Thursday 5 November 2015

Roll to see if you can climb the ladder?

This post is a follow on from what I said earlier about tracking / climbing / pick lock style skills – when the character attempts a complex multi-step task to solve a static multi-faceted problem:
You cannot role play through the resolution of this – sometimes the character will succeed, sometimes fail, depending upon the specific details of the task and the character’s strengths and weaknesses. If the task is simple compared to the character’s ability you would expect them to always succeed. Conversely, if the task is far beyond the character’s ability you would expect it to be impossible for them to succeed.
These two factors give some meaning, some sense of progress, to a character’s abilities; just like the warrior can now slay an Ogre, and the wizard can now cast that cool spell, there should be a feeling that you can now track the wolf when you previously couldn't. You want it not to be too narrow a gap between auto-success tasks and auto-fail tasks – if it is always “no roll” then everything is in effect predetermined by the referee. On the other hand, the random element should not be so large that novices regularly outperform experts. The random element here needs to be a reasonably narrow bell curve, and not open ended, so not the same resolution method used in combat. I’ll talk about the solution for Explore in an upcoming post.
The resolution system for a single attack in a combat is not a good fit for such situations. Single attacks in combat are extremely variable results, but as a combat comprises many such rolls, overall these combine to give much more predictable results. In combat situations, getting +1 onto all your skills means you're twice as likely to win a combat. Hence +1 on climbing needs to be a big improvement.

Some systems address this issue with a "take 10", which is a sticky plaster for the observation that the resolution mechanic used for combat isn't appropriate for the situation at hand. Others suggest only rolling "if your character is under stress". Although this sounds an attractive approach, under this system under non-stress situations everyone is equally capable at every task.

The solution I've developed is: roll a d6 and add your skill level, attempting to beat a score. Do this three times and count the number of successes.

No Successes: Complete failure. No retry. Possibly a bad consequence.
1 Success: Failure. Can retry.
2 Successes: Succeed.
3 Successes: Complete success. You completed it quickly and with panache. No more rolls needed for this skill today until you face a more difficult task.

Three Rolls are Better than One
It is true that if we’re rolling in non-stress situations then it should add interest to the game.

By rolling three times, you get a build up to the final result – a building sense of dread up to the third roll where you know you have to succeed or you fall.

Retries:
If you get a complete failure you cannot retry until you improve the skill.
If you get a failure you can retry, but it takes 10 times as long (1 rnd=>1 min=>10 min=>2 hrs=>all day). You can retry tomorrow but starting threshold is 10 times as long (up to max of 10 min).

Note that you will either get a success or a complete failure with only a few tries, so this process cannot go on for very many rolls, though it may take a long time in game time. 99% of the time it is decided in 5 attempts, but that represents spending all day at it.

For example, if you're trying to pick a lock whilst the count is at the banquet, then the first attempt takes 1 round. Your second attempt takes one minute. The third attempt is dodgy as it takes 10 minutes and there's a chance of a servant passing. It is also your last, as you  only have one hour until the banquet ends. You can come back tomorrow and try again, but the first attempt will take 1 minute this time.

Auto-Success Situations
Because the range of results is low, there are many situations where you don't roll as you're guaranteed to succeed. For example, you don’t have to roll every time you climb a ladder. When you do roll, it is because there is a chance of failure.

Chances of Success
The chances and how you improve are shown below:

Roll Needed
Complete Failure
Failure
Success
Complete Success
7
100%
0%
0%
0%
6
58%
35%
7%
0.5%
5
30%
44%
22%
4%
4
12%
38%
38%
13%
3
4%
22%
44%
30%
2
0.5%
7%
35%
58%
1
0%
0%
0%
100%

At the bottom you are unable to succeed at all. The +1 that gets you onto the first rung on the ladder gives you a stab at it (8% chance of success), but still 58% chance of complete failure. The next +1 makes you 26% likely to succeed, and now only 30% chance of complete failure. Another +3 makes you now 92% chance of success, and over half the time you get complete success, and failure is only a faint possibility (0.5%). Any better and you cannot fail.

The numbers in the table were generated using anydice.com which is a handy resource. For example, "output 3@3d6" gives the highest result of rolling 3d6.

5 comments:

  1. I like the three roll count success method. I used something similar for a shooting contest in my Honor+Intrigue game. I also really like the increase in time for multiple attempts. Tomorrow morning I'll have a post linking to yours that talks a bit about what you have suggested..

    http://explorebeneathandbeyond.blogspot.com/2015/11/roll-to-see-if-you-can-climb-ladder.html

    ReplyDelete
    Replies
    1. Yes, I read your shooting contest and like the idea of contests within a game. I've done it twice(?) recently and it's worked well. Looking forward to reading your post.

      Delete
  2. Interesting approach. Personally I'd roll the 3d6 all at once, rather than one at a time, because I want to know the answer RIGHT NOW. This is actually the one thing that fails to completely satisfy [me] with your open 2d10 roll, by the way -- sometimes it's just one roll of the dice, but sometimes knowing the answer can be deferred for several rolls. And you still may end up with like a 9 result. But that's a minor quibble compared with its many very attractive features. (How did you develop/discover it, by the way? Were you just semi-randomly trying out different systems and found one with nice features? Or did you identify the features you wanted and design a mechanic to satisfy them?)

    W.r.t. Take 10, I think that your open dice mechanic in large part removes the need for it. If there is no "automatic failure" roll then at a certain skill level any given task becomes automatic. It's just a question of setting the difficulties appropriately. Also, there's the principle that you shouldn't call for a roll if the results don't really matter -- if someone is climbing a ladder, and there's no chance of them actually falling, and they're not under severe time pressure such that a retry would significantly affect the end result, then don't bother rolling -- it just wastes everyone's time in an uninteresting manner.

    ReplyDelete
    Replies
    1. In play it's seemed amusing when someone rolls several times but still gets a low score. It took me ages to find that particular method - I had an initial idea of the distribution I wanted and tried out lowest of 2d30 which didn't work well (they roll for ages!). That gave the sort of drop off I wanted, but I decided I wanted a bell curve with an open ended top, and tried all the variants I could think of before stumbling on a d6 version of my system (though I'm sure it's been used before). Re-rolls were too frequent, and no-one liked discarding a 6 for two new dice (originally the re-roll was optional) - switching to 2d10 fixed all these issues as it's discarding a 0.

      One thing this did teach me was that a cool idea isn't enough for a mechanic, it needs to work in play.

      I developed the subsystem in this post because I found that although the open dice system worked well in combat, the results for climbing in particular were too random. The bad climbers were often outperforming the good ones, so I went back to the drawing board to ask what I was wanting the dice to do for me.

      Delete
    2. Well, at least it wasn't obvious. :)

      The long-tailed bell curve distribution of 2d10+ (how I refer to your system) itself does a fair bit to moderate the wide variability in results for some tests, like climbing or (especially in 3rd edition era D&D) jumping. Using a d20 + modifiers where the result indicates how far you jump REALLY doesn't model reality very well at all; forcing a normal-ish distribution in there does a lot to mitigate that and make results more predictable while still leaving room for the occasional surprise. But I can certainly see wanting to further reduce some of the randomness there.

      Delete