RFC: Character creation method and probabilities of custom dice

[oudeis]

I'm thinking of finally putting an idea into form that has been kicking around in my head for... damn, has it really been that long?!?!

Sorry. Anyway, I've had this system involving custom dice in my head for quite some time, but I don't know how to evaluate its worth from a statistical perspective and I'm hoping someone here can help. Before I actually place the order through Gamestation.net I'd like to get your(pl.) feedback.

I've always preferred rolling dice to the point-buy system, though I understand PB's virtues (getting the character you want, keeping lucky players from dominating the early game, preventing envy/resentment, etc). I like the sense of anticipation, the mystery, the fear of random results . However, the character generation methods I've seen either have too much range and downside potential or not enough: 3d6 and its variations can result in a hamstrung and even worthless character and 2d+6 tends to produce overpowered specimens, at least in my experience, and it just feels like the risk and uncertainty have been removed. So I started to wonder how you could retain the excitement of the roll while diminishing the possibility of junk character, yet give the player some of the control of the point build. Here is where my idea comes in.

At character creation, each player would given a pool of dice to roll. In place of the standard d6, s/he rolls customized six-sided d3s with the following possible face values:

• Type A: 1, 1, 2, 2, 3, 3
• Type B: 1, 2, 2, 2, 2, 3
• Type C: 1, 2, 2, 3, 3, 3
• Type D: 1, 1, 1, 2, 2, 3

Instead of rolling 3d6, players will have 6d3 to allocate to each characteristic. This will lessen the chance of an unusably poor result. Of the pool of 36 total dice they can roll, some would be type A and have the standard probability. Most would be type B and skew the results slightly upward (?), because heroes should be exceptional. For the stats the player wants to have a better chance of buffing up (no more than two) s/he would allocate some of the small pool of C dice. I still haven't fully fleshed out the role of the Ds, but I like the idea of the terror they might create.

So, could this work? What is the difference in probability of the various types above, and how many of each would have to be included in the total pool to have the desired effect, assuming it's even possible? Thanks.



PS: It's after Midnight when I write this so it might not be completely coherent or clear. I will re-read this in the morning and make any clarifications necessary.

[Edge of Dreams]

http://anydice.com/ is the best site for calculating custom dice probabilities. Here's an initial starting point: http://anydice.com/program/24ae

You'll notice that Die A and Die B have the same average, but B is skewed heavily toward rolling exactly 2. C and D, on the other hand, skew exactly the same distance above and below 2, respectively.

Now, take a look at this: http://anydice.com/program/24af - 6dA and 6dB both produce average of 12, but B is more likely to produce boringly average results. 6dC produces an average of 14, with about a 1.5% chance of an 18. Compare that to 4d6-drop-lowest, which has a lower average at 12.24, but a slightly higher (1.6%) chance of rolling an 18!

Anyway, play around with anydice and see what you find out.

Also, instead of strange d3s, try allowing players to roll something like 3d6 or 4d6-drop-one, but with the option to pick some number of dice (around 1 to 3) and change it to a 6 after it's rolled, or maybe set it as a 6 before rolling and don't roll that die. Also also, allowing 1s on your d6s to be re-rolled exactly once each (so if it comes up 1 again, you're stuck) provides a nice skew in my experience.

[The Rose Dragon]

Hm. While I can't quite comment on how this would work out, I realized I don't actually play any games that have any kind of random element during character generation other than Maid, which is... different. As such, I'm not sure what benefits, if any, randomized character generation has except to someone who rolls spectacularly well, to others' chagrin.

[erikun]

6d3 has a range of 6-18, with an average roll of 12. This is a bit higher than the 3-18 range and 10.5 average with the 3d6 dice.

As mentioned by Edge, Type B dice will skew results much closer to average (far more rolls closer to 12) with Type C averaging higher (14, apparently) and Type D averaging lower (10).

I might point out that 2d6+6 has an average of 13. (if you like) I note that, while 2d6+6 has a greater chance than 6dC to produce an 18, 6dC has a greater chance of rolling the values 13-17.

[The Dark Fiddler]

Hm. While I can't quite comment on how this would work out, I realized I don't actually play any games that have any kind of random element during character generation other than Maid, which is... different. As such, I'm not sure what benefits, if any, randomized character generation has except to someone who rolls spectacularly well, to others' chagrin.

Dark Heresy does it, and given that your skills are based almost exclusively on your attributes, bad rolls can really screw you over. Even your weapon skills are randomly generated.

Given that Dark Heresy is very much a system about how life is unfair and you are the underdog and cannot make a difference in the grand scheme of things, though, I suppose it does help the theme. Still annoying, though, if you end up with a guardsman with crappy weapon skills.

[JusticeZero]

There are other rolling methods, too. Bingo board works good for a stable luck based system. That is 4d6 best 3, 36 times in order on a 6x6 grid. Pick any array of numbers in order, like a bingo or tic tac toe board. Players can trade sheets.

[Edge of Dreams]

There are other rolling methods, too. Bingo board works good for a stable luck based system. That is 4d6 best 3, 36 times in order on a 6x6 grid. Pick any array of numbers in order, like a bingo or tic tac toe board. Players can trade sheets.

That is hilariously clever. I want to try that now.