Last.fm has been an amazing contributor to my music library. I've found that it does an excellent job of selecting other music I would like. When listening to the radio on one of my favorite artists, I use the skip feature maybe 1-2 times per hour and even those songs aren't really all that bad. Should you be unfamiliar with the system, it basically takes artists you feed it and matches you to other compatible artists(but not in a blind way, genre based way). I don't know how the nuts and bolts of the selection process works, but the results it turns out are impressive.

What I would love is a similar matching system for games. However, I am not asking for something like Gamepot's miserable "You might also like" panel. That thing is about as useful as nothing.

Games often take quite a while to get to a point where you know if you'll like them or not(and that assumes the demo is long enough and an accurate reflection of the game). I'd probably play more games if I had a better chance of liking what I start playing.

So, is there anything like this?

The difference is that music doesn't have platforms. There is 100% freedom to match your stuff up with anything that has ever been released.

When GameSpot recommends games, they have to pick from the (relatively) small library on your platform of choice. The same thing happens to me on Blu-Ray.com when I look for movies that are similar to the ones I like. There is only a limited data-set for it to work with.

let's just overlook a bunch of technical problems(such as platform-specefic games, games that work with your hardware, etc.) and talk about how you'd accomplish this.

first of all, games are vastly different from music, you have a broad array of genre's, and even then, many gamers are really to play games from many diffrent genre's(of course some gamers do make it a point not to play/enjoy certain genre's(such as sports game, or rts's, etc). but it's likely you woudn't be able to use genre's as an indicator for what a user likes.

so, i think the first thing i'd do is begin by allowing users to create tags for a game, just get tons and tons of tags per game, i'd sort these tags by how often they get used to tag a particular game.) from their, i'd begin by creating links between games that have the most common tags. then as a player progressively plays game, systematically modify how certain tags are weighted relative to the user, for example, if the user doesn't like many sports games, then those tags would get less weight for that user.

it would take a good bit to build the initial database, and tag cloud, and it'd take a good bit to learn what sort of things the user likes, but i do believe that over a long period of time, the system would consistently choose games the user would probably enjoy.

but at the end of the day, this would be quite the technical challenge imo.

The first thing that comes to mind that sounds similar is Raptr. Disclaimer: I haven't used Raptr, all I know is it's good for tracking your recently played games and finding what your friends are playing, and search through other games in their database. But I do not know if they have a recommendation feature on there.

first of all, games are vastly different from music, you have a broad array of genre's, and even then, many gamers are really to play games from many diffrent genre's(of course some gamers do make it a point not to play/enjoy certain genre's(such as sports game, or rts's, etc). but it's likely you woudn't be able to use genre's as an indicator for what a user likes.

Most people listen to music at least as diverse as their gaming interests. I don't know a single person that listens to exactly one type of music that's all virtually identical, but I sure know a lot of people that play nothing but

">spunkgargleweewee. But I think you can use genre. If I said "I want something like Starcraft 2" I would generally expect a suggestion of an RTS. It would be strange to say "Then maybe you'll like... Kingdoms of Amalur!".

The first thing that comes to mind that sounds similar is Raptr

I will give it a try. Thanks :)

Last.fm doesn't actually care about the genre of the music you play. All it does is look at bands that are listened to by each person and look for correlations. eg. Lots of people who listen to Metallica also listen to Iron Maiden, so it can correlate them highly and when a new user listens to one of those two bands, it can recommend the other one to them.

The benefit of such an approach is that it doesn't need to try and analyse the music in any way to make the recommendation, which means it can transcend genre squabbles and technicalities. The major downside is that it doesn't work well for music that few people have listened to, because there isn't enough correlation data to make meaningful recommendations. But to make a game version of this, all you'd need is a way to track which games people are playing and a fairly trivial bit of number-crunching

Obviously this type of matching could be easily done by Steam. In fact, it probably already is being done by Steam, driving their Recommendations tab.