for telematics (unplanned stops, detour detection, or unsafe driving), face recognition, or audio recording.

Feel free to address the original point of how co-ops have the same incentives as Big Tech for letting anyone on the platform.

I don’t think you’re wrong but it’s certainly extremely pessimistic to believe this is impossible.

This seems very optimistic about the likelihood of a "community" to vet itself properly vs. covering things up "for the good of the community". Obviously it can work out, but there's bountiful examples of communities ranging from schools to churches to workers unions that have a pretty well established history of systematic abuse and coverups.

The thing is, as a co-op in a small and competitive market, it's not like you have much leeway for reputational damage. Unlike Uber, you can't really survive a scandal by filling in with profits from other cities or drivers.

And what you said is true, but it's also a vapid statement without looking at percentages and incentives. Individual bad examples exist everywhere, but a workers coop of drivers in a City has social incentives (if a driver from your coop robs someone who goes to the same church as you, you'll hear from it) and constraints (if your cooperative gets boycotted in your town, what will you do, move everyone next postcode?) which uber lacks (if there's a boycott in one town, bad for the drivers, but you just put out a press blurb, do a promotion and donate to a charity and wait until things die down)

But now it sounds like you're advocating for Taxis.

We'll, in my native country, Taxis are actually working fine, they are regulated but with no cap on the licenses - uber eventually needed to apply for those as well. And we also have public transport and walkable cities. So it's not like gig-economy driving is the only possible way to have a working system

It'd be interesting to see actual statistics on rideshare crimes before and after the introduction of these intended mitigations. Anecdotally, the number of gruesome news stories seems to have remained roughly constant.

Reply on news.ycombinator.com