Monday, December 9, 2013

Against High Skill Ceilings

Max writes about the common assumption that taking forever to master is a desirable design feature in board games, and why this assumption is wrong.  Easy-to-learn, hard-to-master games like Chess and Go are great, but they're not what most gamers want.

'Skill ceiling' is a term that refers to the maximum skill level that can be achieved by a player in a game.  There are two parts to this concept: how much better a player at maximum skill level is compared to new player, and how long/how many games it takes a new player to reach maximum level.  Often in discussions of skill ceiling, these two are not separated from one other, so a game with a high skill ceiling generally means one where the veteran player is much better than the new player, and it took the veteran many, many plays to get to her skill level.  High skill ceilings are created by increasing the amount of complexity of strategy there is in a game.

Why Is High Skill Ceiling Desirable?
High skill ceiling is often discussed as an unequivocal good.  This is because it can add several things to strategy games.  First is replayability.  So long as players feel like there's more about the strategy to discover (and thus more skill to gain), they may be motivated to play again and again to increase their skill.  This can lead to a feeling of mastery as players gain greater understanding of the game system.  Finally, if repeated plays and learning the strategy can lead to a veteran player beating a complete newbie more of the time, the game is by definition not very luck-based.

Because having a high skill ceiling can ensure these good qualities in a game (high replayability, and low luck), it has tended to always be praised as good design, and not fully considered a design choice.

Why Is It Not Always Desirable?
The first and most obvious reason is that certain audiences don't enjoy the level of strategizing that they will need to do in order to progress at a game with a high skill ceiling.  For example, party games rarely have and shouldn't have high skill ceilings.  They should be easy to pick up, play, and enjoy.  The players aren't playing to achieve mastery of the system, they are likely playing for one of numerous other reasons.

But even within the 'hardcore' board gaming audience high skill ceilings have drawbacks.  In order to have a high skill ceiling, games often have to have a good amount of complexity inherent to the system.  Even among strategy board game players undue complexity isn't desirable, as it can be overwhelming.  If it's warranted it can be great, but a game can have a high skill ceilings because it's simply too hard to understand.  This is, of course, a particular issue during the first few plays of a complex game.  High skill ceiling by very definition also increases the likelihood that in any given game there can be players of vastly different skill levels, which sucks for everyone - the veterans feel cheap if they win, and novices feel like crap because they always get beaten.

My biggest problem with games with very high skill ceilings, however, is that it can easily result in copycatting.  Take a look at three games that have extremely high skill ceilings: Chess, Go, and Magic: the Gathering (building decks, not so much each game).  One thing that each of these games have in common is that it's faster to excel not by playing a lot and experimenting with new strategies (as is the case with many board games), but instead by doing research.  In Go new players are often taught through "problems," board scenarios with correct solutions on how to win them.  Chess players are taught gambits and can investigate pro games.  Magic players "netdeck" by copying winning decks from previous tournaments instead of inventing their own.  Some players really enjoy all these processes, but many (including myself) find that they undermine the fun of the game.  High skill ceilings directly drive players to resort to this mimicry and away from discovering things themselves because in experimentation to increase skill level takes much, much longer (by the very definition of high skill ceiling).  This mimicry is bad because, while it can help players progressing more quickly to a certain level, eventually not being able to think for themselves takes its toll and they have a hard time progressing further.  In addition, it leads to stagnating metagames, as I've written about before with Dota 2.

So how can we avoid the drawbacks of high skill ceilings while ideally capturing the good elements?

3. Bracketed Play (Chess, Go, Magic)
To avoid pitting players of highly unequal skill levels against each other, many games with high skill ceilings do their best to bracket play, ensuring that matched players are on the same level.  At competitive levels this is easily achieved through ranking and elo systems that allow organizations to quantify skill level.  Fortunately, even before reaching competitive levels, the way players are introduced to a game more often than not can organically mitigate this problem; especially with games like Magic, friend groups tend to learn these games together from square one, helping to have players progress in skill level together.

Obviously, formal matchmaking doesn't work for most games as most players don't play at tournament levels.  The organic process of learning a game together is what allows strategy games to work at all, but it can fail if players frequently play with new people - which is a thing that designers should be encouraging, not discouraging!

2. Specialized Play (Dota, World of Warcraft, Team Fortress 2, Sentinels of the Multiverse)
One of my favorite solutions to this problem is allowing play to be compartmentalized such that a player can learn (and master) a specialty relatively quickly and play the game using that specialized character or role, but the game still has tons of replayability as that player can then spend just as much time mastering each other role.  Dota is a pretty good example of this strategy, as it doesn't take forever to master a single character (although I would still argue it takes too long for comfort), but it will take hundreds of hours to master them all.  Different characters, classes, or roles all allow this design strategy to be employed.

The key here is that these specializations are not distinct strategies.  In a game with multiple strategies you cannot only master one: you have to be able to play them all so you can switch between them as necessary.  These specializations are roles through which the entire game can be played, every game.  Players could choose a role, master it, and never have to play another.  This is an elegant solution because it's elastic; it allows players to invest either tens or hundreds of hours in the game and still be able to compete.

1. High Difference, Low Time (Most hobbyist games)
I mentioned at the start of the post that skill ceilings are composed of two parts that people typically treat as linked but, in reality, need not be.  Those parts are: win rate difference between a max-skill player and a newbie, and time to achieve max skill.  The problem in Chess and Go lies in that they both have very large differences between new players and veterans, and very long time to achieve max skill.  Combined together, players feel as if they are treading water, and only getting better slightly after each game.  If we can achieve a high difference between new players and veterans, but have that gap close relatively quickly (say 10-20 hours of play to achieve very close to max skill), players will feel less need to resort to mimicry, as they will feel themselves improving with each game.

In truth, nobody's learning process is linear.  In no game does the 10th game improve a player's ability as much as the second game - it's all diminishing returns.  This is, obviously, because a player's first and second games reveals to her the biggest and most obvious strategic plays.  Each subsequent game allows her to suss out more and more subtle strategems, until her gains become minute or nonexistent.  As with most things in our world, this leads to an asymptotic curve.

Here, players make big skill-level bounds over the first few plays, and gain less and less in each of the further plays of the game.  I think that the vast majority of hobbyist board games have a skill curve that looks like this one.

The problem with games with extremely high skill ceilings, although the player skill curve is the same shape as the one above, to most players (who don't have the time or interest to play enough to master the game), the skill curve looks linear with a very gentle (almost flat) slope.  This shape helps players disengage, as I mentioned above, because they don't see themselves improving and they see no end in sight.

And honestly?  Most hobbyist games have skill ceilings that can be approached in 10-20 hours of play.  As much as players like to kvetch about low skill ceilings, I don't think that hobbyist gamers really want extremely high ceilings; if we did, we would all be playing Chess and Go instead, and there would be no market for new games every year.  At least within the hobbyist board game space, players want to try new things, and they're not looking for a single game that they can play forever.  Given that a game that can be almost mastered in 10-20 hours of play and one that is infinitely replayable are to an extant mutually exclusive, in my opinion extremely high skill ceilings aren't usually desirable in strategy board games for this market.