What makes a game last a generation?
Problems that aren’t actually solvable. Instead, players can only approach optimality. This means there’s always another hill to climb in terms of increasing skill, so people keep devoting the time.
These tend to be problems that fall into high complexity classes. In general, NP-HARD problems that we solve using heuristics make for long-lasting games. Mind you, these problems need to be intrinsic to the core game loop. I refer you to my presentation on that here: Games Are Math slides.
No end in sight for problem variations. New problems using the same ruleset is also a way to give hills to climb. (Yeah, this means that “authored” games with fixed levels are almost certainly not going to endure in quite the same way. A narrative game is very unlikely to last a generation.)
The typical ways of providing apparently endless content are:
- a decently large permutation space. We have an enormous ability to prune possibility space in our mental models. Tic-Tac-Toe is small enough we solve it pretty readily. In contrast, there are a lot of possible games of go.
- a human opponent. Humans add in a whole new set of problems that are also inherently hard, problems of psychology and status.
- procedurality in problem set generation. Every game of Tetris is different. The weather adds random elements to every sporting event. And so on.
Independence from representation. Games that endure a generation or more are ones that are susceptible to the folk process, that embrace the idea of being co-opted by their players.
So: Games that endure are ones that survive reskinning, graphic design changes, etc., in order to ensure viability over cultural shifts. Think chess sets; or the way in which Monopoly stays relevant in part because of silly Monopoly reskins.
It also, critically, includes ease of replication. Battleship endures in part because it can be played on a napkin. Games with finicky custom pieces are less likely to survive. Games tied to an input method on a specific technological platform, way less likely… already we see Wii games with custom peripherals that are almost unplayable. The more things tying the game to a particular incarnation or instantiation, the less likely it will be to endure.
Sometimes platforms are extremely widespread, like the standard deck of cards, or round balls. If you can build on something like that, which is always pretty available, you can leverage its ubiquity.
Accessibility is deeply related to this. A game that is too complex or demanding to learn will provide too high a barrier for entry. Usually games like that are coupled with heavy dependence on representation or technical platform.
External support, via either ongoing marketing funds or cultural support in some fashion, to prop the game up when it might otherwise fade. For example, Monopoly endures not only because of the ongoing marketing, but also because of the social tradition of family game night. Chess survives in part because being good at chess is a status symbol and cultural marker. Many sports are deeply tied to a given culture’s self-image, rituals, or educational systems. In this sense, games aren’t any different from any other cultural artifact — the more they are woven into life in various ways, the more likely they are to become a custom.
Luck. Games endure in part because of popularity, and popularity is itself heavily driven by luck. Studies have shown that “quality,” apart from being kind of an arbitrary judgement, has a relatively weak influence on overall popularity. Social confirmation is the primary driver of popularity. If a game does not pop up to high popularity early on, it likely won’t stick around. And note, popping to high popularity isn’t itself a guarantee of future endurance (see Where does popularity come from, or the Wisdom of Crowds revisited)
For an alternative view with a lot in common with my answer, you may want to read Dan Cook’s article here: Gamasutra: Daniel Cook’s Blog or check into some of the design work of Frank Lantz.
This post originated on Quora…
27 Responses to “What makes a game last a generation?”
Sorry, the comment form is closed at this time.
RT @raphkoster: What makes a game last a generation? http://t.co/P6dJE8Qr8F
@raphkoster constant challenge once you have maxed all levels. Constant change in end game…..as in no end game.
RT @raphkoster: What makes a game last a generation? http://t.co/P6dJE8Qr8F
RT @raphkoster: What makes a game last a generation? http://t.co/P6dJE8Qr8F
RT @raphkoster: What makes a game last a generation? http://t.co/P6dJE8Qr8F
RT @raphkoster: What makes a game last a generation? http://t.co/P6dJE8Qr8F
RT @raphkoster: What makes a game last a generation? http://t.co/P6dJE8Qr8F
@raphkoster Great article 🙂 I was left wondering what digital games (other than Tetris) fit the bill? Rogue and its variants came to mind.
> Studies have shown that “quality,” …. has a relatively weak influence on overall popularity
This assumes you have some indisputable and infallible metric of quality to compare against popularity. The lack of a correlation could just mean that the attempts to define quality failed to capture whatever it was that was important.
RT @raphkoster: What makes a game last a generation? http://t.co/P6dJE8Qr8F
RT @raphkoster: What makes a game last a generation? http://t.co/P6dJE8Qr8F
That’s why I put it in scare quotes. 🙂 Usually what gets used for studies on that stuff is proxies such as reviews at the time, critical opinions, and so on.
That said, at least some studies on music showed that given 50 randomly sorted songs, and multiple completely separated audiences, certain songs did tend to bubble up SOME. But the top ten was basically different in every audience — it was almost entirely based on peer pressure.
Other studies show that there are basically only a set number of characteristic music clusters — and it is clusters that go in and out of fashion. So a song that might have been popular in a given decade doesn’t work a decade later because the cluster is out of fashion.
RT @raphkoster: What makes a game last a generation? http://t.co/P6dJE8Qr8F
Adam: bejeweled, snake, light cycles… a great yardstick is “whatever the very first games always popping up on a platform are”
Great article, Raph. I especially like the reference links!
Skeptical of your use of NP-HARD here; it’s a class defined in terms of Turing machines, games run on human brains, the relationship between these two is unclear. Probing the computational limits of the human brain is one of the basic activities of game design, it’s misleading to point to a different category as anything other than a source of inspiration.
Fair point. I’ve written & presented much much longer explorations of the topic of complexity, and at this point I tend to shorthand it with that.
Most measurements of game complexity actually use idealized versions of the game (e.g., unbounded topologies, infinite boards) and therefore do not actually represent the complexity class as experienced in an actual play session. So right off the bat, that’s a huge caveat too.
Given that caveat, it’s still not an accident, IMHO, that there’s a pretty clear correlation between how intriguing we find a game and how it lands in a list of complexity classes (http://en.wikipedia.org/wiki/Game_complexity has a handy list). PSPACE complete games tend to get pushed down to kids.
Human brains are definitely not Turing machines. In fact, for quite some time, the ability of the human brain to intuit the possibility space was a big reason why we tended to beat AIs in games. Eventually, this started tipping over as the entire possibility space of some games was mapped computationally by the AI. With the advent of neural nets, AIs started doing stuff a little bit more like us, maybe?
But given that the brain is so incredibly plastic, it’s really better to state that the games we find compelling tend to be those that each individual perceives as being at the right “complexity” level (in the non-math sense), given their cognitive skill set (by which I mean, the accumulated set of experiences, schema, and skills they have acquired to that moment). Hence games tend to slide down in perceived complexity for a given human over time.
The games resistant to simple algorithmic play (aka “dominant strategy”) are the ones that intrigue the most, because we fall back to heuristics rather than algorithm. We don’t have a computable solution, we intuit and logic our way to interesting approaches, without actually finding optimality. And that is pretty close to being analogous to uncomputability in the Turing sense. Further, given how brain chemistry around fun and engagement seems to work (not that it’s a settled field at all…) this heuristic-building satisfies the conditions of curiosity, belief in a possible solution (e.g., you don’t give up), and sense of possible mastery that results in positive brain feedback via endorphins.
And it so happens that indeed, those games tend to be found in NP spaces in their idealized cases, though often not in their actual as-played cases. In fact, a run through Karp’s 21 Problems finds ALL of them used as fundamental building blocks of games, the kind of tool we reach for without even thinking about it.
So… shorthand, but super-useful in a very practical sense. It boils down to the fact that high complexity class problems make for better game systems.
But — so do lower complexity problems with high speed requirements (pretty much all physics-based gameplay). Or problems usually considered quite outside the mathematical, such as social relationship problems. Those are great examples of where the differences between a Turing machine and us become apparent.
Given advances in “mathematizing” human behavior (such as complex systems science), it’s quite possible that at some point we’ll see even the social dynamics of playing a game moved into the realm of math and actual complexity class. But in the meantime, I’m happy to just analogize all of these to one another and call them Hard Problems, with the observation that what formal math calls hard lines up pretty darn well.
RT @raphkoster: What makes a game last a generation? http://t.co/P6dJE8Qr8F
And it always comes back to adaptability.
I’d be interested to have refs on those studies about quality and popularity. Maybe I’ll dig into it all sometime.
If I was to try to research the topic, my starting point would probably be that quality means “being good of its kind”, and it is more or less meaningless to compare quality across “kinds” or as we’d say about music and games across “genres”.
Also you’d probably have to say that what makes a thing “good of its kind” is going to depend a lot on the “kind”. What makes a good murder mystery is different to what makes a good historical romance, or a good literary novel, although there might be a bit of overlap.
One thing to be clear about is that to be popular a thing will have to have mass appeal. Things that have mass appeal do so partly because they are accessible to a broad audience. Which in itself often makes them looked down on by critics and such, who find the most accessible works to be too undemanding or not original enough, because they have already traveled far down the learning curve of the genre. So using critical ratings as a measure of quality, is going to have some serious pitfalls.
Some inmortal games: Counter-Strike, Warcraft 3, Starcraft, Age of Empires 2 and more
Some of those are already not getting played. I am highly skeptical they will last a human generation.
[…] What makes a game last a generation? […]
[…] Er hat verstanden, dass viele moderne Videospiele in all ihrer Storylastingkeit ganz und gar nicht aufgrund ihres Gameplays interessant sind. Er schaut sich deshalb kostenlos “Let’s Plays” von Titeln wie Heavy Rain oder Uncharted an, die dadurch kaum an Wert gegenüber dem tatsächlichen Spielen verlieren – gegebenfalls sogar gewinnen, wenn die Interaktivität sich hauptsächlich in trivialer Fleißarbeit äußerst. Es stellt sich jedoch auch die Frage, ob diese Herangehensweise im Game-Design auf lange Sicht nicht auf das gänzlich falsche Medium abzielt. Interessanterweise verschwinden auf diese Zielgruppe ausgelegte Titel in der Regel schon wenige Tage nach Release wieder aus den vorderen Plätzen der Twitch-Zuschauermessungen. Langlebig geht jedenfalls anders. […]
[…] der Lage, Spielen tausende Stunden an Wiederspielwert zu verleihen, wo doch in Wirklichkeit genau das Gegenteil der Fall ist. Dennoch ist die schon im Titel enthaltene Grundthese vollkommen korrekt und […]
[…] “What makes a game last a generation?” – Raph Koster […]
[…] Er hat verstanden, dass viele moderne Videospiele in all ihrer Storylastingkeit ganz und gar nicht aufgrund ihres Gameplays interessant sind. Er schaut sich deshalb kostenlos „Let’s Plays“ von Titeln wie Heavy Rain oder Uncharted an, die dadurch kaum an Wert gegenüber dem tatsächlichen Spielen verlieren – gegebenfalls sogar gewinnen, wenn die Interaktivität sich hauptsächlich in trivialer Fleißarbeit äußerst. Es stellt sich jedoch auch die Frage, ob diese Herangehensweise im Game-Design auf lange Sicht nicht auf das gänzlich falsche Medium abzielt. Interessanterweise verschwinden auf diese Zielgruppe ausgelegte Titel in der Regel schon wenige Tage nach Release wieder aus den vorderen Plätzen der Twitch-Zuschauermessungen. Langlebig geht jedenfalls anders. […]