Networks — Part 2: Network Effects

In Part 1 of this series, we discussed the evolution of networks; if you didn’t get a chance to read it, you can find it here. In this article, Part 2, we explore how we think of “Network Effects” and the unique way that they serve to sustain or slow the growth of networks — both directly and indirectly. Next up after this article will be Part 3: Operator vs Open Networks.

So what are network effects?

Networks suffer from what’s often called the cold-start problem. They aren’t very valuable early on when there aren’t many participants; who wants to go to a party that consists of a few people hanging around a punchbowl and staring at their phones?

The challenge is to get to a network to a critical level of participation, where the network effects exceed the cost of belonging for each participant (in money or time or effort).

Getting to this point establishes a positive feedback loop commonly referred to as the “network effect flywheel” — a situation where network effects alone sustain the network at a steady or even accelerating rate.

Businesses take many different paths in pursuit of this state. A typical early strategy is to target a defined niche of particularly engaged and connection-inclined individuals. Google, for example, made its initial inroads with tech-savvy early adopters. Amazon focused on book readers. Facebook launched on elite college campuses. “Drafting” on existing networks to launch new ones is common as well: PayPal rode the rise of eBay so successfully that the latter ended up acquiring it. And nearly all network-centric startups offer free or heavily subsidized access while jump-starting their businesses, in search of the scale needed to drive the flywheel: Uber’s free rides and $1000 payouts to new drivers; Netflix’s 30-day trial memberships; and the most recent and outrageous example, video network Triller’s efforts to lure influencers away from rival TikTok by offering them partner contracts worth millions — and free Rolls-Royces.

But even these examples highlight the fact that there are complex layers to how network effects work, especially when multiple different types of participants are involved.

First & Second Order Network Effects

There are two layers of network effects: First order, which represent the direct impact of network growth on existing network participants, and second order, which are essentially the impact of network growth on network growth itself. By way of example, let’s take a look at multiplayer online games, such as Riot Games’ League of Legends.

The first-order effect of an additional player joining the game is reduced wait times to start play. That’s because more players means greater likelihood of matching with an opponent.

A second-order effect comes from the fact that potential players who see their friends playing the game are encouraged to play as well. The growth of the network makes the game more appealing. As they join the game, they further decrease wait times for matching, which reduces friction and makes games more fun.

Same-Side and Cross-Side Network Effects

But the example of League of Legends is a relatively simple network. It’s what one might call “single-sided,” because there’s just one type of participant in the network: The players.

Many networks are multi-sided, with multiple types of participants with different roles, agendas and incentives. Marketplaces of all kinds are a good example: In most marketplaces, there are two distinct groups — people who are trying to sell things, and people who are seeking to buy things. To understand how network effects impact a multi-sided network, one must consider not just the effects within a group, what one might call “Same-Side Network Effects,” but also the impact of network effects between groups — “Cross-Side Network Effects.”

To illustrate same-side and cross-side network effects, let’s take as an example rideshare services like Uber or Lyft. The same-side effects are fairly obvious: When there are more riders, the impact on other riders is negative. The greater demand leads to surge pricing, which can cause subsequent riders to choose cheaper alternatives. When there are more drivers, the impact on drivers is negative. A greater supply of drivers causes rates to go down, which means each driver receives lower fees.

The cross-side effects are also straightforward: More riders and surge pricing has positive effects for drivers, because they make more money. And more drivers has positive effects for riders, because they’ll pay less for rides.

What about second-order effects?

Second-order effects are a little more complicated to analyze, but ultimately point to how different types of participants have different outcomes as networks grow. In this case, second-order network effects generally create a virtuous cycle for riders, while they cause the number of drivers to eventually approach an equilibrium. As long as the positive second-order effects on riders are greater than the negative first-order effects, the number of riders in the network will continue to grow.

Generating network effects in games

Let’s take a look at an extremely successful example of a game that has grown significantly based on network effects: Roblox. Roblox is a game that allows its users to explore different experiences, ranging from mini-games to immersive worlds — all of which have been created by other users. Players can simply play the experiences in the game, most of which are free (but with the ability to pay for upgrades). However, they also can essentially become developers themselves, using the platform’s tools to design experiences for their friends or for total strangers.

What makes Roblox’s business model so interesting is that any money spent on user-generated experiences is split with their user-designers. Savvy designers can earn a significant amount from successful creations.

All of this makes Roblox a very interesting multi-sided network: One in which the boundary between the sides is blurry and permeable.

Here’s an analysis of the first- and second-order network effects for Roblox:

It’s easy to see from this why Roblox is such a hit. The Roblox team have focused on making it open as possible — encouraging players to see Roblox as a community to which they belong, not just a service that they pay for — and, most importantly, returning value to the network in the form of direct payments to designers. The incentives for people to participate in developing and growing the network are thus explicit, and they continue to grow as more people join the network, at least until a hypothetical equilibrium is reached when there are too many experiences for people to play and the majority of those experiences are being ignored. But because many designers of experiences are also players, spending time engaging in other designers’ experiences, the bar to reach equilibrium in the designer community is set higher as a result.

Meanwhile, the network effects for players are all positive, with more players and more designers both encouraging more players to join until the positive second-order effect kicks in where Roblox isn’t just seen as a game but as a social platform — a necessary utility for people to stay in touch with their friend circles and meet new people.

Network effects play a huge role in the scalability of virtually any platform in the digital era. Effectively managing first and second order network effects to ensure that the impact of positive effects outweigh negative ones is a critical factor for success.

In the third part of this series, we’ll talk about two types of networks — “Open” and “Operator” networks — how they differ, and how those differences can open them up to contrasting vulnerabilities.

Interested in contributing to our Community Economics series? We’d love to hear from you. Comment below!

Follow us on Twitter @FortePlatform