It’s a buzzword that every venture capitalist and many entrepreneurs use. But what, really, does it mean?

When it was first described to me, my understanding of the network effect was simply that it was like the telephone — the more people have phones, the more utility of any single telephone in the system. This is a self-reinforcing mechanism that makes for great defensibility in businesses. As I wrote about a while back, there is an example of a negative network effect too, where the value of an additional participant in the system actually worsens, rather than improves, the system for every participant.

There is another part of the network effect, at least as its commonly described. To continue the telephone example: at a certain point, the fact that everybody else uses a telephone is why I want to get one. In this case, the system has something of a snowball effect. The existing momentum and scale creates more momentum and more scale. Some refer to this as the bandwagon effect. 

Lyft* and Uber are interesting case studies for the network effect, because their dynamics are actually not what meets the eye. While their global brands create a bandwagon effect — because everybody uses Uber in London, I thereby ought to be using it, even if I’m in San Francisco — the local city dynamics have a positive network effect at the beginning, but I wonder if they perhaps actually have a negative network effect as they scale.

As a driver, it’s actually not great for more drivers to be on the system after a certain point, because that reduces my potential earnings. As a passenger, the more people that use these apps, the more likely I am to get screwed during peak hours. This is the nature of most marketplaces, which have to balance the conversion of demand and supply as they grow. But an interesting factor to consider, along those lines, unlike the telephone, which I can’t really switch off of once everybody else is on it, the value of being an UberX driver is tied to the availability of demand (and associated earnings), but there is nothing that functionally ties me to UberX, versus Lyft. The switching costs are very low, and stay equally low even as the system is liquid and dense. As a result, I have heard anecdotally that many UberX drivers also have a Lyft driver account, and switch between them liberally. On eBay, or Yelp, or AirBnB my seller rating is highly tied to my ability to move my inventory. This creates a strong incentive for me to stay on the platform, particularly if I have a good rating. In this way, then, the ‘rich get richer’ and there is a positive network effect tied to the platform’s rating system. Since driver ratings on Lyft and Uber change so often, and resets after a few weeks, building one’s rating from scratch isn’t ultimately that hard, as compared to the other platforms.

Lyft does a very good job of creating community on the supply side (by allowing its drivers to communicate in-between rides, by encouraging online and offline interaction between driver communities, etc.) which creates a positive network effect against the experience of being an Uber driver. I wonder, though, if that means simply that a Lyft driver is more likely to keep the Lyft app open, or if that’s enough of a strong incentive to not even have the UberX driver app.

Anyway, as I hope this post suggests, the nuances of a peer network’s defensibility are not so simple as they may seem.

*Lyft is a portfolio company of Collaborative Fund.