I’m thinking it could be case sensitivity in some part of the database:

Martoon is written with capital M, but the working link is martoon, in lowercase. This means that username and database name will not be equate as is, because SQL queries are case sensitive. So if it’s returning an error, and that in turn is being translated into a empty field, we would have an explanation.

Anyone else have that?

Okay, so everyone has currently at least 5 ratings and nobody’s getting ignored because of idlers. But it’s just weird how one would participate in a gamedev event and not be bothered to play a single game.

http://img.ie/a9164.png

The “least rated” list is definitely helping to reduce the overall spread.

1266 ratings have been made over the past 24 hours (Wed-Thu).
1291 ratings were made in the previous 24 hour period (Tue-Wed)
So ratings are coming in at pretty much the same frequency as they were yesterday.

As for the spiraling server costs, could the contest get a sponsor, or are the organizers comfortable paying $200/mo?

Seriosuly though: As great as more participants are, going through several hundered of games is becoming a slog.

And here’s another showing a Poisson curve for the data. If games were being selected with uniform probability, we’d expect the distribution of ratings to follow this curve. The difference clearly shows that games are not being selected with equal probability.

This could be down to:

- screenshots are not all equally seductive.
- game names are not all equally seductive.
- people choosing already popular games, rather than random games, because they want to play a good game.
- people preferring some platforms over others. (eg web games take less effort to run than standalones).
- people are interested in some developers more than others either because they are well known or have good reputation.

If I had more data to hand I could test these hypotheses and look for correlations.