You’ve all played Rock, Scissors, Paper, right? It’s a two-player game in which both players simultaneously reveal a fist (rock), open hand (paper), or two fingers (scissors). Rock breaks scissors, paper wraps rock, scissors cut paper. If both players make the same choice, it’s a tie and they try again.
Did you know that there’s a professional Rock, Scissors, Paper tour? Actually, we’re not sure there really is one, but there are international tournaments and two different events each year that bill themselves as the World Championship of Rock, Scissors, Paper. “Professional” RSP players call it Roshambo. It sounds more mysterious, and it’s a little shorter.
Rockin’ the Night Away
On its surface, Roshambo seems like a totally random game. If you programmed two computers to play each other, and one always chose randomly, it wouldn’t matter what the other one chose. In the long run, they will tie. In real life, the game gets interesting because the players try to psych each other out. You either predict the move your opponent will make or use psychological tactics to get him to make the move you want. The game rules are evenly balanced, but players’ minds are not.
Perry Friedman, a Roshambo World Champion and Tiltboy, once gave me a few tips on the game. New players rarely pick Rock as their first move – It feels unnatural to start with a closed fist and keep it that way – so Scissors is a strong first move against an amateur. Of course, he might have told me that just so that I would pick Rock first next time we played so he could be ready to counter it with Paper. Sneakiness is an important attribute for a Roshambo champion.
One Rule to Rule Them All?
We keep reading books and articles on game design that basically say, “All games are variations on Rock, Scissors, Paper.” Everyone nods their heads sagely at such an obvious truth, then works on their latest RSP game mechanic. The only problem with this is that the “rule” isn’t true and lazy application of it leads to bad game design.
The idea is as simple as RSP. Create some variation in the game by creating several different types of units, then balance them so that every unit has its Nemesis. If artillery is strong against cavalry, then make it vulnerable to infantry. If a Wizard easily defeats a Warrior because armor doesn’t stop his spells, then let a Rogue sneak up behind the Wizard and kill him before he can react. This is much more interesting than giving everyone the same strengths and weaknesses.
The concept is simple and obvious… but it’s a little too simple. If all conflict was determined by numbers and formulae, then an RSP model might work fine… and make a really boring game.
Black-and-White Logic in a Multi-Colored World
The problem with “Roshambo rules all” is that the real world is fuzzy, not binary. Bart Kosko, in his book Fuzzy Thinking, talks about the philosophy of fuzzy logic. Kosko claims that fuzzy set theory does a better job of describing the real world than do traditional mathematical models. When we talk about “tall people,” we don’t mean that everyone over six feet tall is “tall” and everyone else is “not tall.” There are degrees of tallness. In traditional Aristotelian logic, everything is either part of a set or not part of it. Fuzzy logic is designed to solve problems that are best described by “partial membership” in sets.
Roshambo has no grayness. There are only three possible actions, and the results of each pair of actions are clearly defined. Great games have many more factors and no black-and-white answers. The excitement comes from uncertainty and the idea that “on any given day, anyone can win.” A combat (or competition) is decided not only by the game rules, but by the relative skill of the players and other factors such as equipment, environmental conditions, and the actions of other players and “monsters.”
World of Warcraft combat is fun because it is unpredictable. There is a Roshambo bias for who “should” win in an otherwise equal encounter, but it’s just an edge, not a guarantee. And that’s pretty much true in every great game. There has to be room for player skill to make a difference. If the game is totally symmetrical, then there is no reason for a player to choose a particular type of unit and the game might as well be simulated on a computer rather than played.
Of course, if the game is too asymmetrical, everybody will choose the strongest units and the game developers might as well have left out the rest. This is why game balance is one of the hardest aspects of game design to master. A seemingly minor tweak can result in some players becoming far too powerful or others becoming hopelessly weak. As characters or units become more powerful and complicated, achieving game balance becomes a tremendously difficult job.
A Hit-Driven Business
Coin flips, Roshambo, and dice rolling in Dungeons & Dragons all have something in common – They have black-and-white rules for who wins. The best game designs hit the sweet spot between such deterministic rules and total chaos. Players tend to see patterns where there are none, and often assume randomness where a game is actually deterministic. This leads to what I think of as the “probability fallacy.”
Is a coin flip actually a 50-50 proposition? Only before you flip the coin. Once the coin falls, the result is known and is either 100% heads or 100% tails. The same applies to most game puzzle situations. Something that is a 75% probability really means that 75% of the players succeed 100% and the other 25% fail… or resort to restoring saved games until they “win.” Role-playing games get around this by repetition – If there is a 25% chance the player will get a particular item on killing a monster, he just has to kill dozens or hundreds of them to find the 10 or 20 items required for a quest. This can get more than a little boring and repetitious, but at least it gives players an extra reason for killing those monsters.
How about the actual combat? Dungeons & Dragons and its spiritual descendants have you roll two dice. The first roll determines if you hit, the second how much damage you deal. There are some problems with this approach, especially in an online multiplayer game. When you shoot at a moving target in Quake, do you hit it? It might look like a very obvious hit on one player’s screen and an equally obvious miss on the other.
A Warm, Fuzzy Feeling
A better fuzzy logic approach would be to apply a series of fuzzy rules to the combat – What sort of attack are you making? What kinds of defenses does your opponent have in play? How far apart are you? How much movement is involved? How good are your stats and equipment, and how strong are the opponent’s? A fuzzy combat system would combine the results of all the rules and come out with an appropriately-averaged result. This might be that you graze your opponent for a minor wound, or that you get in a solid hit, or that you miss entirely. The actual amount of damage would take all factors into account instead of being a random dice roll.
Obviously this isn’t a good system for a paper role-playing game where the players have to make all those calculations. But computers happen to be good at doing a lot of things fast. Why should our video games be tied to restrictions inherited from paper gaming?
Quest for Glory IV: Shadows of Darkness featured a fuzzy logic combat system. We think it was one of the most successful approaches in the series. Players could choose to fight as in an arcade game, or they could set a “combat style” and have the computer take actions according to that style. All of the “monsters” had individual fighting styles of their own. We think this made combat pretty interesting and helped players with poor arcade skills be able to enjoy the game.
We’d like to see more games use a fuzzy logic approach to conflict resolution. It isn’t much harder to program than traditional logic and it solves a lot of problems (such as the effect of lag) that have proven very difficult to solve with binary logic. Most importantly, a fuzzy logic system is much easier to balance because the reasoning can be recorded and observed. In effect, a well-designed fuzzy system can tell the developer or player why it made each decision.
So what will it be? Thinly disguised games of Rock, Scissors, Paper in which the game systems are transparent, boring, and can be manipulated by the players? Or tear up that paper tiger and create rich, complex game systems that use the power of fuzzy sets to balance themselves? Truly great games favor the skilled player, but leave enough room for a lucky, less-skilled player to compete and have a fair chance of winning occasionally.
- Show Me the Gold – Game Economics 101
- “Modern Lessons from Classic Games”
Part One: Chess
- “Modern Lessons from Classic Games”
Part Three: Dungeons & Dragons