Probability of Success and Advancement

[ioihlmn]

Hello all, new guy here to ask a simple question.

Before delving into the creation of a dice mechanic I would like to have a general idea of some specific probabilities.

My initial feeling is that a character of average talent and and average training should only be able complete an average difficulty task 60% of the time under normal adventuring conditions.  Do you feel this is an appropriate probability for a more savage and deadly setting?  I am hoping that such a probability could lead to more tactical decision making from the players in an attempt to acquire a more advantageous situation from which to attempt a skill test.

Similarly, I would like to base skill advancement on a specific event occurring within a skill test.  I have toyed around with a 5% probability of this event occurring with advancement proceeding after 5 occurrences.  This will not be the sole means of skill advancement but one of the more important aspects.

Do these probabilities feel right by your estimation, or would you recommend a different numerology.  I feel I need some additional perspective on this issue.

Thank You - LB

[Luke]

Raw percentages aren't very meaningful. I'm not being pithy with this question: How do players generate random results in your game? 2D6? 3D6? Binary die pool? D20? Percentile D10s? Other?

-L

[ioihlmn]

My leading in with "before delving into the creation of a dice mechanic" in my original post was meant to imply that I had not yet made a decision on the randomization aspect, however for sporting purposes lets throw out a D6 dice pool as the chosen randomization mechanic.

Thank You - LB

[Adam Schiller]

Similarly, I would like to base skill advancement on a specific event occurring within a skill test.  I have toyed around with a 5% probability of this event occurring with advancement proceeding after 5 occurrences.  This will not be the sole means of skill advancement but one of the more important aspects.

I think that, to some extent, the constant rolling of the "advancement event" occurrence whenever a skill is used might very possibly be considered extraneous to some extent. The idea, as a whole, is very nice and provides a welcome alternative to the standard "level-up mah skillz" systems out there. (Now that I know it, I'm sure I'd use your idea if I was running a game.) But what about simply leaving it up to the GM--as in:

SKILL: Marksmanship
LEVEL: 1 (can hit a target at ten metres)
___So basically, the character's skill enables them a ten metre range. Firing at anything below or moderately above ten metres won't really "press the talent" to improve. But if the character is trying something at 2x the norm (20m) then the chances of success are halved, but improving becomes, say, 50%, if the skill-check is successful. If they're at 4x the norm (40m) then the chances of success are quartered, but improving becomes, say, 80%, if the skill-check is successful. After accumulating 100% on an "improvement rating," then the overall skill becomes improved.

Basically, by willingly combining a "success metre," then the system challenges the character to go above and beyond the norm, rather than continuously executing the same skill in order to achieve a level-up. It might not be what you're looking-for, and if these ideas are worthless to your design idea, I understand, but as a player and GM, I tend to think that random events and "chances" are less realistic than keeping track of improvement and relating them to actual events.

[Callan S.]

Hi Lance,

I think the consequences of failure are more relevant than the chance of failure. In play I once said there was a 2% chance of failing a task, but if the character did fail, they died. The player didn't take it - it surprised me at the time. Also if the consequences are nothing much, it doesn't really matter if you only have a 60% chance of failing, you'll just try again or do something else. Rolling wouldn't result in much.

[Egonblaidd]

I want to second Callan here, both the chance of success and consequence of failure are important.  In a more savage setting, I'd imagine, for example, in combat that the chances of striking a successful hit could be quite low, but failing to hit does nothing except give your enemy an extra chance to hit.  Striking a foe, however, would cause serious damage.  A more concrete example might be a combat system in which case characters (PCs and NPCs) have a 20% base chance of hitting, but one hit is an auto-kill.  In this case, missing (the consequence of failing) heightens the tension since your foe now has a 20% chance of killing you before you get to attack again.  If, however, "hitting" merely kept you alive, and it was "missing" that killed you, then the chance of success would have to be drastically altered (or switched).  Or, as in the former case of one-hit-kills, it would probably not be a good idea to have a 60% base chance of hitting, while using some sort of damage tracking system, like hitpoints, would make much more sense with higher chances of hitting, like maybe 80%.

Another example might be spell casting.  If a failure merely means that the spell fizzles, then a 60% chance of success could be appropriate.  But if a failure means that the magical energy being channeled detonates before being released, damaging and/or killing the caster, then you would want a much higher chance of success, probably somewhere between 95%-99%, depending on how lethal it is.

Also, I think it's a good idea to think about probabilities before thinking about what sort of random number generator you are going to use, but looking at different random number generator systems can inspire more creative ways of approaching resolution.  You could have a system where you can get multiple successes or failures (usually by rolling several dice that independently generate a success or failure), which could in turn dictate the quality of the result.  Or you could have the quality of success being independent of whether the action was actually a success or not (for example, the number of successes is any set of dice that come up with the same face value i.e. all 2's, all 5's, etc. and a certain number of successes are required to succeed, while the quality of success is dictated by the face value of those dice, or vice versa).  Especially since both the probability of success and failure and the consequence or quality of success and failure are both important and intertwined, it is important to approach the problem from both sides.  Do actions simply succeed or fail, or do they have varying levels of success?  What are the consequences of success or failure, both for a PC action and enemy NPC action?  What would be best probability be for a success or failure given these consequences?  How should characters improve their resolutions when advancing?  Just some things to think about.

[Luke]

My leading in with "before delving into the creation of a dice mechanic" in my original post was meant to imply that I had not yet made a decision on the randomization aspect, however for sporting purposes lets throw out a D6 dice pool as the chosen randomization mechanic.

Thank You - LB

Well, that's just it. Tossing around percentages isn't the most useful exercise. How the numbers are generated is.

Tell me to roll a 7 or less on 2D6 and I feel pretty good about it.
Tell me roll 60 or under on percentile 10s and I burst into hives.
Tell me to roll four successes on six D6, and I hope for the best.

I'm partial to die pools. There's more information in each roll that can be manipulated.

Anyway, 60% on odds on a starting base ability is fine, so long as, as Callan suggested, the scope of success and failure are properly set. The risk and the reward of a test must seem worth it to the player.

And random advancement works. Jared Sorensen uses it effectively in Inspecters and Lacuna.
How often do you want the character to advance in a session of play? Once? Twice? Once every other?
How many tests will each player make during the session? If he's making 20 or more, with a 5% chance of advancement, he's likely to get at least one advancement per session.

20 seems like a lot, though.

-L

[chance.thirteen]

My response to the original question: Yes, I think that 60% feels about right. I also think that going with your own gut for such a nebulous seeming decision is good. You want your design to vicereally work for you. Also note that you can go the route of saying a typical test isn't easy, but there chances for modifiers, so a character could be rated at 60% to hit a target with a spear, but that aiming at an unmoving target while not in combat raises the chances of success.

[ioihlmn]

Thank you all for the excellent food for thought.  I would like to address some statements.

Regarding advancement:  I envision this particular advancement mechanic to be used on a very low (yet vital) numerical set, perhaps having characters begin with a rank of 1 and having 5 or 6 be the upper end of the spectrum.  I do approve of the idea that the probability of this occurrence should be based on going "above and beyond" a normal skill check so that characters will not repeatedly seek out skill checks solely for the sake of advancement. I will take this into account as I mull the mechanical possibilities.  As far advancement per session, I would imagine with a rank of 5 or 6 being considered "great," advancement should be a slow process (and increasingly difficult).

Regarding consequences: I agree that a "quality of success" mechanic is inevitable.  A slight failure should not have the same consequences as a catastrophic failure be it in combat, debate, or driving.  I should have clarified in my first post that I am assuming a 60% probability of an average character completing an average task with a "bare-bones" success.  Naturally the probability of a "near success" failure would be similar and a "fantastic" success much lower.

I would like to place the importance not on what you can do numerically, but how well you can position yourself to do it in the course of play. By making the probability of actually scoring your skill level around 60% I would like to make the character's strive to improve that probability by making extra effort to plan actions before attempting them. My question is: "Is having a 60% chance of scoring "average" if your skill is "average" motivation enough?

I would like to implement the same motif for opposed rolls (particularly combat).  Before actually entering a conflict your goal should be to surprise, trick, confuse, or otherwise influence your opponent into a disadvantage that makes it statistically harder for them to attain their level of skill (not just arbitrarily lowers their skill or randomization result).

Thank You - LB

[Callan S.]

Well, for myself, I didn't get into anything about quality of success. I was refering directly to ensuring there's a strong conseqence to failure. The issue of climbing a fence isn't about quality of success, it's that bugger all happens if you fail.

[Adam Schiller]

I would like to implement the same motif for opposed rolls (particularly combat).  Before actually entering a conflict your goal should be to surprise, trick, confuse, or otherwise influence your opponent into a disadvantage that makes it statistically harder for them to attain their level of skill (not just arbitrarily lowers their skill or randomization result).

Let's just get one thing out of the way before going any further: What you're proposing is going to be tough on GMs, no questions asked. There are some GMs who don't mind the extra work, and would see ideas like yours as tenets of a "great and innovative system," but there are many others (particularly people who first pick-up the book) who would, in their attachment to other systems, be frustrated by it in gameplay or, if selective enough at the bookshelf, not pick it up altogether simply because it's rather complicated.

But it doesn't have to be.

I would recommend: Make a list of what you want to have happen in the role-playing world itself. Don't worry so much about the game mechanics. Then try to accomplish the goals in a mechanics sense in three, four, or five different ways. The more disparate the ways, the better your chances of making the best system you can. For example:

GOALS
1) Characters improve skills based on "critical events."
2) Characters cannot abuse remedial use of skills to advance.
3) Advancement should be difficult but not near-impossible.

PATH ONE
1) Critical events are defined as emotional outbursts in the heat of a moment where the character extends beyond their normal comfort zone.
2) The player character cannot judge what is critical and what is not; only the GM can.

PATH TWO
1) Critical events are defined as the character attempting a skill they already have but against immense odds, thus incurring a penalty to their normal roll.
2) There is a table for penalties dependent upon challenge.
3) The character will tear a muscle if failing a physical skill or suffer a seizure if failing a mental skill (etc. for magic et al.)

Only you, the designer, can find the right fit for "reality" (for the character) and "enjoyment" (for the player). Try a bunch of different time-lines, follow your own stream-of-consciousness, and after a while, combine if you want elements which you like into a final version of the way to get it done. Best of luck and keep us posted!

[Luke]

I do approve of the idea that the probability of this occurrence should be based on going "above and beyond" a normal skill check so that characters will not repeatedly seek out skill checks solely for the sake of advancement.

Emphasis mine.

This is not a matter of sculpting a perfect probability of occurrence. Very few players are motivated by pure probabilities. Players are motivated to use the easiest mechanic possible to derive the greatest reward.

This is a matter of game design. Your skill system must include more than just probabilities. You skill system must include when a player tests, how a player tests, the resolution mechanics of that test (aka probabilities, but also other stuff), the results of said test and how those results tie into the game at large in the cycle of tests in specific.

Too many RPGs plop the resolution mechanics down in front of the players and expect them to figure out everything else on their own. This is bad game design. If you want to discourage players from "seeking tests solely for the purpose of advancement," your whole system must guide the player, indicating how, when and why it is acceptable to roll. Simply setting a baseline percentage is not enough.

Regarding consequences: I agree that a "quality of success" mechanic is inevitable.  A slight failure should not have the same consequences as a catastrophic failure be it in combat, debate, or driving.  I should have clarified in my first post that I am assuming a 60% probability of an average character completing an average task with a "bare-bones" success.  Naturally the probability of a "near success" failure would be similar and a "fantastic" success much lower.

I would like to place the importance not on what you can do numerically, but how well you can position yourself to do it in the course of play. By making the probability of actually scoring your skill level around 60% I would like to make the character's strive to improve that probability by making extra effort to plan actions before attempting them. My question is: "Is having a 60% chance of scoring "average" if your skill is "average" motivation enough?

I would like to implement the same motif for opposed rolls (particularly combat).  Before actually entering a conflict your goal should be to surprise, trick, confuse, or otherwise influence your opponent into a disadvantage that makes it statistically harder for them to attain their level of skill (not just arbitrarily lowers their skill or randomization result).

Again, these are great goals. But if you're sincere in these statements, then this is what you game is about. Your game is about planning, cooperation and teamwork. Your game damn well better have cool planning, cooperation and teamwork mechanics.

Setting a number and tossing the problem in the players' collective lap to figure out isn't enough. That quickly devolves into "entertain the GM until he gives us a pass."

-L