What Games Are: The Fun Boson Does Not Exist

Editor’s note: Tadhg Kelly is a game designer with 20 years experience. He is the creator of leading game design blog What Games Are, and consults for many companies on game design and development. You can follow him on Twitter here.

Back when the social game scene looked like it might be generationally, rather than merely technologically, disruptive, we game makers discovered Eric Ries. Ries (along with Steve Blank) is the key figure behind the lean startup movement, and at the time his message of fast iteration and customer validation rang true for us for two reasons.

First, we were very frustrated working on multi-year-long projects with no clear goal that seemed to be more about some designer’s ego. Second, we saw it as a new way to look at the games business and found that the investment community was inclined to agree. Game companies are notoriously difficult investment propositions after all, so anything we could do or say that promised to manage the creation process better was music to their ears.

During those days you could immediately tell who got it and who didn’t with the use of the terms “minimum viable product” (MVP) and “free-to-play” (F2P). You’d explain this ideal process whereby a tiny team would iterate on ideas quickly. It would measure everything, too. It would offer this kind of game where the initial experience was free, and that would pull in a lot of users, some of whom would become customers. The eyes of those who knew what you meant would suddenly spark in recognition.

MVP and F2P eventually passed into regular industry jargon along with a boat load of other terms. Most every company involved in the space now talks about DAU, LTV, ARPU, ARPPU, ARPDAU and even ARPPDAU. They talk about performing cohort analyses. Some of them ask whether they are working on an MVP or an MDP? Most don’t really bother discussing viral K-factors any more, and instead obsess about the CPA of players. These are significant changes for an industry that used to worry more about Metacritic ratings.

However they are also often misused. In much the same way that every studio claims to be “agile,” but few actually are, most of them miss the point of all these numbers. They get badly stuck when considering their MVP because they realise that they have no idea what “viable” is supposed to mean in the context of games. So they do what they’ve always done, which is to copy the other guy and invent very little.

How We Got Here

It all starts with the delusion of numbers. One of the axioms of the San Francisco Revolution, derived straight from lean thinking, is that you can’t improve what you can’t measure. In other words, if you add or subtract something and it does not cause a key metric to go up in some significant way, then that change was meaningless.

This axiom is seductive because it promises to expose the game and stop it being treated like a mysterious black box. In theory it’s supposed to unlock a whole wealth of innovation, because we could then know a great deal about how players behave and think, and then use that. Measuring to find an outcome that might scale is, after all, what the entire lean method is about.

When Facebook developer garages were interesting places to be, the sector got very excited by stories that seemed to prove this assertion. There was the story of the Christmas tree put into the game which sold a million units in a week. There was also the story of the object suggested by the community and then put into the game not three days later. All of these emotive images seemed to validate the validation, but the energy around them didn’t last. Now every game conducts regular sales, every game has its holiday boondoggles and none have really taken those ideas to a next level (if there is one).

In practise what the validation-led method actually turned out to be was a sanitised version of age-old processes from the gambling industry. Personally I have no problem with the gambling industry (as long as it behaves responsibly around addiction), but its tendency toward validation of everything means it tends to only focus on a couple of key game formulae that are proven to work. That’s why every casino is identical. That’s also why every social game maker is identical.

Rather than continuing to innovate through measurement, the social sector as a whole rationalised itself into a corner. It knew of a couple of formats of game that seemed to work with measurements (but not really why they worked), knew how to build those, and then continued to repeat the same format again and again. So, just like the gambling industry, social gaming became about who had the best commercial processes in place to push their identikit product around as fast as possible. Farmville really wasn’t about Zynga’s genius at replicating Harvest Moon. It was about their genius at getting that game in front of everyone on Facebook faster than anyone else.

But, again just like the casino business, that kind of thinking can only get you so far.

Local Maxima

The tragedy of social games is that the companies involved discovered the greatest distribution tool in the history of the industry, and yet proved inept at providing great games to go with it. The best things that they’ve come up with so far is Poker, some socialised versions of simple casual games, some super-simple sims and about 100,000 variations of Dungeons and Dragons.

In part that’s because of market conditions. Facebook proved pretty tricky to understand, as some developers devoted almost their entire energies to overcoming visibility issues. All those notifications and cross-promotion obsessions happened because users didn’t really remember the names of the games they were playing, nor how to find them.

It’s also because of technology. PHP games like Mob Wars could do little more than be static click-object games, and while Flash could handle sims well, it was (and still is) very weak for making action-oriented games. Arguably the sector needed a better technology to work with, which in time it got through iOS. The weird thing, however, is that social developers are still making the same limited games. What’s happening on iOS with Supercell is really just a repeat of what happened on Facebook three years ago.

Mostly it’s about development culture. The thinking behind social games is not unlike the thinking behind television. The bean counters in TV land tend to think that there is a number, perhaps not yet discovered, that will one day explain television viewing to them. They believe that attaining viewers is a process, entertaining them is a process, and that if only the right measurement and formula can be found, television would become a predictable industry.

In the absence of that number they look at ratings, demographic data and viewing patterns and try to infer what it might be. They build products based on that inference, to make shows which satisfy those numbers. And when that doesn’t work they fall back to copying other successful show formats and trying to put a spin on them, just like casinos do. And that culture becomes circular and inward-looking over time, so eventually that’s all they know how to do.

The ultimate fallacy of sticking with “you can only improve what you can measure” is that measurements eventually determine all of your creative decisions. I’ve lost count of the number of times I’ve heard a designer complain that they need to enact a deep change in their game, but are not allowed to do so by a manager who demands it be proved with numbers first. Both understand that something is not right, but the axiom mandates that the problem must be expressed in numeric terms, or else it does not exist. The resulting cognitive dissonance leads to frustration and formulaic decisions, and so the game becomes like every other game.

That’s called being trapped in a local maximum.

Measuring The Wrong Thing

Obsessed with measuring everything and therefore defining all of their problems in numerical terms, social game makers have come to believe that those numbers are all there is, and this is why they cannot permit themselves to invent. Like TV people, they are effectively in search of that one number that will explain fun to them. There must, they reason, be some combination of LTV and ARPU and DAU and so on that captures fun, like hunting for the Higgs boson. It must be out there somewhere.

Watch any sport or exciting board game, some gamers playing Call of Duty or Angry Birds, and you start to notice how there’s a certain yin and yang of play. There are many small decisions of little consequence, but they tend to bounce off one another and lead up to bigger moments. In great games this seems to have a pattern, and we often try to describe this in terms of “mechanics.”

Once you sidestep some of the more feverish interpretations of what a game mechanic is, they’re actually pretty easy to understand. A mechanic is an action on the part of, or a rule that affects, a player. When not looking at large meta-metrics like DAU, studios typically measure instances of mechanics. They look at how many players buy an object, level-up within a certain time frame or use an item in-game. Then they try to improve along those vectors, yet for some reason the overall fun of the game seems lacking.

Now to be clear, there are many arguments to be made for soul, culture and the importance of building an identity in a game that tells a marketing story, but this argument is not about those qualities. By “fun” I have a very simple definition (“the joy of winning while mastering fair game dynamics“).

Trying to measure and improve a game through only studying mechanics is like trying to improve tennis solely by measuring how many aces occur, or how many foot faults happen. Those numbers are very useful in many ways, but they are not the game. If, for example, you wanted to increase the number of aces in the game, then perhaps you might lower the net, lengthen the racket or change the type of ball used. That change would have massive knock-on effects through the game, however. And it’s quite likely that you would get more aces but make a worse game.

It is in the interplay of various mechanics that the dynamic of tennis emerges, and that dynamic is surprisingly sensitive to small changes. The dynamic nature of fun is always like that, and is why good game developers often talk about the importance of “finding the fun” in games through prototyping.

Games are essentially chaotic systems. In a chaotic system you can look at the initial starting conditions and the topology of a simulation and try to predict the interactions of strange attractors as much as you like, but you generally don’t really get what the hell the simulation is doing until you observe it in motion. The system is too sensitive, the patterns too hard to interpret and the situation too emergent.

Fun games are a little bit like those chaotic systems that produce beautiful fractals. In some games the balance between all the mechanics produces an inherently exciting set of outcomes, but are hard to predict just from looking at their rules. They have to be played to see what does and doesn’t work, to be genuinely iterated upon in the true sense of the lean startup (not just built) and allowed to be validated in their dynamics. Everything else is just nonsense.

Until you prototype it and try it, you really just don’t know whether a dynamic is fun or whether it produces beautiful outcomes that compel you to play again. Measuring mechanical instances may help you do that, but you still need to accept that there’s an x-factor involved.

Unlike every other major game revolution (arcade, console, PC, casual, MMO, etc.), social game developers have proved consistently unable to understand that fun is dynamic in this way. This is why there is, as yet, no social game that has achieved the genuine love and admiration of a World of Warcraft, an Angry Birds, a Plants vs Zombies or a Super Mario Galaxy. They are hunting for the fun boson, but it does not exist.

This is the hardest lesson that they need to learn if they are to get to generation-two.