"Three bad kinds of randomness, and one good one"

#21
So I was hoping that my post would garner a few more precise definitions of input/output randomness to consider and I'm pleased to see these responses. The public definition that Geoff Engelstein and @keithburgun still use is "input randomness is before the decision" and "output randomness is after the decision." This colloquial definition, as I argued above, doesn't actually mean very much if taken literally. You have to apply some very creative interpretations of the words "before" and "after" in order for it to make any systemic sense, which is why I guessed that perhaps it was more of a psychological distinction than a mathematical one. But I'm now willing to accept that it is actually systemic and that it is just poorly defined.

To this end, I'm happy to see some alternate definitions being offered. Here's what I pulled from the above posts:

(1) @keithburgun wrote:
"... with input randomness, in all cases,the random information has had time to be processed by the player before it changes the game state in a permanent sort of way (this is probably something I should talk about more)."

I actually quite like this definition and it's pretty close to the definition I came up with myself (see below). I look forward to you talking more about this aspect, Keith.

(2) @evizaer wrote:
"You can measure the (usually metaphorical but sometimes literal) distance between what state player actions can effect and what state the randomness effects. Output randomness is very close."

The literal distance version is easy to grasp. In a game where the player has an avatar that exists in space, if my avatar is standing 4 spaces away from an appearing monster that's more input-y than if I'm only standing 3 spaces away. I get this.

The metaphorical version is still a little hard for me to parse however.

I think all random information necessarily impacts the player immediately, as any serious player must immediately base their decisions on all available information. So I think what this definition is still lacking is precise language to talk about different kinds of impacts randomness might have on the player and how those impacts might shift over time.

(3) @BiggJobag wrote:
"I believe input randomness, as Keith describes it, introduces concrete information (that is, factual, and not just odds) to the player before the player uses that information in making a decision that affects the game-state meaningfully in relation to win-chance. Output randomness, however, introduces concrete information to the player after the player has made the decision that will use that information to impact win-chance."

I like that this introduces "win-chance" as a metric, but it still reads to me like just a wordier version of the old before/after the decision paradigm which has the problems already described that all randomness is technically both before and after a decision. Furthermore, this definition seems to imply that specific types of information within the game are in some way married to specific decisions, as in "the decision that will use that information." But don't all good player decisions use all available information?

-------------------------------

After reading these responses, here's my own attempt at a definition:

Input randomness is when random information remains fluid in utility for a longer period of time after it is generated (where 'utility' is defined as the impact of that randomness on the player's win % and 'period of time' is measured in terms of the number of decisions the player has the opportunity to make).

Some examples:

ROLL-TO-HIT
The random information is generated and immediately achieves a fixed utility of either good (hit) or bad (miss).

DICE PLACEMENT
The dice are rolled and then stick around. They get placed on the board over several turns and are used to gather various resources. They block other players from going to certain spots. Perhaps a special power allows the altering of certain die values. Thus, after this random information is generated it has a fluid utility (changing value to the various players) for a while. The final utility of those dice to the various players at the table does not become fixed until the very end of the round when the players pull back their dice.

DUELING CARD GAMES
"Instants" that are played and then go straight to the discard pile are more output-y than "permanents" which stay in play for a while until they are destroyed, which in turn are more "output-y" than longer-lived permanents that never get destroyed but only get altered in various ways.

-----------------

Anyway, the above way of describing things is way more intuitive to me and implies that this has little to do with the word pairs input/output, before/after, or near/far, but rather it has to do with the ideas of "longevity" and "alterability."

Phrased as casual advice, I might say something like "strategy games should introduce random objects that are both longer-lived and more alterable during their lifespan."
 
Last edited:
#22
Input randomness is when random information remains fluid in utility for a longer period of time after it is generated (where 'utility' is defined as the impact of that randomness on the player's win % and 'period of time' is measured in terms of the number of decisions the player has the opportunity to make).
So how do you do this? I think you will reach my definition if you flesh out the implications of this definition.

"Remains fluid in utility" can only be achieved through far/further randomness. Fluidity derives from (at least) metaphorical distance. If something is your #1 concern because it's in your face and going to kill you next turn, utility is not fluid at all. If the monster is a couple turns of movement away from doing anything, or has a long wind-up on its attack/cooldown to wait out, then you have time to interdict the attack--the utility of when you do this is fluid. You have choices. You aren't forced to interdict the attack now. If you were forced to do something now, utility judgment collapses into a stark "do you get hit and die or not?" binary.

I think an interesting example is a design that varies damage numbers on attacks by small proportions (say, < 20%), and damage doesn't do anything until an actor dies from it. Then what appears to be output randomness actually has some distance between the die roll and the terminal outcome. The player has time to react before facing the stark consequences, even though the dice are rolled in her face. This randomness seems to be deceptively far, and only becomes near when the randomness leads to possible damage numbers this turn straddling life and death.
 
Last edited:
#23
I don't believe the concept of being "fluid in utility" really applies to random information. To be "fluid in utility" is for a property of the game-state to be changeable and thereby have a changeable impact upon the win-chance. But if this change emanates from player decisions and not randomness - well, the question has just presupposed the answer: randomness is not being fluid. Randomness happens. Randomness does not linger. If the properties randomness affected are later changed by player agency, then player agency is the source of the change, not the randomness.

In the case of placed dice which can later be changed, the initial values of the dice are generated randomly (more or less). But if you then switch the value on one of the dice, this does not mean the randomness has changed its utility: the win-chance may have changed, but this is purely because of actions you made after the initial values had been introduced, not because somehow the randomness itself has changed.
 
#24
"I believe input randomness, as Keith describes it, introduces concrete information (that is, factual, and not just odds) to the player before the player uses that information in making a decision that affects the game-state meaningfully in relation to win-chance. Output randomness, however, introduces concrete information to the player after the player has made the decision that will use that information to impact win-chance."
...
this definition seems to imply that specific types of information within the game are in some way married to specific decisions, as in "the decision that will use that information." But don't all good player decisions use all available information?
This is a useful addition to my definition, but not a refutation of it: we now need to accept that randomness can be at multiple points on the input/output spectrum simultaneously, when considering that the random information introduced could be relevant to multiple different decisions at multiple points in time. Perhaps an aggregation could be designed, accounting for the position on the spectrum and the importance to the decision?