Twists on Roll-and-Keep

[Tom]

For a new concept I'm working on (hi, I'm back to those who remember), I am thinking about a variant of R&K and would like to bounce the idea off someone or somethree.

I like R&K because it comes very close to the statistical properties that I like in a dice system - higher skill equals not just more success chance but also more consistency.

I also like some R&K implementations where, for example like in the old L5R blood magic you actually choose dice instead of just blindly picking the highest.

Then I like the ORE for its fast resolution, multiple-meaning concept, i.e. reducing the number of dice rolls.



So... combining all these things that I want, I came up with this idea:

Like in most dice pool system, you sum up your skill level and any modifiers and get your dice pool. I imagine 3-6 would be a typical size, with < 3 being crippled or idiots and > 6 being the best of the best. I would also cap the pool at 10 (no matter what modifiers, you never roll more than 10 dice).

GM announces TN, scale based on L5R, so 15 is an average difficulty.

Instead of keeping some fixed number, you now allocate your dice to three things, as you wish. You can use a die to reach the TN (i.e. to succeed), you can use a die to raise the effect (more speed, more damage, etc.) or you can move it into your karma pool.

The karma pool is my version of "roll over" successes, and a way to make very successful rolls have a lasting effect beyond this roll. Basically, if you get all you want from your roll without using all the dice, you can move the remaining dice into your karma pool, as long as they are higher than the current number of dice in the karma pool. So the more karma you have stored up, the higher (and thus more valuable) dice you have to move out to fill it higher, and it can never be higher than 10. This mechanic also allows you to draw something good out of a failed roll. I might have to cap this at 1 die per roll max to prevent abuse, or not, I'll see.

The success is easy. Add dice until sum >= TN. Basically, you will add up the highest dice until you have reached the TN.

Raising the effect is the equivalent of raises or additional successes, but without the math. And/or it is the equivalent of width in the ORE system. This is most clear in combat, where the success (TN) indicates if you hit, and additional dice can be used to cause more damage. And/or for initiative (the way ORE handles it). You have to allocate at least one die to success and one die to effect. If your pool is less than 2, for any reason, you roll 2 anyways and get some other penalty.




The idea here is that a master will have plenty of dice, and on easy tests can use a lot for additional effects (fast, powerful, whatever). Additionally, the system handles trivial tasks well, a challenge for most dice mechanics. If I want to figure out if I can tie my shoes, the chance for failure should be effectively zero, but I can be fast or slow this time. If I want to run 100m, there is little doubt I will make it - but how fast? Simply set the TN to a very low number (setting it to the size of the dice pool guarantees success) and check how much is rolled for effect.

There are some minor complications with 1s that I'm thinking about, but they don't affect the core system.


So, thoughts? I'd like to know if this seems reasonable, intuitive and playable. I like to give players meaningful choices and I would like to extend that to the dice rolls. In most systems, they are too passive, I want to use dice to expand player choice and strategy instead of reducing it. What do you think?

[RangerEd]

You have my attention and my interest with you brief description. Do you have a core system-like write up? I'd like to see a more precise explanation.

Ed

[Tom]

At this stage I have a mess of notes, that's all. The basic idea is roll-and-assign, so instead of discarding dice, you put them into different parts. It could even be more than two. For combat, I was thinking you usually have 3 elements - you want to hit (success), as hard as possible (damage), as fast as possible (initiative).

I see problems with back-and-forth for contested rolls (A sees that B is faster, so assigns more dice to initiative, which makes B assign more dice there, so then A gives up and moves dice back to damage, now B has too many dice on initiative he doesn't need and moves them back...) - but those problems are fixable.

But I also wanted to have a system that is the same for dramatic and combat purposes. So more generally speaking, you usually want to do something good, cheap and fast (pick any two, as the saying goes). Assign dice to those dimensions.

[Tom]

Reading a lot about dice mechanics (again, I keep coming back to this topic and discovering more) I found that Cortex Plus does something similar. It's roll-and-keep plus use the highest discarded dice for effect.

I also found I really like the philosophy of Apocalypse World, that every die roll should move the story forward and "nothing happens" is never a valid result. I will try to incorporate that.

[RangerEd]

Keep going. I may be weird, but the statistician in me finds dice mechanics inspirational for story generation. I like the path you're on and would like to see the waypoints you find. Keep us posted on your journey.

Ed

[Tom]

Yes, the most true thing about dice mechanics I have read in everything I've read is that there is no perfect dice system, it needs to fit your game.

A simulationist hard SciFi system could go well with a complex dice mechanic that would cripple a narrativistic, fast-paced superhero game.


So, merging everything, here is what I have come up with so far for my game. The story is demonic, alternate-reality middle ages. Player characters are demons escaping from hell, trying to do good in a world that doesn't always make it easy (and the game mechanics will actually reward them for being bad, too).

As such, I've decided to use a d6 roll-and-keep system with the keep number fixed at 3.
While this ties my hands a little when it comes to manipulating the probabilities (changing the keep number is really cool for statistics) it gives me two cool things. The first is that expected outcomes are easy to estimate. For 3d6 it's 11 and it goes up by about +1 for every additional die. The second is that the best roll possible is 666 which fits perfectly to my theme, while 777, the number of god, is not possible with earthly dice.

However, my dice are read twice, to resolve everything in one roll. Firstly, you sum them up to check for success. Here I use an inspiration from Apocalypse World, meaning there will be at least 3 degrees of success, and no "nothing happens" result.
But secondly, you also assign the individual dice to mean something. In general, you assign one die for the finesse or quality of your action, one to the quantity or strength of effect, and one to the effort or efficiency. In certain circumstances, these 3 parts will have more precise meanings, e.g. in combat quantity will mean damage, while quality means how well you hit what you wanted to hit (and how - if you wanted to knock someone out, but assigned a 1 to quantity and a 6 to damage, the GM can rule that you've killed him instead).

What this means is that if you roll 6-5-2 you have probably succeeded (13 is an ok roll), but the low die means something went wrong, and you get to decide what.

I will in general use low TNs for success/failure, because my general assumption is that people are perfectly able to tie their shoes and walk up a flight of stairs, unless someone is actively trying to stop them.

[Tom]

Why? Is the question every die mechanic decision needs to answer. So I'll try:

3d6 => because it gives me the thematic 6-6-6 as the best possible die result. As I don't forsee that I need very fine-grained results, d6 are just fine, I don't need d10 or d12.

Sum-and-assign => why do I use both? Because I want the feature that you can reach the same goal in different ways. 1-6 and 3-4 are both 7, but they have different meanings. They both mean you succeed, but one means you succeed at great cost, for example.
I also like that due to the bell curve distribution, there are only a few ways to get a great result, but many ways to get mediocre results. My favorite metaphor for skill in roleplaying is archery, and it fits here. There is only one bulls-eye, and missing the target altogether is all the same as well, doesn't matter where exactly you hit, but in the middle, there are many scores, and you can hit to high, too low, too much left, too much right. You can scatter shots widely (low finesse) or cluster them close but in the wrong spot (high finesse).

roll-and-keep => because reducing the die number that are counted for a result while varying the number that is rolled means higher skill = more dice which fits nicely with player expectations, give a visual and audible feedback on skill, and a simple skill system (more points = higher skill = more dice) and sets me up for a 1:1:1 system like Sorcerer which I like a lot. Changing the dice type with skill, for example, is something I've always found awkward. But counting only 3 dice for the result makes all math easy to handle, keeps all die results within a fixed range.

[RangerEd]

Tom,
I like the idea that 1-6 is different from 3-4 on a sum of 7. How might you inspire the players to understand the difference and bring it into play?

I'm curious, would a 1-6 be different than a 6-1? Have you thought about the order in which the dice land when rolled? The color of the dice pools? Each of these pieces of information exist within a dice roll at the table, and could be used to squeeze story telling potential out.

I have a failed (to playtest) set of mechanics along these lines. The link is here... http://www.herogames.com/forums/topic/90242-a-dramatic-jacket-for-our-hero/?hl=%2Bdramatic+%2Bjacket

If you have any notes for me, I enjoy this kind of discussion a great deal.

Ed

[Tom]

As outlined above, I don't differentiate order, all dice are equal (at least for the resolution).

So 6-1 is the same as 1-6, but not the same as 2-5 or 3-4. Players would understand it with some examples. Let's ignore the 3rd die (resources/effort) for the moment to simplify things.

A roll of 1,6 on a rally speech means you can either give a brilliant speech but only a few people really listened (but they will be very convinced of whatever you said), or you gave a barely convincing speech, but everyone in the room heard it.

Now for combat, I have to add that my combat system doesn't bother with individual blows. One die roll resolves an entire exchange, a series of maneuvers. Instead of combat rounds, I will have combat scenes. So initiative doesn't matter, and the result of the row is actually meaningful, i.e. the roll determines who won the encounter.
So in combat, one die indicates your victory and the other the damage you dealt. You can win but do hardly any damage (you mostly hit armour, maybe, or glancing blows, but at the end the other guy is down on his knees) or you can lose but anyway you roughed him up badly. This is handled by characters staying intentions for encounters, which don't always have to be "I kill him". For example if you want to get past that troll, you have to state that as your intention, or no matter what you roll you'll not be past. In such a narrative combat system, seperating success and damage makes sense (though sometimes the intent can be "kill the fucker", in which case I'll demand both dice high or you've bloodied him, but not fatally).


Again, all this is rough notes and subject to many changes.

[Ron Edwards]

Hi Tom,

I've wondering how to post what I have to say nicely, and I confess I'm not seeing how. So ...

Bluntly, it doesn't matter. Resolution methods are meaningless outside of their context. That context concerns three levels.

Closest in: how does the resolution mechanism relate to what is being said? How is what can be said next subject to constraint by that mechanism? How much?

Next out: what sort of things are being resolved, which is itself a function of situation. Situation means not only external problems or the current weather or whatever, but also what aspects of character/characterization are being brought to bear.

And finally, largest: all of that is in a larger context of what about any element of the imagined fiction is, short or long run, subject to change.

You can talk about this dice thing all day and without these, it will get nowhere.

I do think decisions about dice and resolution matter immensely - there are several dozen games out there whose resolution feature are terribly mismatched for what else is working/going on in each. But that isn't solved by deciding whether a given mechanism is "good" in isolation.

[Tom]

Agree and disagree.

You are right that there's a lot of context missing from this post because I have it in my head or on notes, but I didn't include it.

However, some game mechanics are related not to the full background context, but to fairly simple design decisions. If I wish for a fluid gameplay, my mechanics need to be fast and easy so as to not slow down the gameplay, so I need to know that comparison > addition > substraction > multiplication.


My design decisions for this system are:

I want a narrative, fluid game where the dice mechanics matter, but not in great detail. From this follows that I don't need fine-grained resolution and I don't want excessive lists of bonus and penalty systems.

It also follows that I do not want to roll a hundred dice to resolve something that in the narration is one thing. Any conflict or obstacle should be solved with very few, ideally one, roll(s). From this follows that my rolls need to be more meaningful than "success / failure". They need to show degrees of success and strength of effects.

And finally I want to focus dice rolling on the players, with the GM rolling very rarely, if at all. That has to do with the setting, but also to unburden the GM and let him focus on the story and the characters. This means I cannot use an opposed rolls system.




So, without touching the setting very much, I already have a large number of constraints on the dice mechanics.


Next, I go into the setting a little, juggling little ideas that I like and checking if they fit or not. I decided upon keep-3 and d6 because the background is one of demons in a medieval setting, so it is very fitting that 6-6-6 is the best possible roll. I couldn't decide how to manage the keep mechanics for some days and felt that I just needed to make a constraint to move forward.

Then I felt the Apocalypse World model of "every roll moves the story forward" and the "success-at-a-cost" concept would fit in very well with my concept.


So now I am here, trying to put all the pieces together. I focus on the dice mechanics in this forum because the rest of the setting is very much still in my head and needs to be structured and put to paper. So yes, sorry for not providing more background information. Doesn't mean it doesn't exist. We agree on this, a mechanic does never exist in a vaccuum, even when you're making a generic system (GURPS, AE, Mythic, etc.)

[Ron Edwards]

With respect, that point only applies in comparing dice mechanics that aren't very good.

"Roll single die + add to attribute + add to skill +/- various modifiers vs. TN" is a crappy mechanic, terms of what I've called in the past Search Time and Handling Time. So is "roll 3d6, add, +/- various modifiers vs. roll 3d6, do the same." Emphatically, so is "roll a pool of dice, compare each to a TN, count those who succeed, compare to similar roll made by someone else."

When comparing a number of similar mechanics like these and deciding which is least aggravating, their Search and Handling Times can be compared in isolation. They can also be sanded-off within the assumptions of each. That's what you're doing here as far as I can tell. The problem with that as an ideal is that there is no "ah ha perfect" moment you can arrive at - anything that feels that way is deceptive.

I'm suggesting a different logic that isn't well-known in the hobby at all. It has nothing to do with setting as generally conceived, which is why I didn't use that word and didn't ask to know about it. I wrote about situations in play and what their outcomes have to do with any imagined component changing.

Here's another way to look at it. It is possible that your game includes character improvement over time based on events and possibly mechanics of play. If not, then I would need to know what does change, but for now, I'll stick with character improvement, the most common type of change in RPG design. In some games, the connection between this improvement and resolutions in play is mechanically minimal or perhaps, coarse-grained: as long as your character is still alive, you get a fixed amount of improvement currency per session. But in others it's very fine-grained, down to individual moments of resolution determining chances for improvement for that one skill. The difference is not in good vs. bad game design; it's in whether the designer wants improvement to be "generic session by session, everyone gets the same, and spend as you see fit," or "directly consequential and traceable from single moments of play, highly individualized per character's mechanical history."

The mechanics of resolution are less important regarding success-or-failure than they are regarding their relationship to improvement - which is to say, how you think you want players to experience character improvement. [again, all of this assume that such improvement is the primary changing feature of this game; that feature might just as easily be something else]

That's why I'm asking about situations of play - that lets me know what is being resolved fictionally. The word "task" is emphatically not enough; it's a dodge for not knowing the answer. That's also why I'm asking about what changes.

There is definitely such a thing as a bad dice mechanic. But the issue here is that I think it's better to talk about good ones. And good in this context is not merely the repair of the bad.

[Tom]

Yes, there are a lot of very crappy dice mechanics out there.

I am trying to reduce mine with two approaches.

The first is that I avoid lots of very specific modifiers. Those "+1 for aiming, -1 for partial cover, +1.5 for the scope, 20% deduction for skill and +1d4 because it's Monday" lists. There is only one case in which dice values are ever modified in my system. Everything else is "roll x d6, keep the highest 3, assign them to success, effect and resource depletion".

The second is that I reduce the number of rolls made by handling entire situations with one roll. Instead of 20 combat rounds and individual rolls for each attack and defense and wound, a fight will usually be one, two, maybe three encounters. In movie language, that's one scene. So a combat roll is not one strike, but a series of exchanges. This brings combat closer to other resolutions where it is more common that you make one roll to resolve your entire seduction or research or vehicle repair, not roll for every move seperately.


I'm also trying an approach that will put Search and Handling Time back into the game through interpretation. Instead of comparing dice values with TNs and then looking up the difference on some table to find out what it means, dice have fixed meanings, and all manipulation goes to the size of the dice pool only. By assigning dice, players choose the outcome of their actions - dice values become meaningful.

So, I have success, effect and cost. In all cases, a 6 means total success, 4-5 means success, 2-3 means failure and 1 means catastrophic failure.

If your pool is 5 dice and you roll 1-1-2-5-6 (which I actually just rolled), you drop the two 1s and now you choose. In case you can't decide, you usually want them in highest-to-lowest order, but maybe not. With the default choice, you would have had a total success (you got everything you aimed for), you made a big impact, but it was costly in terms of resources. If you are low on resources, for example, you can decide to get less effect, but save on resources.

In combat, this could mean you hit exactly where you wanted, you made a lot of damage but it cost you a lot of ammunition (ranged) or exhaustion (melee). Or you could choose to make less damage, but save on ammo.

Total Search and Handling time, IMHO very good considering you get the entire encounter resolved meaningfully in less time it takes many systems to resolve one attack.



For character improvement I have a very specific system that depends on the background setting a little. What you need to know is that characters are demons who inhibit humans, so their abilities (dice pools) are the sum of what the human can do plus what they add to it. This allows me to start out characters with ZERO abilities. What they can do is use magic (from a limited pool) to "boost" the abilities of whatever human they are in at the time. Every time they do so, players put a mark near the boosted ability. Once they've done it a certain number of times, that boost becomes permanent, they can use it with zero cost, and they can use their magic to boost higher, with the same rule (do it x times, it becomes permanent).

So in your terminology, my mechanic is directly consequential from single moments of play.

However, character improvement is not by itself the primary changing feature. It is one half of the equation. The game is much about ethics and consequences (as a game with demons should be), so one part is getting better, the other part is HOW. Like in the Star Wars games, characters can advance on the good or the evil, and I also have a third, neutral, axis. These are all mutually exclusive. So in addition to the numbers part, there is also an evaluative part. I'm still working on the details of this one, the mechanics is that your power in either of these three dimensions limits your magic pool and how much you can boost anything. Like the ability improvement, this is a result of in-game actions.

The concept of both of these is to start literally with an empty sheet, and go from there. Character generation consists of exactly one step: Write your character name on the top of the character sheet.

[Tom]

While it's an old topic now, I have made steady progress in aligning this idea both with the game concept and refining it in terms of mechanics and probabilities.

My current working version will probably go into playtesting and it looks like this:

Determine dice pool size (ability + situation modifiers)
Typical pools being 3-6 dice, but the system allows virtually any number (even 0 and negative, hang on for details). Dice pools beyond 8 will be very rare, and while I don't have a a hard cap, I don't think you will see more than 10 in this game.

Roll your pool and keep the highest 3.

Special rule: If the dice pool is less than 3, you ADD dice instead of removing them, but you keep the LOWEST 3. So basically I have:

0 - 6d6kl3
1 - 5d6kl3
2 - 4d6kl3
3 - 3d6kl3 = 3d6kh3
4 - 4d6kh3
5 - 5d6kh3
...

Assign dice (usually from high to low) to goal, effect and cost, as outlined above, with one more clarification:
You need to succeed in both goal and effect to get a real success. If you succeed in one, but fail in the other, you get a hollow success (a hit with no damage, or lots of damage but your target remains unharmed).

Dice are read individually with
6 - critical success
4-5 - success
2-3 - failure
1 - critical failure

This gives me a fairly nice probability distribution:



The green line showing the 2+ success requirement for a real task success gives me a 50% probability for 3 dice, and 3 is the average/default human ability in my system.

I also have really good critical probabilities, with a critical failure being really, really rare (less than 1% for 3+ dice, and even with really low pools it stays low - ca. 6% with a 0 pool which can almost never occur). Critical success, on the other hand, rises nicely starting at 0.5% for a pool size of 3 and reaching 10% at the 7/8 pool size and 20% at 10.

This is the first time I have a dice system where increasing capability increases both the success chance and the reliability (i.e. reduces standard deviation) in a smooth curve.



Handling is pretty good, I believe (playtesting will show). While I'm not happy with the conceptual break between 2 and 3, it should occur rarely as most player characters will roll 4, 5, 6 dice in most cases. Aside from that, rolling and resolving is straightforward. Roll x dice, keep the highest 3, interpret the result, basically as success/failure. No math operations are required, all dice results are interpreted by comparison.

[Tom]

I should also mention that I make it an explicit part of the rules that a roll always moves the story forward, because a roll resolves a task. Any re-tries or try-it-differently is already included in the roll. The result is final and on at least one of the three dimensions, something has to change in the story.