Facebook should publish their list of banned words

I got banned from Facebook again, and this time there are no subtleties and nuances like last time, where regional-specific discourse was involved.

This time it’s something inexplicable without assuming there’s just a “bag of words” for which users are automatically banned after someone reports a comment (and the appeal does not go through a human moderator).

What’s the context: someone had published a racist joke about a building in Germany that caught fire, and immigrant families (from Muslim countries) died, while their German neighbors didn’t, because they were at work (meaning —” immigrants don’t work”).

So my comment was: “if German nazis were telling that joke, one of the victims would’ve been Bulgarian” (I’m Bulgarian and a Bulgarian had posted the “joke”), implying that racism affects us as well when we are immigrants and we should have this perspective when we throw stones at others.

The result is — the racist joke remains, and I’m banned for 3 days. Probably because I used the word “nazis”? There is no other sensible explanation, apart from a set of very dumb human moderators that can’t grasp context.

But if there is such a list of monitored words, we need to know them. In order to avoid them from complex sentences and nuanced thoughts. The algorithm apparently doesn’t understand context either.

In addition to my primary activity as a tech entrepreneur, I’m also a political activist — and my party’s opponents may use the censorship features of Facebook against me. Elections here are in a few months and arbitrary bans with increasing length (first one day, now three days) may affect my ability to express my political thoughts. Bans for comments that are not just non-racist — they are anti-racist.

Maybe this “overpolicing” will stop after the US elections. But Facebook needs to be more transparent on how it functions. If I favor any regulation of online companies, it’s transparency — how and why they take certain decisions (from bans and content moderation to content recommendations).

So, for a start — publish this list of bad words and let me go and cleanup any potential offenders where your dumb “algorithm” can miss the context before someone else does it and I end up de-platformed during an election campaign for no good reason apart from an ill-designed moderation process.