[Design] Fw: relative strength of randomness and skill

[Kyle Schuant]

----- Original Message ----- 
From: Kyle Schuant 
To: Lev Lafayette 
Sent: Monday, October 17, 2005 11:51 AM
Subject: relative strength of randomness and skill


From: Lev Lafayette 

OK, the idea is pretty simple. The basic principle is
the degree of randomness varies according to the
action performed.
...
Depending on the type of activity or their relative
importance of the incident to the story, different
skills will have a different influence of randomness
in determining the Trait Effect. 
****

OH NO, NARRATIVISM!

Okay, so that's, "if it's important to the story, randomness will be important, too. If it's not important to the story, not so much." 
    That's what we miight call a "Narrative" approach. Or "storytelling" or whatever. I'd call it a "dramatic" approach. Problem with these words is that they all come from movies... and in movies, we have the opposite effect. The more important a task is, the LESS likely a PC is to do really well or badly. The less important the task, the more random the result! So Jackie Chan may fumble making a cup of tea, or may make the most delicious dinner ever. But he will almost never fail when sliding down a large ribbon through the middle of a multi-storey mall, and landing on a foe and knocking him out. 
    In rpg design, we keep aiming for the opposite of movies, whereas players keep hoping for action movies... I think this is perhaps a reason for trouble in many game groups!
    In rpg design, we shie away from the idea of making ridiculous tasks easier than simple tasks, because we feel that players would do ridiculous tasks all the time, then, and the game would become... ridiculous. Many systems solve this by having a basic system of, the more difficult the task, the greater the degree or need for randomness (ie, "you need a good roll to succeed"), and balance it for the "movie" feel with some system of Hero Points, allowing you to buy successes or rerolls. 


HELP, EVEN WORSE, GAMISM!

    Then there's the approach of simulating reality. My view is that as you become more skilled in an area, random chance plays a smaller part. So for example there's the 100m sprint. I have no skill in it, I may run it in 14.50 sec today and 12.50 sec tomorrow. My performance will vary by +/-1 second, or +/-1 7%, roughly. But your Carl Lewis will run it at 10.01 sec today, and 9.92 sec tomorrow. His performance varies by 0.09 seconds, or +/-0.05%. His higher skill doesn't just make him faster, it makes his performances more consistent, too. 
    In most games, however, the randomness of the dice rolls is very great compared to the skill/attributes involved, so that in say 10% of contests, I could outrun Carl Lewis. 
    So that's why the dice number range compared to the attribute/skill number range is very important. For example, in Rolemaster, the number range for dice is 100 (roll d100 and add to to skill+attribute bonus), with a 5% chance of going below that, and a 5% chance of going above that. It takes up to about level 10 for most characters to have skills of around +100. So there's several months of play (of weekly sessions) before the skill number will match the random number in importance. A skill +40 guy has a decent chance against a skill +70 guy, because the random roll is on the order of 100.
    Whereas in Runequest 2, which was a roll-under system (attacker roll to hit, defender roll to parry; attack and parry were separate skills but let's set that aside for the moment) a skill of 70% was FAR superior to one of 40%. The higher skilled guy would roll to hit (70% chance) and the lower-skilled guy roll to parry (40% chance), for an overall (70% chance to hit x 60% chance of parry failing) 42% chance of the higher-skilled guy striking for a wound; whereas the lower-skilled guy has just as (40% chance to hit x 30% chance of parry failing) 12% chance of striking for a wound. 
    In Masterbook (skill+attribute, adjusted by d10-d10, compared to difficulty number, highest wins), a skill+attribute of 7 is quite superior to one of 4, because the d10-d10 roll will make the performance cluster around the skill+attribute level. If you had a roll-under system (d10 roll, vs 0 to 10 skill+attribute), then the attacker with skill 7 vs the defender with skill 4, it'd be just like RQ2 as above. But with the skill+attribute being altered by d10-d10, you get a different result. The results are their chances are, -3 (7%), -2 (8%),  -1 (9%), 0 (10%), +1 (9%) and so on. So the chance of the 4 beating the 7 is, success if they get +3 or more. The chance of +3 or more on d10-d10 is 28%. How to illustrate it, um....
    Let's look at it as a ladder. Imagine there are possible ranges of results, 1 to 10, skill plus a d10-d10 roll, and the results are capped at either end. So if you have two guys of skill 4, the two ladders look like,

10   10    (10%, coming from 4% chance of 10 alone, plus 3% for the prohibited score of 11, 2% for 12, 1% for 13)
  9   9      (5%)
  8   8      (6%)
  7   7      (7%)
  6   6      (8%)
  5   5      (9%)
(4)   (4)    (10%)
  3   3      (9%)
  2   2      (8%)
  1   1      (28%, coming from 7% chance of 1 alone, plus 6% chance of the prohibited score of 0, 5% for -1, etc)

Each has equal chances of success. The ladders are matched in probability, the chances of each result for each guy noted. . Their performances will hover around 4, with a 10% chance of its being 4, a 10+9+9 = 28% chance of its being 3, 4 or 5, a 10+9+9+8+8 = 44% chance of its being 2, 3, 4, 5, or 6, and so on. Now consider one guy of skill 4 vs another of skill 7. The comparative ladders look like,

10   10    (10% for the score (4) guy, 28% for the score (7) guy, since results are capped at 10)
 9    10    (5% for the score (4) guy)
 8    10    (6% for the score (4) guy)
 7    10    (7% for the score (4) guy)
 6     9     (8%)
 5     8     (9%)
(4)   (7)    (10%)
 3     6     (9%)
 2     5     (8%)
 1     4     (28% for the score (4) guy, since results are capped at 1, 7% for the score (7) guy)
 1     3     (6% for the score (7) guy)
 1     2     (5% for the score (7) guy)
 1     1     (10% for the score (7) guy)

So this illustrates the way add-subtract dice, with a comparative resolution system, works. It also illustrates the problem of "range of randomness vs range of abilites." Which is all a very long way of saying, do you want the dice to be very important.

For example, you could have ability ranges of 1-10, and use d20 with them; the dice are twice as big as the abilities in importance. This encorourages players to get a heap of low-level skills, instead of a few highly-focused ones, since skill 7 can easily be beaten by skill 4. Or you could have ability ranges of 1-20, and use d6 with them; the dice are one-third as big as the abilities in importance. This encourages players to get a few highly-focused skills, since skill 7 will only with difficulty be beaten by skill 4. 
    And then of course the question is of realism, whether you want that. For my part, as I said, I feel that as skills improve, the range of results narrows, too; the extremely good or bad results are still possible (like, I can beat Carl Lewis in a 100m sprint, in the 1 in 1,000 times he falls over along the way). So in a realistic system, you might have something like,

"Skills range from 1 to 20. Performance is adjusted by a d6-d6 roll. However, when you have skill level 6-10, you get to roll d6-6 twice, and take the result closest to 0 (negative or positive). When you have skill level 11-15, you get to roll d6-d6 three times, and take the closest-to-zero result. At skill 16-20, roll four times, and take the closest-to-zero result." 

This would give the effect of the randomness being one-third as important as the skill level, overall, but that as skill level improves, the randomness becomes less important, not only in absolute terms (d6-d6, -5 to +5, is relatively larger compared to 1-5 than to 16-20) but also in practical terms (rolling d6-d6 four times and taking the closest-to-zero result gives a much greater chance of a close-to-zero result than a single roll). 

Of course, such a system would be unwieldy and a pain in the arse to use, but it's just an illustration I made up just this moment:) Consider it the "average results for a character" version of the WoD's system. In that, you need to roll 6 or above (or whatever it is now) on a d10 for a "success", but as you increase in skill, you get to roll more d10s. So you have greater chances of scoring successes. In this system I just made up, you'd have greater chance of success, but also greater chance of a consistent result. 

In d4-d4, I wanted this effect, of consistent results, and also of the higher-skilled guy feeling he could quite reliably deal with the lower-skilled guy. So I had the Performance Ladder,

Famous
Olympic
Outstanding
Excellent
Good
Middling
Fair
Ordinary
Poor
Terrible
Crap

The normal human range of abilities is Poor to Outstanding. The d4-d4 roll adjusts their performance by -3 to +3, giving them results of Crap to Famous; results are capped at those. So the random die roll has a range 7, the abilities the same range of 7, and the actual performance a range of 11. (By comparison, what Lev is proposing is essentially a range of 100 in abilities, 100 in die rolls, and 4 in results: critical failure, failure, success, critical success). But (in d4-d4) the best an Ordinary guy can do is Good (uni graduate level), while the worst he can do is Crap (zero knowledge level). The uni graduate )Good) guy can at best manage an Olympic performance, and at worst will be Ordinary. He can never be Poor, or Crap, or Famous. Not without help, that is (ie die roll bonuses). The Ordinary guy will consistently be Ordinary, Poor or Fair, and the Good guy consistently Middling, Excellent or Good.
    So it's a bit like the system I just made up above, but less unwieldy. The system I made up above gives you the narrower range of results as the skill increases; d4-d4, you've got the same range of results and probabilities, but because the randomness (-3 to +3) is smaller than the range of performances (0 to 10) you get it that it's impossible (normally, without bonus/malus) for the Good guy to do Crap, or the Ordinary guy to do Famously. 
    It's not entirely realistic, but I think it comes close. It gives a higher than realistic chance of really bad or really good results in performance, but I like that because really good or bad results make for entertaining games. 

Sorry I write so much. I took a disadvantage, "loquacious", and used the points in "handsome." No, really!:)

Cheers,
Kyle
Better Mousetrap Games
home of d4-d4 and other stuff
http://www.rpgnow.com/default.php?manufacturers_id=339
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mimesisrpg.com/pipermail/design_mimesisrpg.com/attachments/20051017/24d02f57/attachment.htm>