[MUD-Dev] Game Balance: Statistical Analysis in MPORPGs

Lovecraft dave at darkages.com
Thu Dec 16 15:42:13 New Zealand Daylight Time 1999

MUD-Dev members,

Could a list member direct this designer to research performed or case
studies of applied statistical analysis to massive multiplayer online
roleplaying games?  Books, links, the more valid the study, the happier I'm
to review it.

I'm familiar with statistical analysis and understand basic methods to
analyze game balance of classes in a D&D style multiplayer game.  I created
massive character analysis software and study the results thereof.  But
before executing my balancing plan, I want to research any alternative
analysis methods in the happy case that the industry has passed--and
documented passed--square one of this analysis.

Wishing to avoid re-inventing a better bean counter,

Dave Kennerly
Dark Ages Director

Addendum for those interested in details:

1. The most important average numbers are compared for the population.  The
massive multiplayer RPG population is large enough to provide confident data
even after divided into 50 categories.

For example, when determining empirical play-balance for classes and levels
of a class:
    level / time online
This is done for each class at regular intervals of levels.

Variations of this premise are applied in an easy to read format, such as
tables and graphs.

Advanced techniques are possible which make #2 feasible; most importantly:
    experience gained [in x days] / time online spent [in x days]

2. The more interesting analysis comes from hypothesis testing of a change
to the game mechanics.  The details of hypothesis testing are better
answered by a text book.  Yet here is a fair idea of the hypotheses:

H(0): This game mechanical change has no effect on rate of advancement.
H(1): This game mechanical change has an effect on rate of advancement.

Standard deviation and normals are known (from #1), and therefore the
hypothesis is accepted or rejected.

This is freshman-quality statistical analysis.  I'm interested in previous
statistical work done; wheels already invented.

Wishing not to be the first along this path in a genre that is over 20 years


MUD-Dev maillist  -  MUD-Dev at kanga.nu

More information about the MUD-Dev mailing list