[Classicmerp] GURPS Normal Distribution Curve vs Linear Approaches

[Michael Cole]

On 26/12/2013 3:13 PM, Lev Lafayette wrote:
> On Tue, December 24, 2013 9:48 pm, Michael Cole wrote:
>> OK.  Lets start with the target at about point blank range.  At a point
>> where Doofus the Recruit should be able to hit it at about 95% of the
>> time.  Lets say that Sharpie the Marksman now has 99% - hey, he could
>> miss.
> Doofus and Sharpie have a 4% difference in ability? They're not really
> that different then.
Now you are being intentionally thick.  They have a great difference in 
ability.  Its just at that close range, Doofus will have a 95% chance to 
hit.  Go back and try it again.
>> Note here that I am not arguing that the numbers all work out - sure, we
>> could fix the numbers.  But a linear system would be even more broken -
>> that is what I am arguing.  The key point is that as you get better,
>> simple stuff becomes extremely simple, difficult stuff becomes only
>> slightly less difficult.
> It is true that for an expert they can perform simple tasks without
> thinking about about ("unconscious competence" in learning theory), but
> that can be easily simulated by a linear system.
>
> For example, let's assume a "routine" task has a modifier of +6 (using the
> MERP/Rolemaster definition of "routine" on a d20 scale). If they have a
> good level of ability (say 15+), they can perform routine tasks in a fair
> environment 105% of the time, i.e., it's unconscious.
Actually, they can only perform it 100% of the time.  Basic mathematics.