<< Just What Is He Up To? | Main | It Could Have Been Worse... >>

November 05, 2003

Rolling Dice

I'm moving my reply over here from Bryant's thread, as I've wandered a bit from his original points.

The essential topic here started as how many times you need to roll dice to resolve an action (usually a combat action) in an RPG, but my focus is more on the complexity of the resolution process as a whole.

On the UI front, there are a few simple rules beyond "number of rolls" to keep in mind (though a low number of rolls obeys most of the rules)...

  • Limit it to simple math, and addition is favored over subtraction or, indeed, anything else. Division in particular should be used with extreme caution, even relatively easy stuff like division by 2 or 10.
  • Keep the numbers small. Single-digit is superior to double-digit and numbers in three or more figures should be avoided if possible. For example, the Synnibarr idea of rating damage for the smallest weapon in the hundreds or thousands then having each level of armor reduce damage by a factor of 10 is mighty stupid (as is most everything in Synnibarr).
  • Do as much of the math on the character sheet as possible, leaving the arithmetic with the dice as simple as you can make it. That way the speed of calculation is less of an issue. Of course, this means you can't go changing thngs on the fly... For example, changing the attribute a skill applies to by use is reasonable, but it slows down play! Kudos to Feng Shui for rejecting this idea!
  • No tables. Just don't do it. There is always a better way.
  • Okay, if you must have tables, keep them simple enough they can all be squeezed onto the character sheet.
  • Apply modifiers to only one side of the contest. How obvious is this? If I have to go through complex calculations (and even a handful of additions/subtractions qualifies), I only want to do it once. All too often, games have a situation where equivalent results are achieved by modifying either side of a contest, but they arbitrarily divide modifies over both rather than reducing problems by putting them all on one or the other side. Yes, d20, I'm looking at you, where some effects actually create modifiers for both sides!
  • Counting is easy, but only if the count is small. As mentioned in my comment earlier, the Whispering Vault idea of grouping dice by like value and counting in groups is a really slow and clumsy process. And the less said about Tunnels and Trolls' "group like dice and reroll" berserk mechanic (as fun as it is to watch the damage pile up), the better.
  • Mixing together items with different rules for each (say, certain dice that have different meanings or that have to be read different ways) should be avoided or made as simple as possible. Ghostbusters' logo d6 is probably OK, while Godlike's multiple die types is pushing the limit (even though one type is "don't bother to roll, just set down a "10").
  • Keep special cases rare or obvious. Feng Shui kinda loses out on the double-ones for jams... double-sixes for potential crits works OK as 6s are already a special case, but except for double-ones, ones mean nothing.

I'd also say that I don't completely discount the "involve him in the game" idea for having players roll a couple dice... Keeping all players attentive during the heavily phased nature of RPG combat is an admirable goal, and while one approach is to make each player's phase go very quickly so you move through the players rapidly, an alternative is to make the process more all-involving. This tends to lead to more broad, abstract resolution schemes like Maelstrom or HeroQuest simple contests. Here, a long time is spent by everyone describing their goals, each of which is turned into a modifier to a single resolution roll, from which the overall result is determined. d20 can be shifted this way by use of lots of "aid another" actions (particularly once the aiding PCs have +9 or better total skill mods), though some parts of the game get a bit weird if you do so.

Another alternative, even more interesting, is the Dust Devils idea. Here, there are two sorts of "winner" in any one round... One (the player who generates the best poker hand) is the one whose action succeeds. Another (the player who played the highest card) becomes narrator of the result, taking on a sub-GM roll to determine the exact details of the successes and failures. As with the abstracted all-in-one resolution ideas, this creates additional involvement in every round.

Okay... that's enough rambling for one morning, I think.

Posted by ghoul at November 5, 2003 08:14 AM

Trackback Pings

TrackBack URL for this entry:
http://noneuclidianstaircase.com/cgi-bin/mt/mt-tb.cgi/92

Listed below are links to weblogs that reference Rolling Dice:

>> Random comments from Population: One
Jack Gulick is the man when it comes to stats and gaming. He also plays a lot of Amber; go... [Read More]

Tracked on November 5, 2003 09:57 AM

>> RPG Die Mechanics from Where Worlds Collide
The Non-Euclidian Staircase discusses Rolling Dice, with a growing comment thread, on what makes a good RPG dice mechanic, and [Read More]

Tracked on November 6, 2003 02:28 PM

Comments

Aid another in d20 allows for some serious shifting, as you noted, and it does require some interesting descriptions from the PCs (something I force on 'em).

T&T was over the top; so was the original Stormbringer game. If you had a demon weapon, it had a good chance of growing in strength, resulting in its damage going up a d6 every time its hunger was sated.

One guy was rolling 30d6 by the time we stopped playing. He was one-shotting dragons, for crying out loud.

Posted by: Scott at November 5, 2003 10:07 AM

I'd add to that "Make the probabilities transparent". One of the reasons I strongly dislike almost all the cumbersome and arcane dice pool mechanics that were fashionable during the 1990s is that they obfuscate the probability curves.

I suspect that in some of the worst not even the designers understood the probabilities of their own systems, which can result in a badly broken games where the results of the dice mechanic don't match player expectations. The way that higher skill characters had a higher chance of critical failure in early version of Storyteller is an example of this.

I've heard it said that an obfuscated task resolution mechanic encourages roleplaying at the expense of minmaxing, a claim I find dubious at best. At least when I'm playing something like Call of Cthulhu and I roll a series of critical failures I know that I'm having bad luck with the dice. With a system like Deadlands (hack! spit!) I have no idea if that's also the case, or I'm attemting difficult tasks my character lacks the skill level for, or whether the GM and/or scenario writer doesn't understand the probabilities either and is setting the difficulty levels too high. And for me that ruins enjoyment of the game.

Posted by: Tim Hall at November 5, 2003 02:11 PM

I do like transparent distributions, though I also like "roll X and keep only the best" systems because of the distribution they create, even if it is tricky to calculate. I probably have another entry to do on the shape of distributions, which are good/bad, and why.

As for designers and odd statements... I've heard at least two well-known designers say they don't mind broken rules because it "encourages people to move past the rules". Hey, people, don't tell me you intentionally made it poorly to encourage me to creatively fix it for myself... If I wanted to design my own game, I wouldn't pay you!

Posted by: Ghoul at November 5, 2003 05:22 PM

The problem with a lot of dice pool systems is that they're not plain "roll X and keep the best" or "roll X and count successes" but they have additional complications like rerolling doubles or ones cancelling successes, which then play hell with the probabilities.

I don't understand the attitude of those designers (was one of the John Wick, by any chance?)

Posted by: Tim Hall at November 6, 2003 08:21 AM

I did a big table of multitudes of die rolling scheme probabilities several years back on CompuServe... I should dig that up (won't be too hard... I'm pretty sure I know where it is at home) and put it somewhere here. Making the tables for old-edition Vampire (et. al.) was, indeed, rather a pain. Doing Whispering Vault was, as you might guess, worse. And some of the more extreme open-ended rolls were as well.

And no, actually, the oddest Wick statement I can recall is the L5R RPG design note where he says he wanted to make it diceless, was told he couldn't, so made a complex die scheme that is so biased toward better skill winning that it might as well have been diceless from the result. Now, to my mind, making a complex, hard-to-learn, hard-to-do die scheme that, in fact, reduces to "higher skill wins" way more often than anything else is a perfect example of game design not to be proud of.

Posted by: Ghoul at November 6, 2003 08:51 AM

I think L5R uses the same system as 7th Sea, a game I wasted my money on and describe as the RPG equivalent of Yes' "Tales from Topographic Oceans" as it epitomised everything that was wrong with the gaming industry at the time.

Posted by: Tim Hall at November 6, 2003 02:45 PM

I remember once when a friend of mine was designing an RPG(*), they asked me about dice probabilities - the question was why giving someone a +n advantage against a fixed goal was more advantageous to the person being advantaged than giving them a +n advantage against someone else making a similar roll. (The context was one in which people were rolling at least 3 dice, so the distributions were already close to Gaussian)

It's a cute little problem - an advantage of +n against a fixed goal is worth about as much as +sqrt(2)*n against another roller - but stuff like that is just lying in wait for the unwary game designer. Dice probabilities are hard.

I'll note that my friend's game, Nobilis, ended up being diceless.

(*) Actually, I may have the timeline messed up. This may have been after the creation of Nobilis, and my friend was consulting on something with another rpg designer.

Posted by: Daniel Martin at November 7, 2003 04:52 PM

I'm with you... even easy looking dice stuff can end up being hard in subtle ways. And contests are really nice, but they frequently just magnify the strangeness, if just because they pretty much double the variance (and, as a rule, variance should never get too large... my feel is that the effect of dice should not exceed the effect of character ability).

Yeah... Though it's clear Daniel already knows about this sort of thing, I do have to find time to put together that entry on distributions, shapes, and why some are better/worse. Maybe after this weekend (which will be spent in lots o' gaming and, therefore, less time to think abstractly about gaming).

Posted by: Ghoul at November 7, 2003 05:59 PM

Matching the range of the die mechanic with the range of skills levels is another issue some games handle poorly. It's one of the problems I have with Castle Falkenstein, which in many ways is quite an elegant game. For those that don't know the system it uses playing cards as randomisers, with Card+Ability to beat a target number. I always find the cards you hold matter far more than the abilities on the character sheet; the 1 to 13 range of the cards is something like double the range of ability values.

Of course, if your randomiser range is smaller than the trait range, you get the opposite problem; a low skill character can never succeed at a task a high skill character cannot fail at.

Posted by: Tim Hall at November 8, 2003 08:33 AM

Lots of games have random range issues, but Falkenstein is one that, I think you're right, serves as a perfect example of the extremes. When I ran CF at GenCon (which I did for three years running because I really do like the game), I regularly changed the ratio of character skill to cards by doubling or tripling the base attribute before adding the card. But that's only a rough patch.

Posted by: Ghoul at November 8, 2003 11:23 AM

Post a comment

Thanks for signing in, . Now you can comment. (sign out)

(If you haven't left a comment here before, you may need to be approved by the site owner before your comment will appear. Until then, it won't appear on the entry. Thanks for waiting.)


Remember me?