Dice Jamming

[tenuki]

Just like any other GM worth the name, I am continuously working on a homebrew RPG system that never gets finished. The background is a world with a roughly 15th/16th century European tech level and some magic revolving around the manipulation of information and bio-modding. So no fireballs, but you can get a reflex booster enchantment, communicate mind to mind, let others see through your eyes, remote-control an AI built into a souped-up grizzly bear, have tree roots grow extra deep to extract mercury from the earth and enrich it in their fruits, stuff like that. The overall feel is gritty/heroic; life is cheap, but you may get to reboot if you're lucky.

I recently came up with this here skill resolution system (which I dubbed Dice Jamming for the time being) and would like to hear people's opinion on it.

* All basic rolls are made on 2d10 against a max based on a character's skill/trait+mods. That's right, rolling low is good.
* Rolling a zero on any die means pick up that die again and roll two dice instead; repeat until you end up with only non-zeros. (Note that your odds of getting at least one zero on 2d10 are 19:81, or roughly 1:4, so it doesn't happen that often.)
* Add up all dice rolled.
* If the total is equal to or less than the max, success level is determined by the number of dice rolled. Succeeding with 3 dice is better than with 2, and so on.
* If the total exceeds the max, subtract the number of dice rolled from the total. If the second result is equal to or lower than your max, you get a partial success (success lvl of 1). If not, you just fail (lvl 0).
* On a contested roll (such as in a fight), the adversaries' success lvls are subtracted from each other to see who hits whom.
* Here comes the good stuff.
* Upping the ante: After each roll, you can decide to convert any one (edit) of the dice you just rolled into a zero and do the double re-roll procedure descibed above. If you do this after a successful roll, you get both a chance of a crit and a risk of partial success or failure. Another application is that you can try and squeeze at least a partial success from an unlucky roll on a strong skill (e.g. if you roll 7 and 9 on a skill of 14, drop the 9 and hope for the best).

Example:
In the streets of Iartiga, Khed the Able Swordsman (skill lvl 18) is facing off against a goon of the Church of the Red Dawn, when he notices two other minions coming around the corner. He decides to try and dispatch his first opponent quickly before his odds get any worse. Khed chops away with his double-handed sabre (a Langes Messer for the weapons geeks among you), rolling a 5 and 3, an easy success. However, our hero figures he doesn't have the time to wait for the goon (skill lvl 12) to make a mistake, so he presses the attack, does a fancy winding maneuver and goes for a crit. He drops the 5 and re-rolls two dice, getting a 6 and 8. Khed ends up with a total of 17 (3+6+8), a triple success that is probably more than his semi-skilled adversary can handle.

What do you think?

(edit: add "one")

[Cynic]

3 things.

1: Apparently if you don't continuously create your own homebrew rpg system, you are a total failure as a dm. I fucking lose at this game, man. It's game over for me. GAME OVER!

2: The system is kinda terribad. the whole low roll is just going to get a little confusing.

3: The unique factor in the whole system is having to roll two dice if you have a zero. So you complete negate the effect of that by having that become the standard after each roll. If I roll an 8, and a 9. I know that I'm probably going to fail. So it is much easier for me to convert the 9 or both the 8 & 9 into zeroes and roll four dice increasing my chance of winning.

[tenuki]

3 things.

1: Apparently if you don't continuously create your own homebrew rpg system, you are a total failure as a dm. I fucking lose at this game, man. It's game over for me. GAME OVER!

2: The system is kinda terribad. the whole low roll is just going to get a little confusing.

3: The unique factor in the whole system is having to roll two dice if you have a zero. So you complete negate the effect of that by having that become the standard after each roll. If I roll an 8, and a 9. I know that I'm probably going to fail. So it is much easier for me to convert the 9 or both the 8 & 9 into zeroes and roll four dice increasing my chance of winning.

1. Precisely. :p
2. Rolling low is what everybody tries to do in any Chaosium game without getting confused.
3. My bad. I should have said "any one of the dice you just rolled". So no re-rolling 2 dice from the same roll.

[Dean]

Alright here's what I can see. It's very multi staged and "complex" for a resolution mechanic. So in some ways thats very good and in others not so.

There are a lot of steps to this system relative to most die resolution systems (which are usually one roll-success/failure) your system has a minimum of two steps and one choice, and a maximum of 6 steps and one choice. What this obviously means is that it's going to take longer, and significantly longer over the course of an entire game night. However it does have something which not many resolution mechanics have which is an element of choice built into it. That's the part of it that's of any interest to me. It puts a little bit of decision making into every roll. Making, in my mind, every roll a bit more interesting.

However it does seem like it has lots and lots of steps and stages. And that may not be wholly necessary.

[Ice9]

I think the multiple steps could work, as long as one roll accomplishes a fair amount.
So in a fight, for instance, you wouldn't want Attack vs Dodge + Damage vs Soak, for each player. Or even Attack vs Dodge for each player. But you could have a single Skill vs Skill roll for both sides, with the winner dealing damage to the loser.

Another thing is that if people are playing cautiously, you will get a lot of ties. At 12+ skill levels, the majority of rolls are going to be 2-dice success. So there will only be a difference if someone screws up or feel like gambling.

[tenuki]

First of all, thank you for your constructive comments. This is the kind of exchange I had in mind when I joined this forum.

I think the multiple steps could work, as long as one roll accomplishes a fair amount.
So in a fight, for instance, you wouldn't want Attack vs Dodge + Damage vs Soak, for each player. Or even Attack vs Dodge for each player. But you could have a single Skill vs Skill roll for both sides, with the winner dealing damage to the loser.

Yeah, that's how I want to do it, skill vs. skill, higher success damages lower, no soak, just a straight armor penalty to damage, if applicable.

Basically, every character gets one action per round. If somebody with a higher initiative attacks someone with a lower ini, lower ini has to roll first, so the guy with higher ini gets extra knowledge to base his dice decisions on. (I simplified the example above because I just wanted it to illustrate the dice mechanism; plenty of information already. Maybe I should've described a climbing check instead.)

Problems arise when somebody gets attacked by several enemies. 'll probably have to add a dodge/defense skill that doesn't cost an action and that is based on a different trait than the melee skill. It would have to be not as good as fighting back with your weapon (maybe a penalty, and you can't do any damage with it, either). However, you can use it if you have already used up your regular combat action or want to use it later.

More simple, I could just say that if you get attacked and don't fight back with a contested roll, your opponent gets a penalty to his attack equal to some fraction (maybe around 1/3 or 2/5) of your defense skill. I haven't done the math on that one, though, so I don't know how to calibrate it yet.

Suggestions are welcome, of course.

Another thing is that if people are playing cautiously, you will get a lot of ties. At 12+ skill levels, the majority of rolls are going to be 2-dice success. So there will only be a difference if someone screws up or feel like gambling.

Yeah, that's definitely a consideration.

One solution would be that a tie (or some ties, e.g. on odd success levels) results in both fighters taking half damage. This would mean that in longer combats, the stronger, tougher, better armed fighter has an advantage. I kinda like the idea because it ties in well with the gritty feel I want to get for the game world in general: If you get in a fight, count on feeling sore afterwards.

The stalling tactic also fails when a zero shows up on a die. In any given round, the odds of one opponent rolling at least one zero while the other rolls none are a little better than 3:7. I.e., about every 3 rounds or so, something will happen anyway. Don't know if that's enough, though.

Any ideas?

(edit: formatting)

[tenuki]

Alright here's what I can see. It's very multi staged and "complex" for a resolution mechanic. So in some ways thats very good and in others not so.

There are a lot of steps to this system relative to most die resolution systems (which are usually one roll-success/failure) your system has a minimum of two steps and one choice, and a maximum of 6 steps and one choice. What this obviously means is that it's going to take longer, and significantly longer over the course of an entire game night. However it does have something which not many resolution mechanics have which is an element of choice built into it. That's the part of it that's of any interest to me. It puts a little bit of decision making into every roll. Making, in my mind, every roll a bit more interesting.

However it does seem like it has lots and lots of steps and stages. And that may not be wholly necessary.

Good point, you're probably right. Maybe I should restrict the convert-to-zero option to successful rolls, or attach some kind of cost to discourage continuous use. E.g., when used to save against a failure, you take stress damage or lose mana/confidence/something along those lines.

I would like most rolls to be straight success/failure affairs, with partial successes or crits coming up only when a natural zero is rolled. (This happens about every one out of five 2d10 rolls.) The whole gambling thing should be reserved for important rolls and intense action scenes when everybody is staring at the dice anyway. Ideas?

As to action scenes, hit points are scaled so that the average character (including the PCs) is likely to be seriously handicapped after the second decent hit with a melee weapon wielded by somebody of similar strength, and down after the third. (Decent hit = 2 lvls difference; e.g. success vs. failure on a straight 2d10 roll.) So most combats won't have that many rolls, or so I hope. My goal is to make those rolls as interesting as I can. Again, suggestions?

Thank you very much for your comments!

[CCarter]

It seems a little over-complex - you might be able to refactor this into something less complex with at leastsame properties?

e.g.

Roll 2d10 (or d20) under [stat+skill]. If you succeed, you may choose to 'Push' your success up one level by taking a penalty of +d10 to the roll. You may Push up a roll multiple times, but you fail if the new total beats your skill. A character's level of success is 2 + the number of times they roll up.
Characters are forced to Push if they roll an 0.


Not sure how useful partial successes are, but these might may be less complex if they're a fixed margin (2 or perhaps 3 over the total) rather than being [# dice] over.

At least, this seems simpler to me. [shrug].

[tenuki]

Hmm, I see a number of problems with that.

If you don't pick up the die, you would have to assign it a value - if it just stays a zero, people are essentially getting a free success level. If you make it a 10, then you get all kinds of oddness in the probability distribution (e.g., the impossibility of rolling a 10 on a conventional exploding 10-sided die). I chose the double-reroll mechanic because it gets rid of the annoying (to me) mathematical issues of the existing systems that use exploding dice.

Another thing is, interpreting a 0 as a 10 instead of as a non-value messes up the scaling. The mean value of a single n-sided die in my mechanic is n/2 x (n-1)/(n-2), or 5.625 for a d10. In order to roughly keep this scaling, I would need to have a d10 explode on a 5 or 6, depending on whether I interpret the zero as 0 or 10. However, this still wouldn't solve the result distribution problem. No like.

As to partial successes, using a fixed margin might be workable. OTOH, is it really too much effort to count the dice before you on the table? It'll very rarely be more than four. And yes, I definitely need a way to have some rolls yield a success level of 1.

Revolutionary idea: Maybe I should just finish the damn rules draft and playtest this shit.

[CCarter]

Hmm...

'If you don't pick up the die you would have to assign it a value, since otherwise people are getting a free success level..'

Well, 2 dice drop 1, add 2 dice = 3 success levels.
2 dice +1 dice = 3 success levels

When I said '+d10', BTW, I didn't specify but was assuming you'd read this d10 basically the way you were currently reading your d10s, with 0 being a forced reroll. I'm not sure that was clear.

In terms of actual number rolled to pick up the 3 success levels, my method shifts the total slightly up (i.e. against the player) since in your original method, the player gets to pick the higher dice and convert it [there's no incentive to pick the lower value as far as I can see]. Hence the total will be [lowest of 2 dice] + 2 dice in your method, compared to 3 dice for mine.
There is admittedly a discontinuity in the potential number rollable, even though 0 counts as +0, at least as long as characters are forced to roll up every time they roll an 0. You could however drop forced rerolling for subsequent rolls, or make it 'reroll at -1' (the standard explodefix).

Personally the gaps don't bother me so much when you're rolling say 2d10, since explosion happens for each dice separately - I don't believe there are any numbers in the potential total that are just unrollable (like you get when rolling up on one die). So its not as bad as with say d20/20 rolls up, where rolling a total of 20 is flat out impossible.
For example with the 2d10 if you got an 18 and have to roll up by at least 1 point on d10, you can no longer roll a 19 now, but it was entirely possible to get a total of 19 when the whole set of rolls to define success of the action started.

As far as setting fixed threshold for partial successes, OK maybe that's not much saving. When I was running through the process in my head I was keeping track of dice added/subtracted mentally, but yeah it is slightly simpler in play since you can just see how many of the things are left on the table, at least if you have enough d10s.

[fectin]

So, why is this better than playing your setting using d20? (or storyteller, or d6, or whatever)
It's kind of a neat mechanic, but it's definitely not lightweight. What does it give you that's worth the extra effort of using a peculiar system?

[tenuki]

Hmm...

When I said '+d10', BTW, I didn't specify but was assuming you'd read this d10 basically the way you were currently reading your d10s, with 0 being a forced reroll. I'm not sure that was clear.

Well, now that you say it, it's quite clear...

I guess I didn't read your post carefully enough.

In terms of actual number rolled to pick up the 3 success levels, my method shifts the total slightly up (i.e. against the player) since in your original method, the player gets to pick the higher dice and convert it [there's no incentive to pick the lower value as far as I can see]. Hence the total will be [lowest of 2 dice] + 2 dice in your method, compared to 3 dice for mine.

It also completely kills the option of saving a failed roll by gambling, since you can't ever get a lower result than you had on your first roll. Hmm. There are all kinds of consequences to that, but I'm not sure they're bad.

For instance, what kind of bothered me about my way of handling the voluntary reroll is that you could turn a failed roll into a crit (e.g. roll 6 and 9 vs. a skill of 14, drop the 9, roll 8 or lower (14-6) on the two new dice, not that unlikely). While this might not be so bad mechanically, it doesn't exactly facilitate the description of the whole process in terms of storytelling - having these miraculous turnarounds all the time can get old fast. I'd even thought about ruling that the best result you can get from saving a failed roll is a partial success, but that would have been yet another rule... no good.

Another effect is that you will get more results of 2 or 0 successes - people will gamble less on a normal success because it's more risky, and you can't save a fail any more. This means that a difference of 2 success levels becomes more likely on a contested roll, so I may have to make the consequences of that a little less severe (e.g., adjust combat damage). However, considering that you can also get a partial success right away on your first roll... this brings back your idea of setting a fixed threshold for a partial, maybe 3, to make partials more likely in the first place.

I like your suggestions a lot better now that I understand them. ^^
Great job!

[tenuki]

So, why is this better than playing your setting using d20? (or storyteller, or d6, or whatever)
It's kind of a neat mechanic, but it's definitely not lightweight. What does it give you that's worth the extra effort of using a peculiar system?

Hahaha, very good question, and one that I maybe should have addressed before launching into the description of the mechanic itself.

Well, here it comes.

I want the following things from my system:
- enough complexity enough to allow for meaningful tactical decisions
- a probability distribution that has at least some resemblance with a Gauss curve
- plausible results that are easy to interpret and describe in terms of storytelling
- no charts with bonuses and penalties for various combat maneuvers, feats, spells, special abilities, critical hits and stuff; nobody should need to look up stuff during the game
- one mechanic for everything, no special rules for certain characters or character classes, just traits/skills

The approach I came up with is to put most of the complexity and options for decision-making into the interpretation of the dice. They're out there on the table where everybody can see them, which makes it easier for people to follow the events and get involved with what the other PCs are doing, both in terms of story and mechanically. Or so I hope.

Also, what lowers the threshold for me is that the people I'm going to be playing with don't know that many systems anyway (not even d20!). They have experience with Earthdawn, Talislanta, Call of Cthulhu, some Rolemaster (eurgh) and Decipher's Middle Earth. None of these mechanics support the overall mood of the setting I have in mind.

Finally, I'm a total sucker for game design and tinkering with probability.

[CCarter]

Well glad to be of service. Interesting and thought-provoking thread for me anyway, whether or not you use any suggestions.

Good luck with playtesting...or maybe I should wish you a complete lack of good luck so the probabilities can be evaluated fairly =P

[fectin]

Fair 'nuff. A couple specific points then:
- Exploding dice prevents you from having a gauss curve. It'll be skewed wierdly instead.
- Tactics also needs fairly deterministic outcomes, or at least predictable ones. This seems really swingy, though CCarter's suggestions help a bit.

[fectin]

After some effort, here are approximate probabilities of results off each d10 under this system. Its a stochastic solution, so the numbers aren't exact, but should be close enough. First colum is end value, second is probability of that outcome in %. Mean result is 5.64, distribution is definitely not Gaussian (so standard deviation isn't meaningful).
1 9.99926
2 10.0839
3 10.20815
4 10.31822
5 10.42476
6 10.50959
7 10.62967
8 10.76162
9 10.84883
10 0.97721
11 0.89936
12 0.80938
13 0.72672
14 0.63636
15 0.53755
16 0.43785
17 0.33495
18 0.22941
19 0.11709
20 0.09957
21 0.08339
22 0.07027
23 0.05562
24 0.04517
25 0.03458
26 0.0266
27 0.01967
28 0.0147
29 0.0124
30 0.00966
31 0.00816
32 0.00643
33 0.00507
34 0.00378
35 0.00329
36 0.00229
37 0.00209
38 0.00144
39 0.00121
40 0.00102

[tenuki]

After some effort, here are approximate probabilities of results off each d10 under this system. Its a stochastic solution, so the numbers aren't exact, but should be close enough. First colum is end value, second is probability of that outcome in %. Mean result is 5.64, distribution is definitely not Gaussian (so standard deviation isn't meaningful).

Thank you very much for providing the numbers. How many rolls did you do for this d10? As I wrote above, according to my calculation using the limit value of an infinite geometric series, the mean result for an n-sided die is n/2 * (n-1)/(n-2), or 5.625 for a d10. I hope I got it right.

When you overlay two d10s with double-reroll explosion on a 0, the curve is roughly pyramidal with an extended, sloping right side, looking a bit like a root sign rotated by 180 degrees. That's close enough to Gaussian for me. It becomes more bell-shaped with more dice, of course, always retaining its extended right side where the failures are located.

Btw, I slept over it and decided that I'll probably use the original convert-to-zero mechanic for pushing (good term, btw). This is because I want very skilled characters to be able to bring their advantages to bear more effectively by pushing with relative ease.

OTOH, I'll completely scrap pushing on anything less than a normal success (lvl 2). Using it for saving failures and partials feels just like too much of a hassle mechanically and too swingy results-wise. (P=0 for a normal success, you get either a crit or a partial/failure. No good.)

You're right in that the mechanic is not very predictable in terms of the actual numbers rolled. However, I only care about success levels. Here the distribution (without pushing) looks pretty nice to me: With increasing skill, failure becomes continuously less likely (no shit, Sherlock...) in an s-shaped curve converging on 0. The probability of a normal success rises sharply at first and then less so (also an s-shaped curve), eventually reaching .81 for maximum rolls of 18 or higher, while partials gradually shift into crits.

Comparing a fight between two noobs with one between two highly skilled characters, the noobs will have lots of normal successes and straight failures, which should make things short and messy with many full hits at 2 lvls difference. Masters will mostly get results of 2 and, with pushing on suitable first rolls (i.e. those containing at least one low die) 3 success lvls. This makes for a larger proportion of light hits (difference of 1) and longer fights where smaller advantages in the skill/damage level make more of a difference. (Mandatory addition: or so I hope.)

[fectin]

That was off 10^7 trials; it should be good for at least the first couple digits.
Remember, this was per die though.