4 ms·

Their principles are fine. It's just that hard-coding a list of English-language bad words, then matching on them in an extremely naive way, is not an excellent

Their principles are fine. It's just that hard-coding a list of English-language bad words, then matching on them in an extremely naive way, is not an excellent way of manifesting those principles in code. The Scunthorpe Problem[1] has been known for over 25 years. Plenty of people have names that incorporate a word that, on its own, would be a slur. Many English-language slurs are also words with entirely different meanings in other languages. Naively regex-matching all incoming text to see if it contains any of a hard-coded list of words means that you're going to get a lot of false positives, which the instance administrators can do nothing to prevent. Meanwhile, you are requiring only the level of creativity of an unremarkable eight-year-old on behalf of those who wish to evade the filter by creatively mis-spelling something. This is all very very obvious to anyone who has ever worked on content filtering, and not much less obvious to someone who hasn't and just sits thinking for five minutes or so. [1] https://en.wikipedia.org/wiki/Scunthorpe_problem https://en.wikipedia.org/wiki/Scunthorpe_problem

nortonham 4y ago
>Many English-language slurs are also words with entirely different meanings in other languages. Possibly. But looking at lemmy, it seems like most interaction is in English. As such, it seems reasonable to expect people to make a good faith effort to learn what isn't acceptable on a mostly english language website. >It's just that hard-coding a list of English-language bad words, then matching on them in an extremely naive way, is not an excellent way of manifesting those principles in code. I imagine there are better ways to do it, but this goes back to who owns the instance, and what type of community do they want to foster. More to the point, the github issue originally linked was resolved (to the extent it can be) by having user created filters. https://join-lemmy.org/docs/en/administration/configuration.html https://join-lemmy.org/docs/en/administration/configuration.... https://github.com/LemmyNet/lemmy/issues/1481 https://github.com/LemmyNet/lemmy/issues/1481
Mezzie 4y ago
Also there are genuine reasons to want to be able to type out slurs that aren't racist. I have an interest in linguistics (complete with degree - so I'm not just saying that to be racist), and discussing something like 'how the use of ethnic slurs has changed over time in English speaking countries' or how the euphemism treadmill functions with regard to insults would be impossible with this filter in place. Similar as to how a hard coded 'no swears' would make it hard to discuss English infixes since the most obvious infix is 'fuck'. Which is the exact kind of niche, pedantic, really nerdy small group discussion that is a good match for this.

nortonham 4y ago

>Many English-language slurs are also words with entirely different meanings in other languages. Possibly. But looking at lemmy, it seems like most interaction is in English. As such, it seems reasonable to expect people to make a good faith effort to learn what isn't acceptable on a mostly english language website. >It's just that hard-coding a list of English-language bad words, then matching on them in an extremely naive way, is not an excellent way of manifesting those principles in code. I imagine there are better ways to do it, but this goes back to who owns the instance, and what type of community do they want to foster. More to the point, the github issue originally linked was resolved (to the extent it can be) by having user created filters. https://join-lemmy.org/docs/en/administration/configuration.html https://join-lemmy.org/docs/en/administration/configuration.... https://github.com/LemmyNet/lemmy/issues/1481 https://github.com/LemmyNet/lemmy/issues/1481

Mezzie 4y ago

Also there are genuine reasons to want to be able to type out slurs that aren't racist. I have an interest in linguistics (complete with degree - so I'm not just saying that to be racist), and discussing something like 'how the use of ethnic slurs has changed over time in English speaking countries' or how the euphemism treadmill functions with regard to insults would be impossible with this filter in place. Similar as to how a hard coded 'no swears' would make it hard to discuss English infixes since the most obvious infix is 'fuck'. Which is the exact kind of niche, pedantic, really nerdy small group discussion that is a good match for this.