ratings and counting number of times played. Total number of plays is
probably one of the most common. As a measure of how "good" a game
is, it's not bad, but it doesn't take into account length of the game
or replayability over time. That is, a 15 minute game that's played 6
times is probably not "better" than a 3 hour game that's played 3
times. On the other hand, using total time playing has problems on
the low end: A 15 minute game played 10 times is probably "better"
than a two hour game played twice. Further, a game played every other
week for a year is probably "better" than one played every day for a
month and then never again.
Various other metrics have been suggested, including Joe Huber's
"happiness" metric which works well, but requires a lot of
bookkeeping. So, I've been trying to identify a metric of game
"quality" that meets the following criteria:
- Doesn't bias unduly for or against long or short games
- Doesn't require ratings
- Doesn't require overly detailed records
- Doesn't overly credit "fad" games that get played a lot for a
little while
and then never again
- Does bias towards "classics", though not overwhelmingly
- Measures "replayability" in some form
- Is easily understood and doesn't require deep understanding of
statistics
- Is easily calculated
- Can be readily used on a per year basis or on a "all-time" basis
So, after careful consideration and analysis, I've come up with a
metric that I like and thought I would share it and suggest it to all
:-) I've actually suggested this metric before, off-handedly but I
like it even more now that I've looked at the actual statistics for my
own games.
The metric is the "month metric", in which a game receives 1 point for
every unique month in which it is played. For example, I played
Medici in both February and March of this year, so for the year it
scores a "2", but it also was played in 5 different months over
2000-2001, so it's total score is a "7".
For my own games, only a dozen games have an all-time score above 10
and there's a nice distribution below that. For the year, naturally a
"12" is the highest possible score, though I had no game above 7, but
a solid group (9 games) at 6 or above. The other nice thing about
this metric is that I find it is (probably unsurprisingly) heavily
correlated with all of the other metrics discussed (# plays, total
time playing, "happiness").