I pythoned a bit to create a full spreadsheet of LD26 games, with percentiles. :)


Compo (CSV) – 1357 Rated Games
Jam (CSV) – 566 Rated Games

* Note that all the coolness data was taken directly from the site, and not processed.

But tables alone aren’t that interesting, that’s what graphs are for!
Here you can see how each category correlates with the overall (data taken from compo only):



It’s very interesting to see where’s the most (and the least) correlation:

  1. Fun (0.89)
  2. Mood (0.73)
  3. Graphics (0.68)
  4. Innovation (0.64)
  5. Theme (0.51)
  6. Audio (0.50)
  7. Humor (0.14)


P.S. I might post some bell curves later.

11 Responses to “LD26 Statistics – or – Make your games fun, not funny!”

  1. HackThePanda says:

    That’s good work man 😀 It seems people who can make a game funny are less good with making it fun at the same time, however I would have thought the opposite. Maybe humorous people have a hard time taking a serious approach to the actual game design?

    • ashdnazg says:

      Not exactly.
      If the overall is correlated with fun and quite uncorrelated with funny, it means that fun and funny are quite unrelated.

      The “center of mass” of all the dots in the upper part of the humor graph (i.e. funny games), is only slightly better overall than the one of all the dots in the lower part (i.e. not funny games).

      That means that people who can make their game funny are maybe just a wee bit better than average in making it fun.

  2. sorceress says:

    Interesting. I did this same thing for LD21, and it gave very similar figures.

    I also looked at correlations between all pairs of variables, which were surprisingly all positive. ie, scoring well in one category meant you were more likely to score well in all categories. If you want to draw more graphs for us, it’d be interesting to see if that’s still the case, . :)

    It’s counter-intuitive really, as you’d expect that spending more time on one feature means spending less time on another.

    So some suggested the positive correlations were a halo-effect of the judge, whereby if they liked a game there’s a tendency to score all categories high, instead of making a fair comparison between games.

    • ashdnazg says:

      Definitely halo-effect, though I used the term “Cross Category Diffusion”. 😛

      For instance:

      PONK, which is a pong game got to #70 on humor.
      Don’t get me wrong, the game is wonderful, but it definitely isn’t funny…

      I’m also sure you’re going to find soundless games with a score in audio (heck, even mine sort of fits this definition).

      • Codexus says:

        I think this is a special case and not really halo effect. It’s not so much the finished product that is being rated as the choice of platform and popularity of the author: “Sos tries to make a C64 game, hilarity ensues.”

        • sorceress says:

          I don’t think it’s fair to be picking Sos out. We should try to talk in more general terms.

          re: popularity of the author
          Halo effects do include the scenario your describe: Our general opinion of a person’s character will influence our perception of their work. If an unknown person made a pong game for ludum dare I doubt if it would score over 2.0 in any category.

          re: choice of platform
          It’s my understanding that choice of platform shouldn’t affect ratings. If we give bonus points to those who take hard/obscure routes (eg, c64 / assembly), then shouldn’t we also take penalty points from those who take easy/popular routes? (eg, unity / content generators)

          • ashdnazg says:

            I agree. Note that the Innovation score is probably where choice of platform should be rewarded (from the rules):
            Innovation – The unexpected. Things in a unique combination, or something so different it’s notable.

            The question should be whether we judge the game’s graphics (or anything else) relative to its platform or to all games.
            My game’s graphics consists of ASCII due to the platform’s limitation.
            Should it be judged versus ASCII art (where it’ll get a somewhat higher score) or versus the wonderful 2d and 3d art in the compo (and it’ll get a way lower score)?

          • Codexus says:

            I didn’t mean to pick at Sos. That was a really nice project.

            But I think the unusual circumstances explain the unusual humor rating in this case. There are several comments on his game’s page that indicate that the voter was watching the twitch stream of the game being made.

  3. klianc09 says:

    It’s interesting how audio and humor are the only categories that aren’t symmetrically. Having bad audio doesn’t seem to stop you from getting a good overall rating, but if you have good audio, you are more likely to score good at overall.
    Seems fairly logically, now that I think of it.

  4. goerp says:

    Pfah! Fun is overrated anyway!
    I scored less than 2 at fun this time, I’m on a roll! (downhill that is).

    Nice graphs by the way

Leave a Reply

You must be logged in to post a comment.

[cache: storing page]