Skill Compensation

Something I’ve been meaning to blog about for a long time is a concept I personally think of as Skill Compensation (I briefly touched on it here). Some alternative names for this game property might be Feedback Accuracy, Skill Measurement or Performance Compensation. Let’s take a shot at a definition:

Skill Compensation is the degree to which a player’s performance is reflected in the final outcome of a match. Phrased another way, it is the extent to which “the better player wins”*. 

The skills measured in a game like basketball might include running, passing, shooting accuracy, passing accuracy and decision making ability. In a game like “War” there are no skills measured whatsoever. Therefore War has no Skill Compensation.

In a match of a game that is primarily a strategy or decision making game, a player’s skill is primarily a function of how valuable their decisions are in terms of maximizing their win probability/expected value (“EV”).

Interestingly, in a game like Yomi, for any given game state/turn of the game, there exists a single “best” play in terms of raw EV. However, because the system has both you and your opponent act simultaneously, it is possible to “win” a hand by doing the “wrong” play. Frank Lantz talks about this concept in his article on Donkeyspace. This means that a given turn of Yomi has less than perfect Skill Compensation, if we set aside any value from the skill “reading” or “predicting” the opponent.

Simultaneous action in Yomi. Notably, Yomi still features a lot of depth/strategic consideration possible beyond “reading” or “predicting”.

However, crucially, despite the fact that simultaneous actions tend to lower Skill Compensation, the depth of the game has not necessarily decreased; there may still be the same about of strategic consideration, calculation and skill possible. Even though you might be rewarded for making the wrong move and punished for making the right one, the “right” or “more correct” move or set of moves still exist for players to find. This is also the case for games that feature a significant degree of output randomness, like Hearthstone or Risk. While a given game may feature more or less of either, Depth and Skill Compensation are not the same thing.

Pictured: lowered Skill Compensation.
Cards like this lower Skill Compensation but may actually increase depth, because players may have to contend with more possible future game states.

There are also advantages to adding a system like simultaneous actions to a system. It allows for weaker players to have intermediate successes even when playing the “wrong” move. It tends to allow for the hope of comebacks. And it allows us to tell stories about how we “just knew” the opponent would do that, which can be fun and exciting. Similarly output randomness, or any number of Skill Compensation reducing mechanisms can fuzz up the feedback and also allow for weaker players or players who have fallen behind to have hope. It can allow for a wider range of “valid” or “justifiable” moves, which contributes to playfulness. And variance in a system tends to contribute to a variety, novelty and excitement.

That is not to say that Skill Compensation is irrelevant. Some players may play games “to prove something”, in which case this might be a highly important property. Otherwise, such players may find themselves feeling cheated or having “wasted time” in terms of evaluating their skill/progress. Also, having lower Skill Compensation tends to lower the rate at which players will learn/gain understanding about your system. Whether this is important to you as a designer is a topic for another day. (Keith Burgun has claimed that a fundamental value in strategy games is gaining understanding, but it seems to me to be unclear whether the rate of this learning is important)

One thing I’d like to mention before closing is that games can also give feedback within a match (like taking a tower in DOTA), reducing your opponent’s life points in Magic or fulfilling an objective card in Wingspan. The extent to which this feedback is valid in terms of contributing to your win-rate is also interesting, but beyond the scope of this article.

To summarize, it is absolutely possible to have a game with high depth (interesting decision space to explore) and low Skill Compensation. Varying Skill Compensation can have both positive and negative effects on your game design, depending on what other properties you’re optimizing for (evaluation, learning rate, playfulness, etc). In addition, some players may find one range or another of Skill Compensation to be incompatible with their tastes. 

Let me know what you think, and whether this brings to mind games as examples of extreme ends of this spectrum, but with variable other properties. If you’re interested in games made with these sorts of considerations, especially with an eye towards playfulness, please sign up for the BrainGoodGames mailing list here.

 

* Notably, human performance can vary quite a bit from match to match even in a game that is very high in Skill Compensation, like Chess or StarCraft (there are other human factors like framing, specific preparation, nutrition, intimidation, etc etc). Therefore it might be more useful in some contexts to talk about performance within the frame of a given match rather than talk about player skill as an absolute fixed value, but we can use skill as a rough shorthand for average performance.

6 thoughts on “Skill Compensation”

  1. Hearthstone moved away from cards like Deadly shot because increasing the variance wasn’t acutally a good idea.

    A good example of high depth but low skill compensation is Texas Holdem

    The best players in heads up sit and go’s (before bots took over) had a 53% win rate. But poker is a very deep game https://www.youtube.com/watch?v=vq5zmZx97Jk

    This seems to be the same thing discussed by richard garfield in his book https://www.goodreads.com/book/show/13183103-characteristics-of-games characteristics of games.

    1. Thanks for your comment. One neat thing about deadly shot is how you can theoretically mitigate the randomness by reducing other options (maybe by hitting smaller things with your minions first).

      Texas Hold-em is a great example of what I’m talking about!

      Still gotta read his book, it’s at the top of my amazon “to-buy” list. I’ve had a hard time finding it elsewhere.

  2. The outcome of the game can be broken down into:
    Luck+Yomi+System Understanding+Heuristic Strategies+Calculation+Memorization+Execution/Reflex.
    Luck always minimally exists since no two situations are always equal, although it might be irrelevant.
    Skill Compensation just means how much Luck and Yomi is a factor.
    The thing is Luck and Yomi also modify Heuristics, Calculation and Memorization as they are part of the Solvability of the game.

    1. If Calculation and Memorization completely overtake the Luck component that means it’s a Solved game.

      From what I see what Keith wants to create is as Strategy games based on System Understanding + Heuristics and minimal Luck and no/less other factors.
      That could be seen as an Ideal Single Player Game in one way.

      But definitely far from the concept of a Multiplayer Strategy Game.

Leave a Reply

Your email address will not be published. Required fields are marked *