A look at the Attack of the Meeplings Highscores


aotmscores1.gif

Here’s a histogram of the 4,842 scores in the Attack of the Meeplings database. At first glance the data looks realitively normal – However, it’s far from it. The horizontal axis is not linear, but exponential (2^n) the data is heavily right skewed which tells us that there are some people who are just so much better at the game than most.

This graph shows us how difficult the game is and we can see where peoples limits begin to emerge – For example, there is quite a bit of drop off in the 4000-8000 area.

The median score is 949 and there are only 295 scores that are greater than 16,000.

What does this say about developing a shooter if you’re an indie game developer? – Implement difficulty levels! The people who are in the top 100 are those who really like these types of games and are really good at them. Unfortunately if you only target those people you drastically reduce your potential audience.

Your players are supposed to be having fun, remember that. Give a challenge to those who want it, but relax on people who aren’t quite as good.

  1. #1 by Erik Hogan on May 18, 2008 - 7:06 pm

    I would have thought the exponentially greater scores enjoyed by expert players was because of the scaling score system (for example a player that reached 50,000 points may have only gone a level or two further then one with 5,000 points) and that if shooting the enemies produced the same number of points all of the way through the game then the high scores would resemble a normal distribution.

    Have you thought about adding final round and wave reached to the scores? I would be interested to see how far the top players are reaching in the game.

  2. #2 by admin on May 19, 2008 - 12:04 am

    There is a certain element to that but the scoring system only increases by 1 each time. The graph shown doubles each time, which grows quite large very quickly.

    I should probably have recorded number of kills to see what that would look like without the score scaling. I would still imagine it to be right skewed.

    Also adding the round and wave number is a good idea. I would retrofit it but I don’t think there will be too many changes to the top positions. For next time maybe.

(will not be published)