Game designers often talk about two types of randomness, input randomness and output randomness. In this post, I’d like to critique this polar model of randomness and introduce some new thinking about it that designers should consider. First, a quick summary for those new to the concepts:
The distinction between input and output randomness in a game is whether the random event happens before or after a player’s decision. When the random event happens first, and a player decides on a response to it, we call that input randomness – the randomness is an input to the the player’s decision-making. Drawing a random tile in Carcassonne and then deciding where to play it is input randomness. Output randomness is the reverse: first the player makes a decision, and then a random event happens that impacts the decision. In many American-style troops-on-a-map game, the die roll for combat resolution is an example of output randomness. The randomness is an output, a consequence of, the player’s decision.
Geoff Englestein discussed these concepts at length with co-host Ryan Sturm as early as June 2012, in Ludology Episode 34 and again in a more concise form in this audio GameTek segment, both of which have done much to popularize these ideas within the tabletop game design community.
Video game designer Keith Burgun, also talked about these ideas in a video at the start of 2015 that offered a subtly different take.
A key difference between Engelstein and Burgun is that Burgun places randomness on a spectrum between these poles. At the far end of input randomness, says Burgun, is map generation, like in Catan. The map is generated at random, before any player decisions are made, it is known to all players at all times, and it doesn’t change.
Burgun’s next example is very helpful in illuminating how he distinguishes input from output randomness. He references face-up cards that draw out every few turns, which is perhaps like the market row in Suburbia. Burgun says that since players have a few turns before they have to respond to this random event, this is closer to input randomness than output randomness. Similarly, drawing cards into your own hand from a deck, places the consequence of the random event even closer to your point of decision, because the card is now not just potentially available, but actually available to play, is seen by Burgun as closer, on the spectrum, to output randomness. Eventually, we get to rolling a die to determine the outcome of combat, which is clearly output randomness.
Burgun emphasizes this notion of distance in, game-time terms, between the event and needing to grapple with it. The less time a player has to respond to the random event, the more Burgun would describe the random event as output random. Burgun believes that for designers who want to create strategy games, it’s important to provide players with enough time to plan for and respond to random events, so in the context of strategy games, Burgun believes that output randomness is bad design.
Englestein, on the other hand, mentions almost off-hand that input and output randomness are actually a loop, in which the random outcome of some event becomes the input to another decision. Your loss of two troops in your first attack on Greenland in Risk is now a random outcome that informs your decision to attack again, or to try something else. Your decision will then expose you to another encounter with output randomness, and so on and so on.
This works the other way as well. When you place a tile in Carcassonne, you’re creating a different set of payouts for tiles you might draw from the bag next, because of what you just played. If you extend a city, for example you’ve raised the value of a matching city-closing tile by one point for each tile already in the city. When you then draw a new tile from the bag, you’re experiencing output randomness – will you draw the tile that helps you? Will the wagers you’re making pay off? The random tile draw will determine that!
This uncertainty loop reminds me a bit of how physicists grapple with time. In their equations, time can go backwards or forwards, but in the real world, for whatever reason, time stubbornly goes only forward. Similarly, though input and output randomness are looped together, there’s a huge experiential gap between games that Burgun would characterize as having a lot of input randomness, versus those that have a lot of output randomness.
I’d like to suggest, first, that we refer to these games as being clockwise (input) or counterclockwise (output) in terms of randomness. For one, it will be easier to say “clockwise games” than “games which predominantly feature input randomness.” For two, it emphasizes the cyclical nature of randomness that Engelstein pointed out. For three, I’ll get to sneak in one of my favorite words, ‘widdershins’ – which means counterclockwise – into more of these posts.
I think the experiential difference between clockwise and counterclockwise games is in how players find their way through the choices the game presents and how they climb the heuristic tree towards improved play. Sticking with our examples, in Risk, a counterclockwise game, players initially learn to look at troop ratios in provinces and decide on whether an invasion is likely to succeed. Beginners will often commit to repeated attacks when they start out with twice as many attackers and defenders, irrespective of the outcomes of individual rounds of combat. They make an initial probabilistic evaluation, and then engage the randomness machine as often as is necessary to yield a decisive outcome. This gives Risk that strongly counterclockwise feel.
As players gain skill, however, and as they become more sophisticated in their understanding of the independent probabilities of successive rounds of combat, they change behavior. The repeated refrain in a game of risk between skilled players is “again?” As in, after you’ve evaluated the results of the last round of combat on your troops, the enemy troops, your objectives for the turn, and your overall strategy, would you like to roll the dice again, and wager two of your troops against two of mine? This decision is what I’d call high noon – the balancing point between clockwise and counterclockwise, where both past random events and future random events crystallize into a rich and interesting decision point.
In Carcassonne, a clockwise game, we see the same evolution in reverse. Initially, players look at a tile they’ve drawn and try to find the best use for the tile relative to the current board state. Does the tile have a city on it? Is it well-suited for extending a city you’ve already started? Should you start a second city? Over time players also learn to use tiles aggressively, to constrain their opponents’ chances of completing features they’re working on. Finally, players come to realize that their tile placements are more like wagers on a few simultaneous pots, and they seek to maximize their odds of getting those pots to pay off.
Experienced Carcassonne players react most strongly to the moment they draw and reveal their tile, because they already know all the possible draws, and which pots might pay off. What an interesting moment of drama in a clockwise game! The random event which decides the payout! These players have reached that tipping point I called high noon before. They know which tiles they’re hoping to draw, and among the best players, they know the odds of drawing them. The moment then, is the resolution of that tension, for good or for bad.
We’ll continue to discuss random loops in our next post, where we’ll examine what happens when there isn’t enough looping to the experience of randomness in a game, and how that feels both for clockwise and counterclockwise games.
