[MUD-Dev] Balance of Character Power

[Shawn Halpenny]

On Sun, 5 Oct 1997, Travis Casey wrote:

> >These are of course the system extremes, yet they show BALANCE even
> >at the extremes of min/maxing development.  This is I think a key
> >point in combat balance.  Give David a chance against Goliath no
> >matter what the odds.  I don't see David EVER defeating Goliath in a
> >vanilla DIKU mud nor do I see it happening in basic D&ngntD rpg.  I
> >remember calculating the odds for a 1st level vs. 20th level warriors
> >and it was around 1 in 15000 for the 1st level to win.  D&singD and
> >Diku being battles of hit point attrition rather than amazing one
> >time lucky shots.
> 
> Of course, this is an intentional design feature of D&D.  Whether the
> original DIKU mud designers also intended it, or just didn't think about
> it when they based their system on D&D, I don't know.
> 
> I thought of a system a while back which has the characteristic of always
> giving the lower-skilled character a chance.  Here's how it works:
> 
>   A character's chance of success is determined by the ratio between
>   his/her skill and the difficulty of the task, the formula being
> 
>     skill/(skill + difficulty) = chance of success
> 
>   (Note that both skill and difficulty are numbers >= 0)
> 
>   Thus, if Theresa has a score of 20 in lockpicking and is trying
>   to pick a difficulty 50 lock, she has a 20/(20 + 50) = 20/70 =
>   roughly 0.29 probability of success.  Hanse, with a score of
>   136 in lockpicking, has a 136/(136 + 50) = 136/186 = 0.73
>   probability of success.
> 
>   The difficulty may be dependant on another character's skill.  Thus,
>   if Hanse has a 75 in shortsword and Theresa has a 45 dodge, Hanse has
>   a 0.625 chance to hit Theresa.

This is very similar to the system I'm using.  So far (in theory) it has
worked acceptably well.

>   A character's skill score should be partially "innate" -- that is,
>   even an untrained character shouldn't have a 0 score in any skill.

Agreed.  I prefer to call this aspect "luck", since having practically no
skill in something doesn't mean you couldn't "miraculously" do it--the
chances of you repeating it a second time, however, are again slim.
Certainly, you learn from what you did, but you didn't go into it with any
real knowledge of the task.

>   Notes:
> 
>   1.  Unless a difficulty is 0, the character never has a 100% chance
>       of success.  On the other end of things, given the accuracy to
>       which most computers will calculate fractions, there will almost
>       never be a 0% chance of success.

Each character's luck comes into play here for me.

> 
>   2.  Builders and coders can easily get a feel for probabilities of
>       success.

This was a design requirement for me.  I wanted a system where one could
somewhat intuitively (or with minimal guidance) decide how difficult a task
(killing a character, picking a lock, etc.) should be _relative_ to tasks
of similar level.

>   3.  Absolute skill doesn't normally matter in competitions -- only
>       relative skill does.  Thus, a character with a 60 is twice as
>       good as a character with a 30, and a character with a 120 is
>       twice as good as someone with a 60 -- and a 120's chance of success
>       against a 60 is the same as the 60's chance of success against a
>       30.

This is particularly attractive, since works somewhat like a sliding scale.
There isn't really a min and max ability, only a better-than or worse-than
X comparison.  And it differs in every case.

> 
>   4.  Changes in skill at the high end mainly benefit the character when
>       fighting other high-skill characters.  For example, a move from
>       80 to 100 skill only increases your chance of success from 0.80 to
>       0.83 against a 20 skill, but increases it from 0.47 to 0.53 against
>       a 90 skill.

This was my largest motivation for using a system like this.  It also
inherently handles what happens when a high-skill character repeatedly does
low-skill actions:  he succeeds practically all the time, but succeeds so
easily that any skill enhancement is nonexistent.

[snip - balance and HnS]

> Something that does need to be avoided is the scenario where high-level
> characters can go into lower-level areas and be able to advance rapidly
> without significant risk to themselves.  A number of ways have been tried
> to avoid this -- in a Hack'n'Slash game, you might want to not award
> characters any experience for defeating monsters that are more than a
> few levels below them.  You may also need to reduce other rewards, such
> as monetary awards.  This might be possible through "realistic" effects --
> for example, if you can hit the orc hard enough to kill it in just three
> blows even with its armor, chances are that the orc's armor won't still be
> in saleable condition afterwards.

Agreed.  The above system for calculating success of an action helps in
this respect.  If you're that much more skilled than your opponent, then
you should have no trouble dispatching him and receive the correspondingly
low reward, simply because you haven't had the time during the fight to
learn anything new and enhance those skills.

--
Shawn Halpenny

"You only need two things in life:
    WD-40 to make things go and duct tape to make them stop."