In deciding on a dice mechanic for my current system, I'm toying with a Silhouette/Shadowrun variant. I'm discussing this under the assumption that I'm using d6s, merely for the ease of semantics. In fact, I've been toying with various die types.

The issue I am having with a "roll X dice and take highest" scenario is that the dice max out fairly quickly, while using multiple max rolls (as in multiple max 6s on a d6) for some kind of bonus diminishes rather quickly as well. Basically, the more dice one adds to the pool, the more likely the player is going to get a 6 (if using d6s), with a rather small possibility of doing better for multiple sixes. The reason why I'm having an issue with this is that it takes some of the tension out of the check. If a player is making a check with the decent sized dice pool (which would be the case for any "skilled" character), he can pretty much assume that he's getting a 6, and maybe a little better if there's a bonus for multiple 6s.

In the end this translates to a diminishing value of checks from 1-5, while checks totaling 7+ are really too rare to "depend" on.

Does anyone know of a way to "flatten" and "spread" the log-linear curve somewhat? I've examined probability models using a d4, which was improved, but only somewhat. I've also tried an "add the next highest die for multiple 6s (or max)" type of thing, but (unless I computed it wrong) the effects were a little hinky.

Hey Mike
  This is just me spitballing here, but if you have made a detailed analysis of the odds for d6's, then why not use the number of d6's where the returns really start to diminish as the max number of dice you get and parse them out into max skill, ability, bonus, whatever. If you can predict where the sweet spot is, where getting more dice is inconsequenctial, use that as your game's upper limit.
  Also, a system like this models a certain perspective on reality. That perspective is that highly competent people easily outclass normal people and the highly competant people are only training, studying and struggling to eelk out a minor advantage over their competition. If that is not your philosophy, maybe you should use another mechanic...

Michael,

What if even numbers exploded?  They keep exploding until an odd number is rolled.

If you want to keep the original numbers (2+4+2+6+5 = 17) this will require a bit more work on the players parts, especialy if they have a large dice pool.

The other way would be to ignore the original number in explosions and add a finite number each time the dice explodes, only keeping the final number with the finite number added for each time it exploded.  (So if you choose 3 and Bob has one dice and rolls 2, 4, 2, 6, 5 he would get 3 + 3 + 3 + 3 + 5 = 17).  You could even include a mechanic for choosing which number to choose for that finite number which is added for each explosion.

Best,
        Bill

Tunnels and Trolls used matching numbers to explode. That is, if you rolled a 1, 1, 5, 3, 2, 2, 4, you would re-roll the 1, 1 and the 2, 2.

As I recall, we played that multiple re-rolls weren't put back all together, but were rolled separately, but I can't see this in the rulebook right now.

That is, on the example above, I might re-roll the two 1s and get 3, 6; and then re-roll the two 2s and get 4, 6. But the two 6s aren't a double, because they happened on different rolls.

Now Tunnels & Trolls used additive dice rather than number of successes. (That is, the above set of rolls would result in a total of 34.)

But there might be some tricks there; for example, the two 1s and two 2s might become a single 1+d6, and a single 2+d6. Obviously, it means that a dice pool of 1 will never explode. I just tried to do the math for what happens with a dice pool of 3 or more, and it made something else explode.... my head. So that might certainly be a "less predictable" mechanic.

Part of the way that the original Shadowrun did it was that the success depended on two things: the target number for each die, and the number of dice that met that target number. Defenses (such as armor) raised the number of successes required; It was possible for some actions to require more than one success, so that knowing you were going to get one success didn't translate to knowing that you would succeed.

Jerry

This is just me spitballing here, but if you have made a detailed analysis of the odds for d6's, then why not use the number of d6's where the returns really start to diminish as the max number of dice you get and parse them out into max skill, ability, bonus, whatever.

Also, a system like this models a certain perspective on reality...

Part of the way that the original Shadowrun did it was that the success depended on two things: the target number for each die, and the number of dice that met that target number.

You could try something like this:

<Skill> (n,m) meaning your character rolls n dice, rolling an additional d6 for any that are below m, repeating the roll-additional process. The result of the roll would be the number of dice that are on the table when you're finished.

So, as an example, a weaponsmith with Blacksmithing (5,0) would always end up with a 5 as his result.

But a kid with a lot of potential as a swordsman, with Swordfighting (1,6) would have a good chance of ending up with a lot of dice on the table, but still a good chance of getting a result of 1 as well.

(Granted, I know nothing about your setting or game, so these examples are purely illustrative).

Seems like a good way to allow for relative simplicity of rules, but still allow a good tweak to the shape of the probability curves (note, I haven't actually done any of the math for this yet).

Still, that does give me an idea to play with that might work.

IagainstI wrote:
The issue I am having with a "roll X dice and take highest" scenario is that the dice max out fairly quickly, while using multiple max rolls (as in multiple max 6s on a d6) for some kind of bonus diminishes rather quickly as well.
<snip>
Does anyone know of a way to "flatten" and "spread" the log-linear curve somewhat?

Basically, what I would like to know, is what are the "problems" with Fudge? Am I off kilter to think I could use it for a Sim?

Have you considered a Sorcerer/Donjon-style dice pool.