Wednesday, June 19, 2002

Great Games, Hot Games, and Replayability

Often, when I introduce new players to these games, I'm asked "Which
is your favorite?" or something equivalent. Usually, I don't have too
much trouble identifying a handful of "favorites", but it raises a
question of how to rank games. One obvious issue is how to compare
games of dramatically differing lengths. An outstanding 10 minute
game is going to be a very different experience than an outstanding 2
hour game, which is a different experience from playing the
outstanding 10 minute game 12 times in a row. What might make some
reasonable metrics?

Ratings. One way to do it would just be to rate games, say
from 1-10 or from 1 to 4 stars. The problem with this is it tends to
be fairly variable even for one individual. That is, sometimes a game
may be an 8 and other days it might be a 6. Further, it doesn't
really catch the difference between games of different lengths. You
could try to incorporate length in somehow, but that doesn't
distinguish between the short game that you want to play over and over
and the short game that you enjoy a great deal, but only once in a

Number of Plays. I've got my games
that get played a lot
page. This obviously works fairly well, but
has a natural bias towards shorter games. I would guess that most of
the truly great games (to me) will make this list someday, but for
some of the longer games, it will take a while. Further, a number of
great but not-quite-as-great games make this list because they are so

Total time playing. This seems like a pretty good measure, but
it has bias against the short games. If I make the decision to play a
game 15 times, that implies it is pretty good. Two plays of a two hour
game doesn't to me imply it is better than the 10-minute game I play
15 times.

Replayability. This is a metric which I define as "the number
of 1 month periods in which the game has been played". This means a
game which gets played 7 times all at once and is never played again
gets a "1". On the other hand, a game that gets 2 plays and then is
played again once a month for the next 5 months gets a "6". This
works a little better than number of plays, but also tends to favor
short games.

Happiness. Joe Huber suggested this formula: take a rating of
a game, subtract a baseline, and multiply it times the total time
spent playing a game. It avoids the bias against long games by
including length in the calculation, and mitigates the bias against
short games by introducing the rating. Overall, this is probably the
"ranking system" I'm most fond of. So, without further ado:

Top 20 Games of All Time. RoboRally, Battle Line, Lord of the
Rings, Settlers of Catan, Vinci, Traumfabrik, 6 nimmt!, Schnappchen
Jagd, Bluff, Euphrat & Tigris, Pente, Ra, Princes of Florence, Puerto
Rico, Can't Stop, Medici, Bohnanza, Industrial Waste, Ricochet Robot,

The list doesn't change all that frequently, but it certainly does a
little bit.

Of course, then there's the notion of "what's hot now", and a little
tweak to the happiness formula to incorporate "newness" and number of
recent plays. This year's "hot" games for me have been: Puerto Rico,
SpinBall, Industrial Waste, TransAmerica, Crokinole, Lord of the
Rings, Alles im Eimer, Pueblo, Haste Worte? and Hamster Rolle.