[Design] Why random rolls?

[David Cake]

At 8:37 PM -0700 9/10/05, Lev Lafayette wrote:
>What function does the random roll have in skill
>resolution? What range should that roll be?
>
>Some games, like Amber (obvious example) are
>'diceless' which means that unless a character uses
>some director's stance technique (amusingly described
>as "bullying the GM") a character of skill 19 will
>*always* lose to a character of skill 20.
>
>Clearly this should not always be the case. But by the
>same token (and this is based on witnessing
>yesterday's game) I don't know whether it is necessary
>for a character to make a language skill roll on each
>piece of text that they read.
>
>Think also in terms of scale of the random effect.
>Take for example the Cyberpunk system where skill
>rolls are STAT (normal range 1-10) + SKILL (normal
>range 1-10) + RANDOM (1d10). Compare this to d20
>systems where the nominal range is 1-20 (stat plus
>skill) plus 1-20 for the skills roll. (i.e., 1/3 to
>1/2 effective input of the random roll).
>
>How much effect should 'randomness' has then
>resolution of skills?

	A simulationist conception of why you include randomness is 
that randomness is simulating all the dozens of factors that have an 
effect on the outcome, but are too fiddly/trivial in themselves to be 
accounted for (ie for hand to hand such a list of factors might 
include footing, exact design of all weapons, minor injuries, exact 
topography, etc), the relative effectiveness of the second to second 
variance in tactics used, and also the normal random variance of 
sampling a single trial -- but take a situation in which these minor 
differences are controlled for and its over a long enough period that 
the individual sampling factor is statistically minimised, and the 
results will be as predictable as Amber.
	Which is a simulationist explanation for a rule that says 
'roll dice where it is a quick confrontation, use less randomness the 
more sustained the effort is and the more controlled the conditions 
are'.
	The gamist and narrativist justifications are different, and 
accordingly have different rules. The gamist justification is that it 
makes the tactical choices much more interesting and less obvious and 
predicatable, the narrativist one is that adding an element of chance 
can be a useful creative technique (which is a weaker argument than 
the other two, hence the relative popularity of diceless techniques 
in narrativist games like Amber or Puppetland).

At 1:51 PM +1000 10/10/05, Eric McKibbin wrote:
>Personally I've always believed that 1-20 is far to great a random 
>range for a skill, and something like 1-10 would be better. However 
>in this case do you have a character always fail on a 1 and succeed 
>on a 10? Or should in some cases people never fail/succeed?

	A friend of mine has a cool variant of the HeroQuest rules 
where he uses d12s , and does 1-10 plus critical and fumble (or in HQ 
terms, bump up/bump down a level of mastery).

	Its worth remembering that there are other ways to do 
randomness besides dice as well. I loved the card based combat system 
in Lace and Steel.
	Cheers
		David