[MUD-Dev] Re: Less numbers, more roleplaying.

coder at ibm.net coder at ibm.net
Fri Dec 12 15:05:34 CET 1997


On 11/12/97 at 12:38 PM, Matt Chatterley <root at mpc.dyn.ml.org> said: >On
Wed, 10 Dec 1997 coder at ibm.net wrote:
>> On 27/11/97 at 11:28 AM, Richard Woolcock <KaVir at dial.pipex.com> 
>> >Adam Wiggins wrote:
>> >> [Richard Woolcock:]

>> >> > So called 'Intelligent' mobs should go for the weakest opponent.
>> >> 
>> >> They should?  I consider myself intelligent, yet I always consider
>> >> the most dangerous opponent to be my first target in a combat
>> >> situation.
>> 
>> >Hmmmm I'm not sure now.  Certainly, you'd be more worried about the most
>> >dangerous opponent, but equally, which would you rather have?
>> 
>> Run some examples thru a simulator.  This is actually a non-trivial
>> simultaneous equation.  Some variables under consideration for each
>> protagonist are:

>This is interesting (I seem to be saying that a lot!). Basically the
>definition of 'dangerous' is not as trivial as it sounds by this
>reasoning (which makes some sense).

Quite.  Even more annoying is that the solutions change every time the
curves change.  Given a system with percentage chances on blow, and other
semi-random side-effects, the model needs to be re-evaluated every
iteration for the optimal pattern.

>This could probably be tackled by integration if you have a defined
>formula, and wanted a rough assessment of the total 'danger' a target
>offered. I intend to approach this by simplification, and by making NPCs
>attack a random target. If this begins to present problems, I will weigh
>the random variables by the *size* of targets - they'll swing at larger
>things more often.

Once I get a settled and semi-stable combat design and set of curves, I'll
probably setup a basic simulator and run a few hundred million variations
thru it on spare CPU cycles while I'm at work, tabulate the results, and
slap that in a mini-DB in the game for NPC intelligences to access as
"inherited knowledge".

It could make for some truly nasty antagonists.  I don't even need to get
real optimal solutions -- just a "near enough" along with correct
selection of weapon/blow etc should be enough to translate minor fidos
into feared killing machines.  

--
J C Lawrence                               Internet: claw at null.net
----------(*)                              Internet: coder at ibm.net
...Honourary Member of Clan McFud -- Teamer's Avenging Monolith...




More information about the mud-dev-archive mailing list