[mwsherman]

The network effect of ride-sharing is considerably weaker than that of a social network. A social network’s effect is O(n²), as that is the number of meaningful connections.

Ride-sharing networks are O(n). They benefit from liquidity, but I derive only secondary benefit from being “connected” to others on the platform. It's not nothing, but it feels linear, not superlinear.

Ride-sharing apps are just apps, and the barrier for adopting a new one is low, for both sides of the market. Further, Uber’s political operations benefit the other apps equally.

Uber can defend its brand and its satisfied users. Size helps here. Think iTunes. But the “network” is replicable by others.

[lpolovets]

IMO you can have network effects that don't have to do with connectivity. According to Wikipedia, a network effect is "the effect that one user of a good or service has on the value of that product to other people."

In this case... - more passengers lead to more drivers - more drivers lead to shorter pick up times - shorter pick up times lead to better UX/value to passengers

So even though passengers aren't connected to each other, additional passengers significantly increase the value of Uber for existing passengers. This is actually very hard to replicate for new entrants.

For example, let's say when Uber started, there were 20 drivers and 1000 passengers, and the average pick up time was 6 minutes. If the average ride was 9 minutes, then a driver could do 4 rides per hour. Those 4 rides had to provide a decent wage -- let's say $20/hour. That means each 9 minute ride was $6.25 ($1.25 for Uber's take rate, $5 to the driver).

Now, let's say 4000 additional passengers signed up over a few months, which led to 80 additional drivers. Now, because there are more drivers all over the city, the average pick-up time might be 3 minutes. That means a driver can now do five 9-minute rides per hour instead of four. That means each ride can now cost $5 instead of $6.25, and the driver still makes $20/hr.

For passengers, this is awesome: their wait times got cut in half while the cost of a ride dropped by 20%. For new entrants, this is awful: the two main levers for competition are cost and waiting time, and Uber broke both of those levers. New entrants can still match fast pick up times and low costs, but they have to bleed much, much more money to do so when they launch in a new market.

In a way, Uber's network effects are like economies of scale, but they're typically classified as network effects because the means of production come from more drivers and passengers signing up, not from Uber doing any "real" work itself.

[MarkPNeyer]

> This is actually very hard to replicate for new entrants.

I've seen many drivers running both lyft and uber at the same time. A third company comes along that pays drivers better? Why not keep all three apps up. There's no cost to running an additional app. Literally none at all. And if it pays slightly better, or has a fairer reputation system...

The same thing works on the rider side as well. If you've got an app (like what's the fair) that can let you query any number of providers and get the cheapest price, maybe you're happy to wait a few minutes to save money.

[lpolovets]

The problem is that Uber/Lyft make it so that they are cheaper AND faster than a new entrant. I think if a 3rd company came along and it paid drivers better, then drivers would install a 3rd app. But Uber/Lyft's network effects makes that a very expensive proposition. For example, if an Uber driver gets $5/ride and can do 6 rides/hour, they make $30/hour. If a new entrant can only offer 4 rides/driver/hour (at $5/each) because there aren't enough riders using its app yet, then either the riders or the new entrant have to subsidize drivers the additional $10/hour. Riders won't pay extra -- why would they when they can use Uber instead? -- so the new entrant ends up paying 1/3 of each ride to get the network effect going, and that can be quite expensive if you're doing thousands of rides per day in a single city.

There are some complications to consider. Maybe a 3rd company can pay drivers a little less because it treats them more nicely than Uber does, or maybe riders will be willing to pay more or wait more for a 3rd company's cars for some reason (e.g. Lyft's tends to be friendlier while Uber is more formal, and some users are willing to pay more for the former). But in general the more passengers and drivers an app has, the better off both sides of that app's market will be.

[gnicholas]

True in theory, but incumbents can (and do) reward drivers for completing lots of trips. This makes it easier to lock out a new entrant, especially if they're attacking city by city. Just ramp up the rewards for drivers who complete X trips per week, and you can fend off the new guy.

[mkagenius]

Someone if goes bankrupt and then making the app open source and not a taking cut at all..can it kill Uber eventually?

Or the tech benefit of actively working people in Uber will be much better ROI for drivers to give a 20% cut -- which other industry gives such high a cut? (airlines or movie ticket booking doesnt)

[JoshTriplett]

> Someone if goes bankrupt and then making the app open source and not a taking cut at all..can it kill Uber eventually?

Uber and Lyft provide a substantial amount of infrastructure and staffing; they wouldn't function without that. Someone has to keep the server infrastructure running and scaling, and even more importantly, someone has to maintain the support, rating, and qualification system to maintain quality.

[sharemywin]

Do more drivers lead to shorter pickup times or more drivers without a passenger.

Also, I did some math and it takes n^2 drivers to half the distance between drivers areas.

Also, I think as the food delivery companies get bigger it all starts to merge into one industry. Imagine some kind of loyalty program were if free ride or delivery after so many meals delivered.

If you had the drivers out delivering flyers for the local food companies they could make money while waiting on a fare.

ebay and amazon have network effects because it has so many different items for sale, not just drivers.

[lpolovets]

> Do more drivers lead to shorter pick-up times or more drivers without a passenger.

That's an interesting question. I think in this case it's shorter pick-up times because the system can somewhat self-correct in real-time:

1) If a someone is thinking about going out and driving, they check the app, see that there are lots of drivers out and not many passengers, so they might decide not to go out at that moment.

2) If there are too many requests and not enough drivers, Uber increases surge pricing. The greater the increase, the more drivers come out and the more requests get withdrawn until an equilibrium is reestablished.

Finally, there's some data at https://newsroom.uber.com/uber-expectations-as-we-grow/ that shows that the longer Uber is in a city, the more pick-up times drop and passengers expect shorter pick-up times.

I agree with you about food companies: at some point they might become competitive with Uber, or they might outsource logistics to Uber while they handle the food ordering and preparation aspects.

[JoshTriplett]

> The network effect of ride-sharing is considerably weaker than that of a social network.

If Uber actually did ride-sharing, in the social sense, it would have a significant network effect: it would get more valuable for everyone the more people used it. But Uber isn't "ride-sharing"; it's just a normal service business, with no significant benefit from network effects.

[deciplex]

> Ride-sharing apps are just apps, and the barrier for adopting a new one is low, for both sides of the market.

It's easy for two people to start using a new social network as well. That doesn't mean they'll want to do it if they're the only two people on it.

More Uber users -> larger market for Uber drivers -> more Uber drivers -> higher availability of drivers for Uber users -> more Uber users -> etc etc

> Further, Uber’s political operations benefit the other apps equally.

They could push for monopolies at the municipal level, much like cable companies do.

[mwsherman]

Also, FWIW, the article really doesn’t explain much, and might be plain wrong. It’s far from a zero-sum game, the overall market has very little ceiling in the foreseeable future.

[adrianN]

Social networks are sparse graphs with a linear number of edges.