Social Games vs Multiplayer Games: A Commentary on Raph Koster’s Social Mechanics Presentation

[Editor’s note: Guest columnist Tadhg Kelly has an in-depth review, below, covering a presentation recently given by veteran game developer Raph Koster at the Game Developers Conference happening this week in San Francisco. Please note that in order to follow along with Kelly’s commentary, you’ll need to reference Koster’s original presentation. Please view the presentation itself, here, for reference.]

The excellent Raph Koster delivered a long presentation (190 slides) at GDC on Monday on what he called social mechanics. The presentation covered topics such as common social actions that occur in social games and potential innovations in social structure or gameplay that developers could adopt.

Raph often delivers talks about game design issues – especially as they apply to online gaming – and since he has been at Playdom for a while, social games and social structures have very much been on his mind. The slides are presented in a PDF on his site, which makes them quite hard to read, but they are well worth digesting.

I thought that since I’m not at GDC, it might be interesting to write a review-from-afar. I looked at the core of his ideas, many of which center around cooperative gameplay, then raised questions over whether they were appropriate to the social game space. Below, I suggest some alternatives that may have been missed in the presentation, and finally I ask whether social games have a next level of sophistication to them at all.

Initial Impressions

I dislike the term game mechanic. It means anything to anyone, and so it tends to represent only a very loose umbrella of things that happen in games. From an informal perspective (i.e. someone who doesn’t make games), mechanic is just a sales term. You use it in presentations with investors along with discussing your unique selling points, and everybody thinks they understand what you mean.

So from a sales perspective, I understand why Raph might use the word social mechanic but it doesn’t actually mean anything, especially not to game makers. What the presentation actually covers is:

  • Observations on social structures in games
  • Musings on what might work in social games
  • Thoughts on why some games work the way they do
  • Some common action types in games

The presentation is subtitled ‘The Engines Behind Everything Multiplayer’ and traces a path from single player games through to different kinds of collaborative or competitive multiplayer gaming.

I find the emphasis on multiplayer odd. In a number of places (Twitter mostly) I’m reading opinions from developers and players that the next stage of social game development is great connection between players. Everyone from Mark Pincus on down has talked about increasing the social dimensions in their games, but if you actually look at the games that work on Facebook, practically none of them are multiplayer.

Pretty much every successful social game is a parallel game (a variant of the single player game). The typical uses of social aspects in parallel games is purely incentive-driven. As I wrote about in my analysis of CityVille, the key to understanding social in this context is to realize that these games are selfishly social. They’re like a networking event rather than friends hanging out: Everybody has an agenda.

The core of successful social game design thus far has been to focus on the player playing her game effectively alone, and then allowing light, incentive-driven collaboration. Poker is pretty much the only exception to this, and is not even that much of an exception. Poker lets players join and leave tables very quickly, and although they are playing against strangers, the game is actually about building stacks of chips. It is played multiplayer, but in effect Poker is only one remove away from any other role-playing game.

This makes me wonder whether the discussion around bringing more socializing and multiplayer activity to social games is actually a problem that users have, or whether it’s a gap that classical game designers think they see. Should social gaming really mean ‘group play’? Do Facebook users actually care about that kind of interaction? Are we to hear endless riffs on why social gaming needs to be multiplayer and analogies about family board games?

These are the sorts of questions that Raph’s presentation bring to mind.


A brief introduction to the presentation topic is followed by an opening quote:

A good game should focus entirely on it’s single-player aspect first and foremost. Then if it’s a simple game like a shooter or racer, use the remaining time and space to fit in a multi-player aspect to it… In other words, multi-playing should never take away from the single-player aspect of the game. Pure multi-player games really should be few and far between.’

This sets the stage. Raph then proceeds to define single player games. Repeatedly. He starts with: A game that is not played in opposition to, or in parallel with, someone else. (which he thinks describes social games today)

But expands to: A game that is not played in opposition to, or in parallel with, or collaboratively with, someone else. (Which is where he thinks the opportunity lies)

What Raph is basically saying is that a single player social game is played by yourself, but with some crossover to your friends, and he thinks that there is much more potential to be had with that crossover than what we’ve seen so far. Raph is seeing the connections that exist between gamers in the social game space, and then looking at how multiplayer games like World of Warcraft are much more collaborative, and drawing a line between the two.

He draws this line by looking at the mutability of game boundaries. A game is not bounded by its rules, but by anything that can legally affect the gameplay. Training is an example that he gives, by which I think he means either learning to play or practising. If we can accept that training is ‘in game’ but not ‘of game’ then it follows that other things like collaboration are ‘in game’ but not ‘of game’ too.

That, I think, is tenuous.

On the one hand, I don’t agree that training is within the boundary of the game. When learning to play a new card game, players will often play a practice round before they play ‘for real’. The implication is that the frame of the world is being understood, but once the real game is on then consequences and sense of win or loss will start to be applied. Dummy-play, guided-play (such as the intro to My Vineyard) and so forth are the introduction into the game world, but they are done so with stabilizers. The full engagement with a game only really happens when the stabilizers come off.

I also think that doesn’t really mean anything when it comes to understanding the social component of social games. Just because one kind of boundary is mutable doesn’t mean that the other is also, so the syllogism doesn’t really stack up.

My own definitions would be:

  • A single player game is a game that a player plays against a game system alone.
  • A parallel game is a variant of single play, allowing single players to mildly assist each other.
  • A cooperative game is a team variant of single play which allows players to majorly assist each other.
  • A multiplayer game is a game in which players (or teams) oppose one another in short bursts.
  • A tournament game is a game in which players (or teams) oppose each other over extended periods.

Parallel, multiplayer and tournament games are all represented on Facebook. CityVille is a parallel game, Poker is a multiplayer game and Bejewelled Blitz is a tournament. Single player games exist on the platform but seriously struggle to gain traction, and co-operative games are largely absent.

Raph seems to think that co-operative games are the missing link. I see what he’s getting at, but I’m not convinced that it’s feasible. The Facebook experience is built on asynchrony, which is why light collaboration between single players works, but deep collaboration between players pretty much requires synchrony. If 25 people are going to take down the big demon, then we need to be coordinating their attack in real time, but that’s a deeper and more full-screen kind of engagement than Facebook normally generates.

Even many of Facebook’s own advanced features which try to build deeper engagement don’t seem to really take off, so how would deeper multiplayer games fare in those waters?

Some of the 40 Things that Could Happen in Social Games

After the opening discussion, Raph gets into the meat of the presentation. I won’t go through each and every slide, but some I found insightful or disagreeable were:

Races: This might mean a physical race or a winning-goal board game. They are, as he notes, absent from social games. Arguably I’d say this is because synchrony is a factor in race games.

Tournaments: One of the interesting things about tournaments is that the deeper they become, the fewer users they typically have. The Blitz games from Popcap are light tournaments, for example, in that they encourage players to compete for high scores within a week-length time-frame. However the deep football management game Soccer Stars Football has always had only a small audience. I think this is because tournaments form pyramidal structures, and pyramids discourage lower class players from participating (but are great for spectators).

Flower Picking, Dot Eating and Tug of War: The interesting observation here is the different perceptions over how they work between players and developers. It also led me to wonder whether there is room in Facebook for a multiplayer version of Amidar.

Last Man Standing: This is where we start to properly get into synchrony. Whether in real time or turn-based games, synchrony has never scaled well on Facebook. Scrabulous was certainly popular as a novelty, but Lexulous has miserable user numbers and even the official Scrabble games can barely manage 2m MAU between them.

Bidding: Raph notes that no game seems to use silent auctions. And yet we know from eBay that silent bidding is phenomenally powerful. Is it perhaps something to do with the tangible aspect of bidding for real things that makes a game like this work?

Deception: Raph talks about several kinds of deception, bluffing and moderated experiences. Each of these relies on a live effect, i.e. synchrony territory. Games like Werewolf only work because people are in the room, whereas turn-based or long-wait versions (like various forum PBEM versions) tend to experience a lot of player fall-off. Time gaps also give players a lot of opportunity to cheat.

Gifts: The interesting thing about gifting in Facebook games is that it often doesn’t involve transfer. The gift I send to another player does not come out of my stash. The game simply invents it. This means that parallel play has no negative side-effects, and so players can be selfishly social and good for each other at the same time.

What doesn’t work as well is if gifts involve transfer. If they do, then they require a social connection to overcome the loss of stash. Raph tries to say that the gift request is analogous to stash, and so all gifts are actually not free, but I think that’s a rationalization. It tries to prove that gifts are more genuinely social than they actually are, but the reality is they function best purely as incentives.

Reciprocity: The realization that gifts are a source of free stuff then make sending gifts easy (which leads me to wonder whether any player syndicates have formed in CityVille to overcome the punitive Energy requirement).

I think Raph is over-analyzing why re-gifting happens though. Reciprocity is not a social expectation among friendlies (most of your Facebook graph) as it is with actual friends (the minority). So for reciprocity to work it really has to have the ‘no-skin-off-my-nose’ quality that free gifts entail. The implication is that the game is bringing friendlies closer together and turning them into friends, but that only happens rarely. Most of the time, reciprocity has a pure profit motive.

Ostracism: Ostracism in Poker could be deliberately near-bankrupting a player so that they cannot bid any more. However, ostracism is also a perfect vector for optimal players to exclude opponents and otherwise game the game system (as it were), and so it has problems when it comes to trying to encourage player retention.

Trust: Cooperative games rely on trust, as do team-based multiplayer games. Outside of that, multiplayer games are deliberately built to either discourage trust, and most single player games are built to ignore the need for it either way. So while it’s an interesting observation, both of these are (I think) still coming back to that issue of requiring synchrony.

Trade and Contract: What surprises me is that these do not already exist in social games. Many MMOs already have player-to-player economies. For me, this is the most insightful of all Raph’s slides.

Elections: Perhaps this could work, although I think American Idol is not a great example because it isn’t actually a game. Nor is politics. The problem with electioneering is that for most of the players they only have one action (vote) and after that they win or lose. This is why most political games tend to focus much more on process.

Reputation: Reputation only matters if it’s in relation to something that viewers or other players already care about. Bieber fever relates to teenage girls and their boy mania, while Barack Obama is the President of the United States. On the other hand being Level 50 in Mafia Wars means nothing to nobody because it is not related to anything other than the player themselves. This lack of relevance is a critical failing of many achievement schemes, and so the better kind of reward to give a player is an extension.

Communities: These are more outcomes rather than designed. Because the game of Poker is built on bidding and suspicion, it will attract some kinds of players and so a culture around the game will have a certain character. On the other hand, Belote is a different kind of game and will attract a different crowd.

Strategy Guides and Teamwork: Arguably these are repeats of earlier ideas. ARGs are essentially cooperative games but they have serious scale problems, while teamwork is implied as a layer for most of the other ideas above. Teamwork tends to need synchrony though.

Brainstorming From Slides 41-50

As a final note, Raph challenges the reader to fill in the blanks. So here goes:

41. Gating: Gating is the practise of not allowing the player to proceed until he gets some of his friends to sign up to a list. This can mean friends already playing the game or friends outside the game and can be completed through public or notification requests channels. The result is false sociality, in that players will click to help out their friends, but it creates no social bonds.

42. Tolls: A variation of gating is to charge a toll. A toll offers the choice of either going through with a gating requirement or paying 5 of the game’s cash (meaning the purchase-able rather than the earn-able currency) to proceed.

43. Sharevertising: A crummy name. This is the mechanism of using prompts to constantly share either achievements or incentives to your friends. Sharevertising’s purpose is to overcome Facebook’s problem of visibility by getting players to do your free advertising for you.

44. Groupcentive: Not seen too much on Facebook, but the heart of Groupon, groupcentive (I know, I know) is the act of getting friends to collaborate for a prize. 50% off if 500 people sign up is a very powerful motivator. Many games could use this I think.

45. Begging: A combination of gating and sharevertising, begging is obliging players to ask their friends to click-to-help-out. A building may require a special item to be completed (like Hard Wood in FrontierVille) and players have no choice but to beg for it or pay a toll.

46. Sponsorship: This was seen more in the early days of social games (it got banned). It was simply a promise to reward a player if he shared or invited his friends back into the game. Daily rewards, bonus cash or levels and similar incentives were the promise of sponsorship.

47. Visitation: The game turns clicks (‘Energy’) into a valuable resources by billing the player for each one he uses. Then it encourages visitation to other player profiles as a way to earn more clicks.

48. Rental: A form of continuous gifting, players can allow each other to use some of their own territory as a rental space, and both players reap the reward. As with gifting, this requires a lack of side-effects to work.

49. Harem: More of an early-social dynamic than a commonly seen one today, the buying and selling of friends encouraged sub-culture networks to form. Human Pets and Friends For Sale are two examples: Players buy, sell and look after their bought friends, resulting in a sort of harem-like social structure.

50. Indulgence: Finally, the economic model of most social games is based on the selling of plenary indulgence. Players can get ahead but for the price of a few dollars, get exclusive items and more energy, and so earn more status and cover over failure than other mere mortals.

Is There a Next Level for Social Games?

Raph is reaching for the next level, and so he should. Few are the game designers whose only ambition is to create Poker or yet another city simulator, and boundaries exist to be pushed upon.

And yet at the same time, I feel that much of the design talk currently floating around social games (and gamification) are a bit flaky. It’s true that there are many kinds of social structure and it is possible to get very excited about what it might mean, but take a step back and look at what we actually see on Facebook.

A few dominant companies that make fairly ordinary games that coax some amount of virality out of their users and provide amusement-level entertainment. While there is certainly space for amazing things to be happening there in theory, the striking ordinariness of the games is telling.

Facebook seems to be much like television, in that the way the platform works results in a few very top-heavy channels, with many other smaller companies scrambling for a little bit of the juice that the big boys get.

In such a landscape, many long-running games have sat on top of the tree for a couple of years. There are very few breakout hits, and other platforms like the iPhone are much more active on the innovation front.

This leads me to say that while social games and social structures are interesting in theory, Facebook’s visibility problem is discouraging that innovation from happening. The more that I work in the social games space, the more I think Facebook’s lack of dedicated editorial to assist discovery of interesting apps is a primary reason why many of the more adventurous ideas that Raph thinks should be there are not there.

A Wall of Eryx stands between where social games are and what the next level could be, and it’s Facebook’s fault. The theories of social structure that Raph and many other great guys are sharing these days are thus unattractive to developers. Discovery is a problem that’s sitting in Mark Zuckerberg’s inbox, and the real question is what is he going to do about it?

An Irish lead designer and producer living in London, Tadhg Kelly is the author of a forthcoming book about, as he describes it, “Reclaiming games as an art, craft and industry on its own terms. The book is titled: “What Games Are.” The blog for the book is You can also follow his tweets on Twitter (@tiedtiger).