God save us from meddling politicians!

More than two billion people, approximately one third of the world population, use Facebook, and each spends an average of fifty minutes a dayconsuming and creating all kinds of content, meaning what they do has to be monitored to avoid certain behaviors. It is probably impossible for anyone not involved directly with such a task to imagine the amount of resources and difficulties related to that task.

An interview with company security director Alex Stamos reveals that Facebook eliminates an average of one million accounts every day for violating its usage policies, spam, fraud or scams, and also for hate speech, incitement to violence, threats, harassment, etc.

Eliminating one million accounts every day means monitoring many more than that, and using a combination of algorithmic and manual tools that that is unprecedented technological challenge. The company is planning to enroll thousands of people to monitor content, giving an idea of the magnitude of the task, bearing in mind that manual supervision is only part of the solution. In fact, the main problem is not to establish policies or security norms, but to design the technical procedures. Imagine what it means to monitor a network where, every day, a million accounts are disabled for all sorts of reasons: obviously, mistakes are made: both type 1, false positives, or type 2, false negatives, which the company tries to avoid. Each time the company closes an account that was simply talking about killing a mosquito on the grounds of incitement to violence, for example, it risks being ridiculed or accused of unjustified censorship, and so has to invest valuable hours in manual oversight and reviewing individual cases.

At the same time as Facebook is trying to solve these kinds of issues, some politicians in Germany, in their ignorance, have decided that as the social networks can be used to spread hate speech, something must be done quickly. And of course, their solutions are simply political solutions: making it a criminal offense and imposing fines of up to €50,000 euros on social networks that do not eliminate such accounts immediately. The measure is clear proof of the ineptitude and ignorance of some politicians: because the social networks, which are being used on an unprecedented scale, are seeing isolated episodes of hate speech, the solution is to impose fines. And that will solve the problem.

What is the impact of such stupid laws, passed by technological idiots, incapable of imagining what the task they are demanding entails? Along with a major PR problem, Facebook now faces criminal responsibility and huge fines. As a consequence, the company is forced to rapidly scale up its efforts to supervise accounts, closing more of them, and thus increasingly the likelihood of error. To fix a problem that was being solved but simply needed more time to develop better systems and to devise other protocols, another major problem arises: accounts are being closed incorrectly for reasons such as satire, humor, irony, or simply for using certain language.

In fact, hate messages on social networks are increasingly being dealt with through collective reporting systems, leading to offensive accounts being closed quickly. There was absolutely no need for the politicians to get in on the act. As a result, the social networks are likely to apply more censorship and make more mistakes. A great achievement: thank you, politicians, for saving us… from ourselves?

(En espa?ol, aquí)