I like making charts, it’s a… thing. So I took some time to see how coolness is spread out over people, and how many ratings each person received. I *might* make updated versions on the one week mark and when it’s all over. Also I *might* make something regarding ratings received/coolness, but for now I got my fix. These values are from sometime today.

## Coolness and nr of ratings charts

Posted by Andrew (twitter: @secret_tomato)

August 28th, 2011 11:50 am
It would be interesting to plot in log-log and semi-log y scale! Thanks!

Done 😀

cool! Looks like there is some sort of power law there – which keeps popping up in similar settings. Basically it seems that the more popular games are more likely to get MORE votes…

And also, you can see it is capped from below – probably due to the “least ratings” sort which was just added to the LD system. cool!

So what you’re saying is that the games with most votes are most likely to win? But they still might average low ratings.

Can you expand a bit on your idea? I’m afraid that my skills in graph analysis are a bit lacking 😀

Because a lot of people will try popular things first, something somebody other already said is good. Notch’s game has the most rating by far. Randomizing your own list for rate and slowly adding new ones is good, but since there were already enough games I couldn’t rate (mac, ipad, android things or something which just crashes on my computer), I had to use “show all entries”. It’s still randomized, but if you go slowly over the list and then hit five bad / annoying / frustratingly hard entries, you’ll start to hope for something better… and higher number of entries usually means this might be good.

It means that people are not selecting games with equal probability. (If they were, then we would be getting a Poisson distribution.)

So some games have a higher probability of being reviewed than others, and we may expect those games to continue to be selected with higher probability.

@digital_sorceress: actually, if people would choose games with equal probability, we would get a uniform distribution (which is what ideally we want). Poisson Distributions arise under different circumstances.

I’m talking about games per number of ratings, not number of ratings per game, because that’s what relevant to the graphs which Andrew has provided.

Even with equal probability, some games will get more hits than others ~ that’s just how the dice fall. The Poisson distribution describes that natural variety.

You’re right, I got it the other way around…