Digg caught gaming its own system, claims it was just a test
Digg may have had some trouble getting users to like its latest redesign, but a report claims they may have been up to sneakier things in the mean time. One Digg user has taken the time to do some data mining using Digg's API, and he found a whole bunch of dummy accounts that helped get stories onto the fabled Digg front page with almost no activity from real users.The 159 dummy accounts all have obviously-fake names (such as the "dd1" pictured), and they curiously seem to have only contributed to submissions from Digg's publishing partners. The suspicious activity began after an algorithm revision that took place on October 15.
The submissions that the dummy accounts helped get to the front page do not represent a major percentage of the stories that made it to the front page, but the fact that they were all links to Digg's commercial partners makes this all the more suspicious. If Digg isn't behind these accounts, then it surely chose to ignore them, since it's unlikely that their activity would go unnoticed.
The methods used to collect this data and the data itself is available from Digg user LtGenPanda on his blog.
Digg has issued a response to these allegations, stating that Digg employees were indeed behind the dummy accounts. Digg claims that these accounts weren't built to game the system, but to test for vulnerabilities that may lead to influencing how stories appear on the front page. Apparently, these accounts created a big learning opportunity for Digg staff, and they've prompted some changes to the code over the past few weeks, with more tweaks to follow. Digg founder Kevin Rose himself took the time to stress that Digg has been doing such tests "since day one" and that Digg never received any money from any publishing partner for anything other than standard ad units.












Comments
4
Subscribe to commentsikeasangriaOct 26th 2010 8:04PM
I call shenanigans. Kevin Rose has been eager to get popular stories and more users to digg once it was revealed that reddit had higher traffic numbers.
Zaraki KenpachiOct 28th 2010 6:54PM
lol, it's definitely strange that it was tested on production and the guy who uncovered it definitely has some legitimate questions. I'm not the biggest fan of user vote-up as the only way to determine value. Check out YourVersion.com as a way to get personalized news without dealing with that same gaming the system issue.
lassiOct 27th 2010 4:32AM
"partner for anything other than standard ad units."
"I didn't take bribes, just made certain stories get more hits in exchange for ad value on them"
JoshOct 27th 2010 7:50AM
Makes me continuously happy that I gave up on Digg when it was censoring its users over the whole Reddit thing after the v4 launch.