[MUD-Dev] Metrics, Data, and Statistics

[Jon Morrow]

In the recent past, there has been significant discussion over
developing a set of MUD metrics.  I presume that data would then be
gathered based on the metrics and published.  Then the data from
various sources could be compared, perhaps creating a need for metric
revision and new data, and, from the comparison, we could gather
statistics.  I believe this kind of cooperation and information is an
*essential* step for our industry.

I'd love to see a group of us (Matt? Raph?) officially start a project
for creating metrics, gathering data, and creating statistics.  The
discussion of information sharing has shown me that there *is*
interest, and I'm hoping I'm not the only one with growing initiative.
So, in the spirit of contributing something useful to get the ball
rolling, here is the information and questions I've gathered from
recent hunts through the archives:

  1) Percentage of Total User Base Online

    This is the freshest one.  It's also very important from a
    business perspective, if not for everyone.  Here are some quotes
    from the recent discussion:

      "...the rule of thumb that says that the players online
      represent 15% of your total player base (this estimation given
      by the $10/month people)." -- Matt Mihaly, 5/21/01

      "I think the problem here is the view of 15% as a fixed
      percentage.  It surely must change during the lifetime of the
      game.  When a game is brand new, it's brand new to all of the
      players, and thus very likely to hit numbers much higher than
      15%." --shren at io.com, 5/22/01

      "It does in fact change, from a max of near 100% at lancuh, to
      around 40% after three months, and then over time trending down
      to 15-20%. Now, my understanding is that many games over the
      course of many years have shown that the 20% figure is a good
      rule of thumb--but I admit that's received wisdom from folks who
      have been around the commercial side of this stuff longer than
      I." --Raph Koster, 5/22/01

    I think we have a metric, a little data, and an old statistic on
    this issue.  Here's a summary:

      Metric: Percentage of Active User Base Online at a Time 
      Average Number of Users Online / Size of Active User Base * 100

      Data: GemStone 3 Percentage of User Base Online at a Time 
      Approximately 3.33% (possibly based on inadequate data)

    Old Statistic: An established game has 20% of its active user base
    online at a time.

    It's good to redo old statistics periodically.  I would like to
    see this one challenged with some new data.  The current thread on
    business models will hopefully give us this new data.

  2) Volunteer to User Ratio

    From what I've seen, most games fill this ratio based on need and
    availability.  While that may work for a noncommercial game, it's
    important to know the number of volunteers and paid administrators
    for budget planning purposes.

    Creating a metric for this ratio is also difficult.  A PK-oriented
    environment probably has a much different ratio than a
    socially-oriented environment.  Some of you have also reported
    that users from different countries require different
    administrative strategies and amounts of attention.  I'm sure
    there are other variables, such as the time of day, that affect
    the ratio as well.

    Here is the only statistic that I found in the archives:

      "...one GM or admin per 20 players, though it is preferable to
      have smaller ratios." --Zak Jarvis, 05/25/00

    I think the first step toward creating a reliable statistic here
    is defining who a volunteer is, who a paid administrator is, and
    what their responsibilities are.  The definitions are usually
    different from game to game.  Some games have elected government
    officials with law enforcement power, but they are still very
    close to my definition of a volunteer.

    After that, the next step is determining which variables affect
    the ratio.  Then we can start gathering data.  Looks like a
    complicated and tedious process from where we are currently
    standing.

  3) Average Bandwidth Consumed by Each User

    Metric: Bandwidth per User = Average Bandwidth Used per Minute /
    Average Number of Online Users

    Data: (from the archives)

      "Achaea, we've discovered, is actually transmitting an average
      (very rough average based on probably not enough samples) of
      about 3.3k/player/minute, which seems high to me." --Matt
      Mihaly, 02/04/01

      "3.3k/player/minute is about average." --William Katzell,
      02/05/01

      "Approximately: 199486637 byes.  We have been up for 8 hours and
      27 minues.  508 minutes.  Which means...: 392690 byes per
      minute.  400k per minute...  Hmmm.  This is with roughly 140
      people on.  So 2.8k per player per minute." --David Bennett,
      02/05/01

      "POL (a uo emulator) uses generally less than 2k/player/minute.
      I'd expect the other UO emulators (Sphere, UOX, etc.) to be
      about the same." --Jeff Freeman, 02/05/01

    Statistic: Approximately 3k/player/minute.  Should probably have
    more data to make this one official.

  4) Processor Time and RAM per User

I'm guessing these statistics needs to be calculated for each
server/core.  It would be nice if someone could do a study and publish
the results.  Comparing the results might eliminate a small portion of
the confusion over choosing a server/core.

Regards,

Jon

_______________________________________________
MUD-Dev mailing list
MUD-Dev at kanga.nu
https://www.kanga.nu/lists/listinfo/mud-dev