Game On? Australian Researchers Think Spam Filters Can Learn from Game Theory

If you’re as much a fan of games as I am, you’re going to love this one. Spam, which is some of the most low-utility, extraneous, non-social contribution type of stuff in the world, may be combated in the future by another low utility, extraneous, non-social contribution type of stuff, video games. Or game theory, to be more accurate. Now don’t misperceive me. I’m being facetious when I compare video games to game theory; the two are entirely separate concepts, although video games have been a subject of game theory since their inception. And I do love video games, so I’m not slamming them by calling them low utility. They’re low utility in the sense that a common phrase heard in households around the world is “why do you spend so much time playing those things? Go outside and play!”

I’ve written several of my own games, some of them not even crappy. And I’ve made a career of the things, becoming both a bit of an expert on the business of video games and as a consultant. And I’m well aware of the many different ways video games have benefitted the modern world, from education and simulation to dealing with serious issues like obesity and depression. But game theory has been around longer than video games, as early as the 1930s, and if you’ve seen the movie A Beautiful Mind (mathematician John Nash, played by Russell Crowe in the film, was a well-known game theorist), you already have a layman’s understanding of what game theory is. And game theory’s contribution to society is undeniable. It’s used to create models for economics, geopolitics, psychology, logic, and biology, to name a few practical applications. But now we can add spam detection to the list, and that’s just, well, too cool!

According to the venerable folks at The Register, a couple of Australian scientists have postulated that game theory can be used to thwart spammers in the future. “The new spam classifier, developed by Professor Sanjay Chawla, Fei Wang and Wei Liu of the University of Sydney, outsmarts would-be spammers by predicting the likely pattern of future spam runs by learning from past attacks.” The two researchers, who work at the Capital Markets Cooperative Research Center (CMCRC), have built a model for spam filtering that “uses ideas from “repetitive game theory” to achieve better results in junk mail filtering than existing commercial spam filters.”

According to The Register, in a statement, Professor Chawla said that “applying game theory allows filters to stay ahead of spammers (SIC) tactics as well as offering other advantages that involve constantly updating junk mail filtering rules. Typical spam filters make more mistakes over time as the spammers work out how to get around the filter. An example of this is spammers using misspelt words in the title.” The CMCRC has published a synopsis of the research here, and if you feel really adventurous, you can read the paper here.

Now, it makes a lot of sense that game theory could be used for spam detection. Game theory uses math that’s more sorcery than math (at least it seems like sorcery to me, one who’s not adept at math) to create predictive modeling based on uncertainty. While that’s a slippery statement, it might be unslimed a bit to simply say that we can learn from our past mistakes, and that’s as layman as it gets.

Of course, like every scientific theory, it’s just a theory, and there are skeptics, and rightfully so. Virus Bulletin’s Martijn Grooten, points out to The Register that “the approach would probably only yield improvements for certain classes at spam. Grooten said he would like to see how the filter works in practice, rather than relying on marketing claims about the power of new approach compared to conventional multi-stage junk mail filters.”

“It seems to me that they understand existing spam filters to be static engines that get updated every now and again,” Grooten told The Register. “In fact, most are highly adaptive to both the mail they see (so they create some kind of pattern for that particular customer) and to emails received by external sensors such as spam traps and spam reports. So they get updated in real-time, usually without any human interaction.”

The Register points out that this isn’t the first time game theory has been suggested as a bedmate for spam detection. Back in 2005, some researchers at the Athens University of Economics came up with the same idea, but not much has been heard since.