Tuesday 29 June 2010

It's not dead, it's resting...


Larisa's musings on the recent resurgence of WOW is dying posts have got me thinking. As she rightly points out, the game has been "dying" for over three years, with no noticeable impact on its actual subscription numbers.


Let's take a look at the current figures from MMOdata. The important things to look at here are the actual data points, not the smoothed extrapolated curves (I'll get onto them in a moment). The real subscription numbers are pretty good. WOW's figures may have levelled off, but this is the kind of "death" that any other MMO maker would ... err ... kill for.


But what about those smoothed curves? Surely they prove that WOW will decline in future - they all turn down post 2010. Well, no. Seeing curves like that sets my boffin-sense a tinglin' *

(*Boffin-sense is like spider-sense, but with the ability to locate Kirsten Dunst replaced by the ability to spot bad use of statistics. It's not a good deal.)

Extrapolation from numerical data is a tricky business and using a crude tool such as as spreadsheet's trendline feature can give highly misleading results. Let's look at some example data to show what I mean. I've created some artificial WOW-like subscription numbers that show an underlying trend of continuous increase, with a random variation on top. As you can see from the graph below, using a crude spreadsheet extrapolation appears to show a future decline, even though the real trend is actually upwards.


Putting it simply, those curves that appear to show a decline are meaningless. WOW is bound to join the choir invisibule at some point, but the curves from MMOdata don't prove anything.

1 comment:

  1. Ohhhhhh so much math(s)!

    I have to say I read the OP and just had my usual 'ok' reaction. I'll wait for the servers to shut I guess, before I believe that Azeroth is going to end.

    ReplyDelete