[Ar-list] Human ranges in AR

[Carter T. Butts]

Matthew Selznick wrote:
> The range of attributes, as I understand the rules, is -1000 to 1000 for
> a normal human. 

Well, actually, +/- 1000 is really just the range which corresponds to 
1% and 99% probabilities under the DRF.  Most human attributes would be 
well inside of that range, although in principle one could have 
arbitrarily high or low attributes.

> What I'd like to do, for the attributes where this
> makes sense, is be able to take the known maximum in the real world for
> a given attribute (for example, strength, or lifting, for which the
> current world record is about 800 lbs) and tack that on to the range.

You _can_ do this, but the way to think about it is in probabilistic, 
rather than deterministic, terms.  Consider the lifting example.  If you 
took the average person and asked them to lift 800lbs, the chance that 
they would be able to do so on any given trial is _very_ small.  Wild 
things do happen -- the proverbial car-lifting mothers, for instance, 
though those stories may be apocryphal -- and the chance on any given 
attempt is not zero, but it isn't large.  If you took this probability 
(whatever it is) and put it into the IDRF, -1 times the result would be 
the difficulty of lifting an 800lb object.

Now, if you know that difficulty, you could use it to get the strength 
of the world's strongest person.  Most likely, even the person who set 
that weight record couldn't lift 800 pounds with 100% certainty...they 
would have some success rate, albeit one which is higher than the 
typical person (who has a STR of 0).  If you took this probability, put 
it into the IDRF, and added the difficulty of the lifting task, you'd 
have an estimate for the STR rating of the world's strongest person.

The key here is that the difference between the world's strongest person 
and the average person on the street lies in the differing 
_probabilities_ with which they can accomplish particular tasks...not a 
deterministic ability to accomplish a given task or not.  At the 
exteremes, these quantitative differences are almost qualitative; 
nevertheless, they are not strict dichotomous differences (at least, in 
AR!).

> 
> My reason for doing this:  in the game world I'm creating, there are
> individuals who far exceed the human norm in certain attributes.  A set
> of standards, as much as possible, for the normal human range of
> attribute ratings would help conceptualize this.

That sounds like a very reasonable thing to do...again, the trick is 
going to be to find ways to set the scale using relative ability, rather 
than tasks which one can/cannot perform.  OTOH, if you're looking for a 
quick and dirty approximation, you could always pick a particular 
threshold of success probability (e.g., at least a 5% chance per 
attempt) and use that.  It might or might not be easier to get the 
numbers, but that might be easier to conceptualize....

Hope that helps,

-Carter