Fighting Spam on Tagged

Spam is a problem that plagues most social websites, and Tagged is no exception. On the Security team, we think about spam all the time. For those who aren’t as familiar with the issue, I’d like to share details about our team’s motivation for building the solutions we have put in place.

Fighting spam is a constant battle between security agents and spammers. It’s an area of software development that is truly a virtuous cycle of advancement since both sides are actively trying to outsmart each other. These challenges have helped me grow professionally by forcing me to become accustomed with some of the leading-edge techniques being used in the industry right now. Machine learning in particular has taken the spotlight in the software industry as a means to present users with more relevant information. In the case of spam on Tagged, our system learns what content looks bad and then blocks or deletes it.

Spam detection and prevention is a perfect opportunity for investment at Tagged since the area directly ties back to providing a great user experience. Nobody should think that the Internet is a scary place where one wrong click can be a costly mistake; our goal is to create a safe environment for our users. We continue to have room for improvement in this area with blocking spam and removing fake profiles. At the core, Tagged should be a site for real people to come and connect with other real people.

The most important takeaway I’ve learned from working against spam has been that solutions do not need to be as daunting as the problem. It’s easy to get caught up in solving edge cases or trying to think too far ahead, and sometimes it doesn’t hurt to just try something and see how well it works. Some of the more successful techniques being used against spam are surprisingly intuitive and simple at their core. For example, rather than using some complicated heuristic for labeling what content should be called “spam,” we can just crowd-source the problem to our users by looking at the results of the spam reports they submit. Although there is a lot of noise in this data, the important cases do bubble up. These results can then be used to feed data into our automated systems.

The Security team is dedicated to protecting our users, and one of our core values at Tagged is “Users are #1.” With over 330 million members, we are always trying to stay ahead of the game so that anyone who visits Tagged can have a great experience.

Andrew Neilson is a Software Engineer on the Security team at Tagged with a passion for making ice cream and playing hockey. You can follow Andrew on Twitter.