[MUD-Dev] Continuous versus Discrete Functions

Dave Rickey daver at mythicentertainment.com
Fri Dec 21 10:21:24 New Zealand Daylight Time 2001

-----Original Message-----
From: Daniel.Harman at barclayscapital.com <Daniel.Harman at barclayscapital.com>
> From: Dave Rickey [mailto:daver at mythicentertainment.com]

>>   1) Players seem to hate discontinuous feedback from the system.
>>   To them, something either works, or it does not.  This tends to
>>   reflect itself as extreme criticism of any portion of the game
>>   system that provides a spectrum of outputs.

> Can you give some examples I can't think what makes you say this? 
> I can perhaps see people being upset if a trade skill can produce
> a wep which is inferior and then has no value to vendors or
> players.

AE Spells in Camelot used to fall off in effectiveness from the
center.  Even when the falloff was as low as a 25% reduction at the
edge, players felt they were "useless" because the effect couldn't
be predicted.  We eventually redid them to have two radii, the first
at full effect and the second at 50-75% (depending on spell), and
you were either affected or you weren't.  The base melee algorithm
had to be redone to apply a lower variability because the only thing
the players hated more than missing was hitting for 1 point of

>>   2) Linear functions require less CPU.  A discrete function can
>>   use Boolean or integer math, a continuous function almost
>>   demands the use of floating point math.  When handling hundreds
>>   of players and thousands of NPC's on a single box, fast is
>>   good.

> Not strictly true, FPUs are so fast now, they can beat integer
> calcs in a lot of situations. Also you could always used fixed
> point hacks to get around this (as everyone used to have to do
> before FP got so good). Anyway, I imagine the way to implement
> this stuff is through linear interpolation on a line graph drawn
> by a designer rather than working out polynomial curves.  Or the
> alternative where you just ratio attack skills vs defense skills
> etc (This was mentioned 3 months ago on this list as a nice
> approach).

We do a lot of fixed point math, in fact the rounding errors from it
turn up in the damndest places.  And in fact, our base algorithm
uses exactly the suggested system when comparing the weapon against
the armor to determine damage.

Most players seem to dislike ambiguous systems, which is how they
view discontinuous results.


MUD-Dev mailing list
MUD-Dev at kanga.nu

More information about the MUD-Dev mailing list