Fighting Spam at Flickr

Flickr has always taken the spam problem seriously. Ignoring spam is unacceptable for any social site, as it destroys credibility, ruins the user experience, and stifles communities. As a site grows it becomes increasingly lucrative to spammers, and they continue to get incrementally more creative and resourceful in abusing it. Sadly, they have caught us with our pants down on several occasions. We’ve learned a lot over the years, and would like to share our experience in the hope that it helps others head off spammers more quickly.

We will talk about:

1. the evolution of spammers and the captcha arms race that has resulted
2. being prepared for a massive spam attack
3. past anti-spam efforts at Flickr and lessons learned
4. the economic incentives behind spam and how that can inform spam fighting
5. a new approach to fighting spammers, derived from #3 and #4

We’ll discuss how technology can make anti-spam efforts more efficient using clustering, machine learning, and input from humans, both endusers and staff.