Everyday tech products lead a secret double life as weapons for human rights activists. The same type of camera phone that captured your Nana’s 90th birthday also captures evidence of tortured bodies; shared, online spreadsheets used for T&E reports also collect names of missing dissidents; and the GPS that steered you to Disney also tracks natural disaster victims buried alive.
Behind one of the lastest suites of rebel-ware are the self-proclaimed “IT guys” at Benetech. They pioneer life-saving and democracy-promoting uses of technology, making them more affordable, accessible, and impactful. They specialize in cloud-based storage of corruption evidence and a James Bond-like device that erases evidence of spying, and they employ sophisticated statistical techniques that bring dictators to justice.
Unlike easily monitored Facebook and Twitter streams, Benetech’s private database software gives activists an invisible cloak, so they can amass evidence in secret. Once the evidence is collected, Benetech’s mathematicians go to work building a statistical legal case that often reveals, for example, that unusually high death rates among a certain group were the result of systemic murder.
When Jim Fruchterman began exploring the needs of organizations such as Amnesty International and Human Rights Watch, he discovered that information processing is “all human rights groups do”–they aggregate data and translate it into reports, court briefs, and raise awareness for crises.
They don’t need CIA-quality surveillance and sophisticated sharing platforms. But they do need reliable ways to digitize and safely store text. “There’s a group in Sri Lanka that had lost five years of files because it got eaten by termites.”
Enter Martus, a highly secure, cloud-based database for keeping track of incriminating evidence. The idea of cloud-based storage isn’t new, but with Martus, only the activist has the password and access to the online data, and the encryption’s so tight, it’d give an NSA hacker a headache (according to Benetech’s resident statistician, Patrick Ball). Martus is free, open-source, and translated into many of the languages connected with abusive leaders.
Privacy may not have been what Fruchterman had in mind when he started Benetech, but he quickly learned how its an essential element to everything he does. Internet transparency can be as beneficial to snooping dictators as it is to pro-democracy activists. Evgeny Morozov, for instance, recently warned cyber-utopians how convenient social media is for secret police.
In the past it would take you weeks, if not months, to identify how Iranian activists connect to each other. Now you actually know how they connect to each other by looking at their Facebook page. I mean, KGB (and not just KGB) used to torture in order to actually get this data. Now it’s all available online.
And speaking of torture, the idea of protecting cloud-based data with a password begs the question of whether that password could now be wrenched out of a detainee with the same torture. “Absolutely,” warns Fruchterman. But that’s where Benetech’s eraser tech comes into play. The program immediately wipes an activist’s computer of incriminating spying evidence, should authorities break down the door. The so-called “panic-button” was developed in partnership with the State Department as part of their new global citizenship focus, and Fruchterman is hopeful it will be available for cell phones sometime in the future.
Additionally, data is more likely the target of an attack than people. “It’s hugely embarrassing to repress human rights groups” says Ball. In the Congo, for instance, the head of police is being tried for killing a prominent activist. Save North Korea, China, and few brazenly authoritarian regimes, many corrupt leaders have to at least keep the pretense of legitimacy–which means nixing loud behavior.
An Equation for Sentencing Dictators
Patrick Ball has an uncontainable passion for eliminating sampling bias. Doing so saves lives and imprisons dictators, he says.
That’s sampling bias, as in survey groups that somehow differ from the larger population they purport to represent–like when conservative pundit Sean Hannity showed that 80% of respondents to his survey thought Sarah Palin should run for president or, when his liberal counterpart, Ed Schultz, had a poll revealing that 98% of respondents thought that Dick Cheney wanted the country to be attacked again for political gain. In these cases, the polls don’t represent American as a whole, but the show’s highly partisan audiences.
So, when Benetech was contracted to determine whether the Guatemalan government had committed human rights abuses, it had to overcome the sampling bias in local news media that, according to Benetech, overlooked the systemic murders of indigenous people. Government killing squads often are not as brazen as they are in Libya; sometimes, genocide happens silently.
“If you looked at the newspaper, you would have had no idea whatsoever that the genocide had occurred,” Ball says. The mathematical trick, then, is to compare the mortality rates of indigenous peoples to the general population and show how the ratio is unlikely to be different due to disease, poverty, or simple chance–leaving only murder.
Reading Ball’s technical report is an astonishing juxtaposition of set theory and probability interlaced crudely with words such as “killing” and “violation.” Every 0 or 1 in the expanded matrix could be a murder too horrible to imagine.
In Guatemala’s case, the numbers on right side of the equal sign proved beyond a reasonable doubt that the deaths were systematic.
These numbers are immune to tampering. Fruchterman says that governments can intimidate a handful of witnesses; they can’t suppress a thousand, or even come close enough to skew the results of thorough analysis.
Crises and Obsessions with Data
A misunderstanding of sampling bias can be catastrophic in national disasters as well. Ball worries that the obsession with mobile messaging and open source mapping data (like Ushahidi) can lead rescue teams in the wrong direction, like mosquitoes to a bug-zapper.
In Haiti, for instance, there was a “negative correlation” between building damage and mobile text message streams, meaning that many of hardest hit zones were the most silent.
“If you had a map of buildings in Port-au-Prince, you had a better statistical prediction of where damage was than if you used the text stream,” Ball says, having run analysis comparing satellite info, mobile messaging, and online mapping tools.
Benetech has been very clear that Ushahidi and SMS are “great for documenting specific requests for help.” But, using them as an aggregation strategy has the potential to be terribly misleading. “Because we get excited about the presence of data, well, we make bad decisions, or we could have made bad decisions,” Ball says.
While the technology may not be as buzz worthy as whatever Apple’s new gadget is, Benetech thinks the less-than-sexy technology has an important mission when repurposed or more smartly applied–or as he describes it, giving “David better Tools against Goliath.”