Tilt to Live 2: Redonkulous

[Kamikaze28]

(05-10-2013 03:02 AM)Necrocat219 Wrote:  Also, using a 3rd part host and not gamecentre for the highscore! It's somewhat sad to find a game that would be great to grind against and practice to get a good placing in top places, but then find the top 100 places occupied with the same scores.

I have to agree with your disappointment at fake scores. Call me naive but from my point of view, a little bit of statistics and some kind of bayesian filter could improve scoreboards tremendously. Most hacked scores I've seen have been so outrageously exaggerated that it is pretty obvious that they are faked. Training a machine to identify and handle such kind of data is a well researched topic. It's actually a lot simpler with scores than with email since scores only have one data point.

Here's my thinking: every score distribution follows an exponential curve to some degree. There are a lot of small scores and as the score increases, their frequency decreases. See the following figure:

In the beginning of a game (when there are no scores) you accept any score for a while until this curve establishes itself and then you can apply a simple classifier to all existing and future scores to decide whether they fall within the established distribution (i. e. are valid scores) or if they are severe outliers (i. e. hacked).

I do know that these methods can also be subverted but it would be better than what we've got right now which - as far as I know - is humans looking at scores and cleaning up manually.

/discuss