Last week I was thinking about how different game systems use different dice resolution mechanics. I spend a lot of time thinking about game mechanics. As in, maybe, more than an hour a day, usually walking to and fro the office. It’s one of those topics my mind wanders to when left idle.
The point is that this time I had a little insight that might prove useful to the aspiring game designer/kitbasher/modder. It’s something I’ve never read in any of the books about game deisign I read. Here it is:
Pick the combination of random distribution and modifiers appropriate to the expectations of success for the different levels of expertise and difficulty
That’s a bit opaque. But reread it. It’s composed of many parts:
- random distribution
- expertise and difficulty
- expectations, which is what’s important here.
The first question is: why do we want rules?
Without them, someone has to adjudicate everything on the spot. It means rulings all the time, and a gigantic lot of trust, and a lot of effort from the referee. It’s not bad! It can be fun.
But sometimes you want rules to help players, and as a crutch for the referee. If you use some resolution mechanic the players have an easier job of understanding the risks and outcomes of their actions, and the referee something to lean on when they do not want to think too much about something. This is important because evaluation and risk assessment is something we do in a form or another all the time.
And everybody at the tables has expectations about how things should work out: being better at something means succeeding at it more often, harder tasks should be successful less often and, in case of a challenge, the best should win more often.
The trick is picking the mix of dice rolls and modifiers to make sure that the “often” is the right amount… and this varies from game to game. Actually, even more than that. Unless you’re trying to simulate a real world system, there’s no “right”. It either pleases someone, or it doesn’t, and there are many people at the table with different opinions.
Example, in D&D 3E a strength challenge between two people with strength 18 and 3 has the latter win 8% of the times, and a draw about 2%. This happens because, in layman’s terms, very high and low dice results happen often, and the two modifiers are small compared to the big spread of dice results. D&D does not use stats as straight modifiers, but halves them, because the designer decided that they did not want stats to be too relevant.
GURPS enters from the left. Ignore character building and look only at the resolution mechanics: 3d6 under the appropriate stat. There is less than one chance in ten thousand that the puny weakling will prevail. This is because stats in GURPS are taken at face value (there’s no halving the stat like in 3E) and the dice spread is way narrower, so extremes happen way, way less often.
So we have two factors impacting the result here: a less spread distribution making odd results less probable, and a greater impact of numeric stats on the roll. What’s best for your game?
Rulesets leave, expectations enter the scene and start ignoring each other. What players want differs. Some don’t want the strong to lose, because, hey, they’re stronger. Some want some kind of spread. Some want the stronger to lose only, say, in a narrow case that amounts to a critical (“I sat down to strongarm that puny weakling and pulled a muscle!”), other are ok with a game where there’s less control over the outcome because it fosters shenanigans. Expectations are not simply the “I don’t want to lose”.
(By the way, such high stats in GURPS are uncommon, but do happen, even without rolling stats at random (yes, in third edition it’s a valid way to generate a character). And, yes, there are many ways to handle checks in GURPS and 3E, please don’t remind me, thanks.)
While the above might have seen some discussion, here’s something more interesting and original: how should task difficulty be handled? For the sake of the discourse, move from strength checks to a skill. Say, climbing. And I’ll structure them out as questions because I guess at this point you can understand what I’m saying and I’m afraid I’m going to waffle:
- should difficulty affect the novice more or less than the expert?
- is the expert ever going to fail the simple stuff?
- is the novice ever going to manage the easy stuff?
- is there a ladder of proficiency so that there are many levels of expertise? so that everybody always succeeds with lower levels and fails higher?
- what are the chance of success of a normal task? Should they always be the same for every task? and why 50%?
- should a harder task impact worse an expert or a novice?
- what actually happens on a failure?
- the expert fails surely less often, but do they fail less badly?
- and, most importantly, how does all the above impact on your game?
Let me rewrite it:
how do game mechanics impact on your game?
There’s something we need to address too: the random aspect. Dice. We use dice when we are not sure about the result. The better should succeed, but sometimes they do not, and we need that tension. The random element here is used to model factors not otherwise accounted for. It’s a shortcut that allow us to keep the game moving. Diceless games exists, and they use their own ways to handle this issue. Some other systems, including for example Ars Magica and d20 (both by Tweet), interpret extreme dice results with different outcomes depending whether the situation is controlled (almost no other factors), semi-controlled (few other factors) or completely chaotic (you are on fire on the back of a flying whale fighting a giant robot). Both AM and d20 are 1 die + modifier > difficulty:
- fire-whale-robot scenario: in AM any 1 can be a disaster, while in d20 you have autofailures on 1, and autosuccess on 20, and critical hits.
- some factors: both AM and d20 have the provision for simple roll + modifier, without automatic failure or disaster on extreme results. So, if you’re good enough (the modifier is big enough), there’s no need to roll.
- no other factor: in d20 you can take 10 and take 20. no randomosity. I don’t think there’s anything similar in AM.
Ok, enough waffling. That’s it.