2 ms·

Voting algorithms are hard. If people game Digg and Reddit, you just know they're going to be all over you when there's money involved. We've got some pretty

Voting algorithms are hard. If people game Digg and Reddit, you just know they're going to be all over you when there's money involved. We've got some pretty good metrics that we're looking at. The obvious ones, like number of contributors, voting times, and so on. We also look at activity, and construct minimum spanning trees for various subsets of users as an automated "Someone should look at this" flag. Contribution times and rates are also something we look at. Probably the biggest single stopgap measure is decidedly low-tech: there's a short delay (a day or so) between project release and release of funds. If any real people at all are downloading the stuff they paid for and it doesn't work, hopefully they'll shoot an e-mail off to us and we'll take a look.

sanswork 18y ago
The last solution you mentioned was one I ended up implementing but I just wasn't happy enough with the rest of the solution to work out better fraud detection. One other suggestion for going forward might be to implement none software bounties as well. A large portion of my development was based on the idea of community bounties. Say Bountyville wants their park cleaned up so their kids can play they all chip into a bounty on your site. Receive proposals from companies then choose the best one to do the job. This would help you get more "normal" people involved in the site which might translate into these people discovering the software side of the site as well. The only difference you need for those are a two stage voting process where by users vote on proposals before the action and also on the result after. Anyhow, best of luck.