Network Effect

The network effect is one of those critically important, foundational concepts needed to discuss the success or failure of multiplayer games intelligently, one that gets more “what?” reactions than it should. I’m hoping people know it but perhaps not by that name; knowing it by name lets you tap a century of research and discussion rather than re-inventing the wheel.

The idea is that adding someone to a network creates value for everyone else. If you have no friends on Facebook, it is just a platform for solo Farmville. Every additional user creates additional potential value for every other user. In many of our online games and social network, the primary value we are seeking is that connection with other users. It is not just that users create content; users are the content.

Via network effects, winners keep winning. You play World of Warcraft because all your friends play it, and your friends play it because all their friends do. Hopping games means losing that huge network, which is why you are trying to talk your entire guild into making the jump to the new game (take the heart of your network with you).

“Critical mass” is the first major sub-concept you want. There is some threshold at which a network becomes truly valuable. If you cannot consistently find a group, your game lacks a critical mass of players (or you’re an asshat). Once you have critical mass, you start seeing all those benefits from additional players.

Good blog topic: tiers of critical mass in your game. Is it solid during peak hours but not as playable for an Australian? Does your game need a low-level redesign like LotRO just had because there is no longer a critical mass at the lower levels? How much does it suck to reach the level cap and wait for that critical mass to develop? You could also run with game lobbies, where they either fill quickly (gogogogoogogog!!!) or flounder at 1 or 2 people as other pop in, wait 30 seconds, and leave. A Team Fortress 2 server I often enjoy engages in the contemptible practice of always displaying itself as almost full; it is a self-fulfilling prophecy, because why bother being the 3rd person on an almost-empty server, while getting the last slot on an almost-full server ensures a lively game.

If you hate WoW, lock-in is the next sub-concept you want. Once the major value comes from the network, what the network is organized around matters less, so crap that seemed like a good idea at the time sticks around. Reference hatred of everything Microsoft makes: if only we could get everyone to switch to Unix, the world would be saved, but until we do, people keep getting Windows because everything is made for Windows, and everything is made for Windows because everyone has it.

A related topic is that the effect can end hard and fast when a new network replaces the old. If everyone is on MySpace because everyone is on MySpace, when you think everyone is going to Facebook, you might see your entire network flip in an avalanche (and you took your whole guild with you). Again: self-fulfilling prophecy, potentially with a new lock-in.

The interesting point there is what it takes to overcome lock-in. WoW had to be far better than EQ to replace it, although WoW seems a special case since it expanded the market by an order of magnitude. If you want to take over WoW’s place as the default game, you cannot just be 5% better. You need to be better than WoW+network even before getting your own network effect. This is hard.

Because MMOs separate players, the network effect differs a bit. WoW may have the biggest network, but it has many sub-networks because you cannot reach your friends on other servers or across Horde/Alliance. Paid transfers mitigate that. A Tale in the Desert (until recently), EVE, and Darkfall have smaller but proportionately more powerful network effects because everyone is on the same server, so you really do have access to everyone. Segregating by level reduces your network effect because you still cannot play with your friends; alts, sidekicking, and flatter leveling curves reduce this, and at least you can send your newbie friend some toys from your level-capped character.

Negative effects can exist past the critical mass, notably crowding. Instancing is intended to evade that, but you have certainly played in an area that has too many people in it (and all those other guys are stealing your kills). Larger networks are also more attractive to griefers, trolls, and other people your system needs a way to deal with.

Good blog topic: when are we going to see games that take advantage of interoperability to increase their effective network sizes? You see publishers with stables of games, but they have few interactions. If I can take something from Game X when I go to Game Y, Game Y is a lot more attractive, and I am also more likely to go back to Game X with my toys from Game Y. (I need to find the Charles Stross book that uses this.) Every veteran achievement-whore would love to show off his (our) trophies from the last game, to further than illusion of permanence, create a feeling of continuity, and prove that you are not some newbie idiot. Have you seen someone pick up a Steam or X-Box game just to add more achievements to their list? Some games are advertising that harder than gameplay. Linking networks increases their effective size, and while many software developers are taking advantage of that, MMOs remain mostly separate bits of art. Could this contribute to the rise of Facebook “social games”?

As the network leader, interoperability seems like something that could only hurt Blizzard by aiding others’ networks, while they gain by forcing the choice between the biggest network and all the rest. If Blizzard were more pioneering than adapting, however, I could see them trying to link their next MMO to WoW. Keep your achievements, cross-game chat, pre-reserved guild tags, that kind of thing. If you want people to try your new network without necessarily killing the old, you can connect them. It will at least funnel people from the old to the new without risking losing as many to competitors.

I think I have gone on enough. You could get a post as long as this one from any of the sub-topics within. Please, take one and run with it. This is not just an educational opportunity; once we have a critical mass of readers who understand network effects, you can refer to the topic without rambling on at this length, taking advantage of the existing MMO blog infrastructure.

: Zubon