Shall. We. Play. A. Game? Predator Alert Tool as a game theoretic simulation of countermeasures to rape culture

In “Strategies Without Frontiers,” one of this week’s BSides LV information/security conference talks, software engineer and co-originator of the language-theoretic approach to computer security Meredith L. Patterson used Predator Alert Tool as an example of “an organic response against predatory [societal] games.” Or, in simpler words, Predator Alert Tool was cited as an example of how we can change our cultural environment from a relatively safe place for (sexual) predation into one that’s actively hostile to sexually predatory behaviors. And we can talk about that process using math, like this:

Normal form of the classic Prisoner's Dilemma game theory problem displays a matrix of outcomes for a given combination of player strategies ("cooperate" or "defect").
Normal form of the classic Prisoner’s Dilemma game theory problem displays a matrix of outcomes for a given combination of player strategies (“cooperate” or “defect”).

That’s why myself and a group of volunteer culture hackers have been blanketing the Internet’s social media websites with numerous different variations of Predator Alert Tool prototypes. We’re dissecting rape culture and using what we learn to devise game theoretic counter-strategies encoded as software tools that help people avoid undesirable outcomes.

That sounds complicated, but it has very humble origins: scale protective mechanisms that already work.

For the future, to use Meredith’s words:

Predicting your adversary’s behaviour is the holy grail of threat modeling. This talk will explore the problem of adversarial reasoning under uncertainty through the lens of game theory[. …] But as a tool for the real world, game theory seems to put the cart before the horse: how can you choose the proper strategy if you don’t necessarily even know what game you’re playing? For this, we turn to the relatively young field of probabilistic programming, which enables us to make powerful predictions about adversaries’ strategies and behaviour based on observed data.

In “the transparent society” of the public Internet, we can observe a lot of data. After all, the Internet is a record-keeping archive at the same time as it is a telecommunications medium. And this data reliably reveals patterns about who behaves in predatory ways:

People who try to break tools designed to support rape survivors are extremely likely not to support those who have had their consent violated. And it just so happens that identifying people who are likely to be unsupportive of those who have had their consent violated is what Predator Alert Tool is designed to do. Letting people attack PAT and then identifying who launched those attacks turns out to be an exceptionally reliable indicator—undeniable, even—that those attackers should be included in the database itself.

For the more mathematically minded, Predator Alert Tool can be approached as a reputation system coupled with a societally iterated prisoner’s dilemma. That is to say, it’s a tool designed to help you make dating choices that take into account all the past interactions a given person (like, say, the cutie you’re scoping out on OkCupid) has had. As one oft-targeted woman put it, “PEOPLE CAN SEE WHAT YOU TWEET AROUND HERE and some of us can’t afford to have short memories.”

And wouldn’t you know it? A lot of hackers are already working on this problem. Unfortunately for those of us who think rape is kind of a shitty thing to do, those highly-skilled and well-paid mathematicians and computer hackers are usually employed by secretive government agencies that are famous for sharing sexually explicit photos of attractive women intercepted from their private Internet communications. (Also, hey, thanks for letting us know about that, Edward Snowden.)

Sadly, we live in a world where sociopathic behavior isn’t just tolerated, it’s rewarded. Until that changes, I can guarantee you this: violence prevention needs more hackers.