Game failure rate by O/S: 2008 statistics

Here’s our statistics for 2008. Sample size roughly 19,000 downloads. The figure given is the ratio of failures-to-run (caused almost always by no OpenGL drivers) vs. successful installations. All logged installations are unique so the figures aren’t being skewed.

[tr][td]O/S[/td][td]Failure rate %[/td][/tr]
[tr][td]Windows 95[/td][td]0[/td][/tr]
[tr][td]Windows 2003[/td][td]2.7[/td][/tr]
[tr][td]Mac OS X[/td][td]3.7[/td][/tr]
[tr][td]Linux[/td][td]7.7[/td][/tr]
[tr][td]Windows Vista[/td][td]9.3[/td][/tr]
[tr][td]Windows XP[/td][td]13.2[/td][/tr]
[tr][td]Windows 2000[/td][td]16.7[/td][/tr]
[tr][td]Windows 98[/td][td]28.6[/td][/tr]
[tr][td]Windows NT[/td][td]50[/td][/tr]

So what’s new…

  1. Yes, that actually was a Windows 95 installation, and it worked. Just one though, accounting for its surprising 100% success rate ::slight_smile:
  2. Linux sample size disappointingly small: < 300 downloads. I really don’t know why we bother!
  3. Vista is nowhere near as bad as it used to be, and considerably better than XP was at this stage in its lifecycle.
  4. XP has improved dramatically; the prevailing rate was about 35% failure fairly steadily for several years and now we’re seeing that dropped to just 13%.
  5. We don’t actually support Windows 98 but it’s nice to see some of them still run against all the odds. Albeit only 7 attempts :-*
  6. These figures are subject to the oddities of my logging system and are as such representative of how that works with its odd quirks. That is, it won’t work behind a firewall, it treats any game play as a success and any installation with 0 plays logged as a failure.

Cas :slight_smile:

Awesome, thanks for posting the stats.
What’s the total failure rate %?
Does the failure rate include people who downloaded, but never started the installer? (I’ve heard this happens occasionally - what’s up with that?)

No way of knowing if they didn’t bother installing after downloading tbh… or if the VM failed to start at all. A log of failure or success means the VM worked, the game ran, maybe popped up a “You need to download new video drivers” download, then successfully sent me a log via RMI and cleanly exited.

Overall failure rate would probably be a bit of a misleading stat as the division between Windows and Mac OS is so huge and clearly defined. There were over 2:1 Mac installations compared to all Windows ones combined.

Cas :slight_smile:

Wow, that’s a lot of Mac installs. That’s great.

And here I was worrying that the fantastically small failure rate on OS X might have been due to low volume! In any case, it makes sense because OS X has most of what you need to run Java games built in and set up without any intervention, so as long as the game is set up correctly it’s almost certain to run (not too much local state that people can screw up!).

Cas, any idea how this odd ratio came to be? Are you linked to heavily from official or unofficial Mac software sites or something like that? Or are Mac users more likely to download Java games than Windows users because of the lack of games in general on the platform?

Might be skewed by our presence on Apple.com, but mostly I think it’s because Mac users either like our style and/or are starved for games.

Cas :slight_smile:

I think Mac gamers also come from a long line of shareware game downloads, having grown up with stuff like http://www.ambrosiasw.com/ and http://www.spidweb.com/ (at least I did). Windows users have always had their mainstream games to play, whereas I always had to find my own little gems. I think Cas’s games fit really well with the overall tradition and style of Mac gaming, which is rapidly dissolving and being replaced by traditional PC gaming, in my opinion.