On the Rates of Success

Nick writes, regarding the Skills: The Middle Road system:

"One thing I'd tweak us the master die; it should be higher than d12, imo. (d16, d18, or d20)"

I understand the impulse. Don't you want to be really good at what you do?

No. No you actually don't.

I work off the following assumption.

My players are real human beings, and I would prefer to present interesting content and allow them to play the game. This takes precedence over verisimilitude or simulation.

Maybe you don't think that.
I mean, you probably think your players are real human beings. Unless they are aliens. But if they were, they probably would be hiding from the history channel, and not, you know, playing D&D.

But perhaps you think it's more important to get a statistically probable result rather than one that actually is conducive to interesting play. The experience of rolling ten or more times when nothing interesting happens for an event that has minimal impact on play will likely be described as boring.
Maybe that's what you want people to say about your game? I don't know!

My sarcastic point here, is that it takes the time and attention of a living creature to roll those bones, and I'd prefer to have results of those rolls be meaningful.

What is Meaningful?

I'm looking for consequences that meaningfully impact play. Making someone roll ten times to open a door, isn't meaningful beyond the initial choice to bang around. Why not? Because if it's important, you can just roll for how long it takes to open the door.

It's important that this is a moving target! For some people, having players not know if there's actually a monster there is of high importance, so it's worth having the player roll spot checks at random intervals. Though usually with this design, it's likely an attempt to prevent the players from reacting to information their character's wouldn't have ("Wait? You're calling for a roll? Hold on guys!") and I find that's usually better handled by having players more interested in seeing what happens rather than winning.

Sometimes the result of tedium is meaningful enough to the people in the group. De gustibus non est disputandum. It's important to note, that this is decided by the Dungeon Master actually saying out loud: "Hey, are we interested in this outcome?"

On Success Rolls.

So how does this relate to our skill system?

Generating interest isn't arbitrary.

"Specific variations of intermittent reinforcement reliably induce specific patterns of response, irrespective of the species being investigated (including humans in some conditions) [emphasis added]" Reinforcement Schedules, Wikipedia

Consistent reinforcement is used to teach people the results of an action. Once they know the result of the action, it no longer becomes an object of interest. This leads to the quickest rate of extinction. (Positively: You will only use the behavior to meet the need. Negatively: You will cease the behavior to prevent the response.)

Intermittent reinforcement is used to drive behavior. The variable ratio schedule has the greatest and most stable response.

What variable ratio is best? High success? Low success?

The answer is complicated.

The test on single schedule training which most closely matches role-playing game schedules indicates that the highest interest in the schedule is maintained when there is a lean variable enforcement schedule.

What this means is, players are more likely to be invested in the results when they are rarely (~30%) successful at the start. This interest will remain the longest as their success moves towards certainty in the result. Once they are succeeding frequently enough the behavior will begin to extinct itself, which is why for infrequently used abilities (like skills) it is best to have a maximum of about 80%. This helps it maintain the variable reinforcement, and hence the interest.

  • What about combat?
    • Combat happens frequently so a 85%+ rate of success occurs often enough to be considered variable enough reinforcement. If you had a 95% rate on skills that are only rolled once or twice a session, that will likely be interpreted as a consistent response, due to the interval between rolls. If rolled 3 times a session, if you play weekly, you will only experience a failure every 2 months. 
  • What about improvement?
    • Well, there's a great example of what consistent improvement results in, the problem with skills in epic play is well documented. What happens is that skill checks cease to become a factor. You either always make the check, or always fail. Note that this can be a perfectly valid approach to superhero level style play. But it will mean that around the time players become great at skills is about the time they cease to matter.


The point here is, we don't want our success rates on infrequently used skills to be that successful. For infrequently used skills we want to start off succeeding about 30% of the time, and work our way up to success about 80% of the time - with competency around 60% to maintain interest. For combat (a more frequent roll) we face issues of frustration, so it's better to start off around 60%, and slowly up our effectiveness to about 90%.

So reaching for that d12 to roll that skill gives you a 66% chance of success (higher with statistics and good play modifiers) and that seems just about perfect to make the game engaging for the players.

Joskey Tax

Opening Door, roll high, houserule.
Declare you're going to open the door. Roll your die. On a failed result, the number shown is the number of rounds it takes to open the door.

Hack & Slash
This post took a couple of hours to write, and I loved writing it! If you enjoyed it, how about throwing some support my way, becoming a patron, and help give me the opportunity to write more like it.


  1. Well, the one thing that bugs me most in role-playing games is when infrequent rolls fail on a regular basis. I hate it when it feels like my character isn’t competent at anything. These days I tend to think anything worth rolling dice for is worth having a combat-like system to make the results less Boolean and depend upon several rolls instead of one. Otherwise, auto-success and move on.

  2. I agree with the meat of the post, but I don't like "roll for number of time units it takes" systems because they break the sunk costs/gambling dynamic by providing certainty.

    It's not like rolling a strength check (or a d6), with the knowledge that a random encounter might happen, is all that arduous. In fact, it can even be engaging, if players are on the edge of their seats about potential outcomes.

    1. How's this instead?
      Roll high for success. If lower than success your die roll is added to the next result.
      So, not certain, but contributes towards certainty.
      If you need a 4 and roll a 3, you'll get it next round. But if you roll a 1, you've only just increased your chances by 16%.

    2. Brendan, I agree.

      When I run Numenhalla, it is simply the attempt that prevents surprise. So in that case, taking 4 rounds or whatever doesn't change anything past the first attempt.

      In the case you state, it matters.

  3. Adding to the above arguments, there are way to generate interest by having a roll yield unexpected results. That is part of why we have varying degrees of success in the first place. To provide an example: a character that has a 100% chance perform stealthy actions in a world where the average detection rate for the characters preferred targets is 10%. If the player is really used to being able to avoid detection, then suddenly any situation where the skill is not useful is interesting. Even more interesting is something that has a detection rate of 100% (or better, like RQ), not because of combat or something else common, but because of an unexpected outcome.

  4. This comment has been removed by the author.

    1. For me, the best way is to provide both. In my game, average hero has about 45% to be successful at a task. Then, we give modifiers with a 15% increment to reflect difficulty of a situation (normal=0, simple +15%, easy +30%, etc).
      Basically, on character creation, the players roll d6/2 for stats, so the stat range is from 0 to 3 (or 4 in rare occasions). We multiply these by 15 to get a base skill % chance. Then players roll for skills. Skills have ranks from 1(trained) to 3(master). Each one of these gives 15% success chance as well.
      So the idea was that only true and gifted achieve that {(stat 3 + skill rank 3) × 15 = 90%} almost certain chance of success. We don't use classes or levels.

      Success formula is: (relevant stat + skill rank) × 3 + any misc. mod + d20 = 21+

      Let's say Bjorvik has DEX 2 and swords 1. His chance to connect the cut with it is, without give or take, 45%, or 12+ on d20.

      Anyways, cheers :)

  5. I agree with the gist of the post, but like you say, it can't be so much about simulation. This philosophy requires players to "center their purpose" in a way that is pretty different from almost every other major RPG out there. It is engaging though, and that's ultimately what matters to me.

  6. That pyramid image is awesome.

  7. I find it interesting that a similar argument was made by the coach of the New England Patriots this week. He suggested eliminating/changing the extra point kick in the NFL because they are converted over 99% of the time. It ceases to be interesting or meaningful until at that point.


Related Posts Plugin for WordPress, Blogger...