[aetherson]

Ride-sharing no more has strong network effects than Doritos do.

Which is to say: Is it an advantage for Doritos that they have tons of brand recognition and a large customer base? Sure! They get economies of scale, and it is easy for them to convince retailers to carry their product. In turn, their customers know that they can get Doritos anywhere, and that's an advantage for their customers.

But no reasonable person would regard Doritos as a network-effect business.

Passengers for Uber do not make the service better for other passengers. In fact, they make it worse, by driving surge multipliers. If another business provides a rival service to Uber, passengers experience no negatives for switching, the way they do to a brand new social network, where their friends aren't.

The existence of Uber's brand, and the demand that Uber experiences, is of course an advantage for Uber. Obviously, it is a good thing for Uber that they have passengers, and can sell their services. But that's not a network effect -- it's just a regular old "advantage of being big and having a brand."

[tuna-piano]

I'm going to have to disagree.

1. Drivers want to have passengers close by for pickup. Passengers want to have drivers close by for pickup. The more drivers, the quicker passengers will get picked up, the more passengers will use the service, and the more drivers will use the service.

2. In addition, UberPool/Lift Line have obvious and strong network effects. Sharing a ride is only possible with a large number of users. The more rides that can be shared, the cheaper the rides can be.

In both these ways, additional passengers do add value to other passengers.

These services are really just curated marketplaces of drivers and passengers. Uber has its driver/passenger marketplace, Lyft has its marketplace. It's tough to replicate the power of these marketplaces, just as its tough to replicate the marketplace power of Craigslist or eBay.

You need a substantial amount of both buyers and sellers for the service to be worth anything. If Doritos was not yet in Canada, they could provide value immediately by selling Doritos in one 7/11 shop, no problem. For Uber to expand to Canada, they would need a minimum number of drivers and passengers in a given geographic area.

[rbranson]

I think you're confusing economies of scale with network effects.

For example, a hotel chain is more valuable if it has a global presence because it can leverage that to create customer loyalty. That is not a network effect. To the consumer, it does not matter how many people use those hotels. They could be completely empty for all they care, they just want a room for a competitive rate.

More efficiency through higher utilization (more passengers lowers the cost of pooled rides) is a classic example of economy of scale.

There are counter-examples to your "Uber expand to Canada" argument. For instance, a mobile provider is not going to be very successful in a market until they have coverage across the area of that market. Nobody would say that a mobile provider has a network effect because they have invested in build-out.

[tmh79]

Rideshare providers experience both network effects and economies of scale in their mature markets.

1) Economies of scale: You hire an ops team of 10 people, they manage a city of 10k riders, to get to 50k drivers (5x), you only need to add 5 ops people to manage the city (1.5x).

2) Network Effects: A city starts with 10k drivers. Wait times on average are 10 minutes for a rider, and during rush wait times grow to 25 minutes (less supply elasticity), and the average time from hail to pickup is 12 minutes (IE, app blings the driver, and they are 2 miles away from rider, takes them 12 minutes to drive to rider), and the total pickup loop is hail -> 12 minutes to pickup (unpaid) -> 2 minutes waiting for ride (unpaid) -> 12 minutes on ride, so ~50% of a driver's time on a trip is unpaid. The city grows 5x (50k drivers). Now, average wait time is 3 minutes, and during rush, the average rider wait time is 5 minutes (more supply elasticity), and the average time from hail to pickup goes from 12 minutes to 4 minutes, while the wait time and trip time stay the same, so now the driver is unpaid for only ~20% of a trip. As the percent of time that a "driver is being paid" goes up, the ridesharing fares can go down, to attract more riders, until a sort of equilibrium is reached. The benefit to the consumer of more drivers is lower wait times and lower fares (because you're paying for less driver 'non productive time').

[aetherson]

1. Chip-sellers want to have chip-eaters to purchase from them. Chip-eaters want to have chips of known characteristics readily available to purchase. The more chip-sellers of Doritos, the easier it is for chip-eaters to reliably get Doritos, and the more chip-eaters will become Doritos customers, which in turn incents more chip-sellers to carry Doritos.

And, I mean, that's true. It really is an advantage. But it is again, obviously absurd to describe Doritos as having a network effect. A network effect means a scale advantage over-and-above the effect that is enjoyed by everyone with a big brand and built in demand.

2. Uber Pool and Lyft Line have a genuine network effect, that's true. But as far as I can tell, they're just not compelling products.

[linkregister]

Many drivers bypass the network effect risk by running both Uber and Lyft apps simultaneously. If a driver prefers Lyft but doesn't get frequent riders with Lyft, he or she can still get some riders.

A rider can do the same by attempting to get a ride with their preferred app, then the dominant one. It's annoying, but surely possible.

How many Duck Duck Go users say "I search with the default search, don't find what I want. Then I do !g and I find what I'm looking for."

They're still making the most of the non-dominant platform.

[prostoalex]

As long as "network" is defined as "single-city network" - sure. But being the best ride sharing network in SF does not guarantee you Beijing or Moscow or Barcelona.