Bluesky now checks against this array of slurs when creating a username | Hacker News Reader

Bluesky now checks against this array of slurs when creating a username

github.com

34 pointsKoftaBob84 comments

R.I.P. my sauerkraut fan club

Let me be the first to bring Jung into this, "What you resist, persists", in trying to avoid the usage of slurs, you had to go and commit them to a public branch of source code to be saved in the arctic vault for eternity.

Funny enough I went to a Krautrock mini-festival a few weeks ago, here in Germany. Nobody I know cares about the term "kraut". At this point it is just outdated WW2 lingo that never caught on here.

I feel like "Kartoffel" is often used in a more offensive way. Although I've never really seen someone getting really offended.

I think that is one group of people where English-based slurs haven't stuck. But on the other hand the languages around Germany have still plenty of slurs for Germans, which funnily won't be caught by the bluesky filter.

Funny, I’ve heard potet in Norway as well. Is this a common slur in Germanic countries?

It's supposedly used by neighbor countries and foreigners to refer to Germans. But it's common to see people of other areas using potato-related slurs against Irish and Russians too.

R.I.P. my friend Tito

Well, at some point they are going to learn about 1337sp33k.

Also, I learned some new and interesting words today.

I should open a pull request to add "karen" to the list and see what happens.

S*yb*y is now a slur on the same level as racial slurs now?

No one has said that they are on the same level?

They're on the same level insofar as it's been deemed serious enough by someone to put on this list. Just funny to see it on a list filled with ethnic slurs and a couple disability slurs, looks very out of place to see an extremely-online insult alongside slurs with tons of historical baggage.

Does it really surprise you to see an extremely-online insult being included in a black list for an app for extremely-online people? I genuinely can't understand the confusion.

They are literally in the same flat array

That just means that they both meet the same bar. It doesn't follow that they are equally horrible.

Murder and tax evasion are both crimes, does that mean they're on the same level?

I don’t know why an array in source code would imply an ordering to you; it’s much more likely the author just picked a a list online.

Note that this list also contains several terms that are either reappropriations or contextual uses, and only applies to usernames. This is more or less what any social media network of a nontrivial size does.

not an ordering, but a threshold, "as bad as" to the extent it's on the same list

“Threshold” means “at least,” not “as bad as.” I don’t think anybody thinks that the two are meaningfully comparable in badness.

It took me almost a full minute to figure out what this is supposed to be.

I think it’s “SoyBoy” for anyone clueless as me.

the whole userbase of bluesky, judging by how they're acting right now.

Isn't a blacklist typically a bad idea for all of the reasons people are listing in the comments (like scunthorpe)? I've never implemented a social media platform, but why would you use a blacklist as opposed to a flagging system with manual moderation (where maybe the flagged user gets sent to a quarantine where they're shadow banned until approval)? That way, as language evolves and new slurs are made, or somebody tries to circumvent the system, you can just manually ban that user. Is there a technical reason this is a bad idea?

It’s one layer of mitigation that makes some users very happy. Potentially there should be some kind of AI detection here instead but people haven’t thought to do that yet.

Manual flagging is probably how the user got found but since it’s too “obvious” that it’s an issue with certain words, it makes users upset.

Hopefully they can avoid the Scunthorpe problem.

As I follow the code, it's just a literal match. So "coon" is disallowed, but "CarrieCoon" (or "DirtyCoon"...) is still allowed.

That's great news for Maine Coon fans.

I think that the reappropriation of slurs, like yankee, whig, tory, queer, etc., is a positive thing. But this practice serves to lock them in place, helping to insure that they remain slurs.

https://en.wikipedia.org/wiki/Reappropriation

This is proof that vampires are real and they walk among us. You need to prune slurs more than two centuries old. Otherwise the mortals will catch on to you.

Yankee is a slur? I thought it was just an endearing name for Americans. (I know in the US it only refers to North Americans.)

Luckily only one language is spoken on this planet, english, or they would need to introduce many more such lists ;)

> "gyp",

There goes the GYP project handle[0]. At least GNU Gimp is still safe :)

[0] https://gyp.gsrc.io/

A pal's surname is on the list, guess he just has to change his name in real life

Last week, while on Long Island of all places, I got an alarming text from a friend of a friend whose name is pronounced Key-Kay.

There's an MLB player named Kiké Hernandez and there's been some fretting over his name, including calls for him to change it. Which is... something.

Context: Yesterday a user created a Bluesky account using the N-word as their handle and, unsurprisingly, people were not happy and there was clamor for the Bluesky dev team to add a filter at the account creation level.

My only surprise is that it took this long

At least they don’t have Guido in it… yet. I have major problems signing up for services these days. There’s some list out there that programmers use that says that my name is a slur.

The name was so popular in Italy it ended up becoming a “slur” (though personally the only time I’ve ever heard someone called Guido is when that was their name). I wonder if that will ever happen with other languages and their common names, I’m Russian so I immediately think of “Sergey” as one of our ultra common names.

It's not a slur exactly, but in the United States, "Ivan" is a common name used to refer (usually dismissively) to Russians.

I've heard the same thing happen with Mohammed

It is a slur. Italians just don't really care that much though so it loses most of its power.

WOP is another one, but again it's gets laughed off for the most part.

I always find it strange how some slurs get clung to like people don't want them to go away.

Some slurs have large organizations behind them telling people they need to be mad. Others don't, and people end up not caring. My background is split 50/50 and both of them have common slurs associated with them. Nobody in my family cares cares, including me.

It's actually just a name. https://en.wikipedia.org/wiki/Guido

You might want to actually read that wikipedia post.....

"The slang term Guido is used in American culture to refer derogatorily to an urban working-class Italian or Italian-American male who is overly aggressive or macho with a tendency for certain conspicuous behavior.[3] It may also be used as a more general ethnic slur for working-class urban Italian Americans.[4]"

I'm not American and I know plenty of Guidos. It's just a name, and banning people named that is incredibly discriminatory.

I feel the same. Am I supposed to legally change my name so I can continue to sign up for stuff on the internet?

Okay? Strawman much? I'm not arguing that anyone do anything about it. I'm just telling you objectively what it is. The Wikipedia article YOU cited disagrees with you...

> add more words that should be cought

I hope they have fewer typos in their slur list than in their code comments :)

I started reading the list from the bottom up, and saw the N word written with a W a first letter. I first thought they were going to do a .replace("W", "N") to avoid spelling the word in plain text.

They'll have issues once github bans slurs from source code! ;)

It's not a typo. Google it. It's a derogatory term for white people who adopt black culture and mannerisms.

I think they mean cought vs caught

GitHub already bans accounts and repos for using no-no words in commits, PRs, and comments. See: https://www.techdirt.com/2021/02/03/content-moderation-case-...

This is so laughable, like these pajeets think that blocking the word golliwog on their so called open protocol of future communication will stop it from turning into a cheap 4chan clone

The solution to slurs and insults is always to let them speak and let them be ridiculed, not arbitrary censorship that will never be effective other than to give the platform a veneer of "we've done something" credibility

Add this to the list of "running a social network is hard" issues

"running a social network and not making silly blacklists of names is hard"

running a social network for people that can't handle rude words on a screen is hard.

Forgive my naïveté but isn’t bluesky open source and decentralized? So wouldn’t say some group just spin up an instance without these restrictions?

Yes, but they're currently prefederation (can't run your own instance yet) and still making architectural decisions about how moderation will work.

they're balancing onboarding new users with keeping up with high expectations of their current userbase. I'm not sure how the federation story compares with mastadon, I've mostly heard talk of composite block lists (everyone you block is public on bluesky so you could have rules that say block everyone that x,y,z have blocked, minus people I follow)

> Yes, but they're currently prefederation (can't run your own instance yet)

What's stopping you?

You actually can now, but it's still in "sandbox" stage.

https://atproto.com/blog/federation-developer-sandbox

So you could take their code, tell them to screw off, and run your own instances independently of them?

Yes, but since they’re by far the largest instance right now and would prob degenerate with you, you’d need to be cool with the fact that only people on your instance and some small other ones would be able to see your posts.

If you want to use the words that badly, then yes.

Pretty much. To clarify, Bluesky is a client that uses the AT Protocol. The instance that Bluesky the client uses is bsky.app, but anyone can create their own instance that doesn't refer to this slur list when someone creates an account.

What'll happen though is that the bsky.app instance (which is the largest one and basically the flagship right now) will probably de-federate with instances that have usernames that include those slurs.

Maybe they can untangle the vanity plate situation as well: https://www.aclu.org/press-releases/court-rules-ban-offensiv...

It’s strange to me that the block is username specific. Having it in your username seems like you’re self identifying as such, potentially owning it and “taking it back”.

Like I guess your username could be “BobHates<Slur>s” but that just self identifies you as an idiot.

What the devil is a "spade" other than something you make sandcastles with?

Watch “A Bronx Tale”

a card suit

They are missing the T word, the F word, the D word, the S word, and the M word. Now obviously only some of those are racial slurs but I think keeping this in the code base rather than a database is definitely asking for issues

> T word, the F word, the D word, the S word, and the M word

God, this is so childish. Perhaps you think that list should also only list words as "f-word" and have software interpret what it refers to so as not to offend anyone reading the code.

Funny anecdote about this: in the podcast "Blocked and Reported" one of the presenters tells a story she heard from a mother. Her child learned in school that "under no circumstance is it okay to say the n-word because it's the most vile, horrible, unforgivable thing anyone can say". The kid is 7 years old (or about that age) and has no idea what the n-word is so he asks the teacher. But the teacher refuses to say it because she's bound by her own logic of "under no circumstance" so kid is has no idea which of the many words beginning with n he's not supposed to say.

I also have a funny anecdote: I made a comment in Hacker News saying a list of slurs was incomplete and only mentioning the initial letters of a few slur words not everyone understood me.

The point of not mentioning a full word was because for a slur to be a slur it needs to be in a shared context. Otherwise maybe it's just a shortened version of Michael or an embankment used to prevent flooding. The child in your anecdote is a perfect example of that. Thank you.

Tesla, Ford, Daihatsu, Suzuki, Mitsubishi?

tranny, faggot, dick, slut, motherfucker

mind you, I only condone usage of the 3rd and 5th.

Thanks, albeit I was tongue-in-cheek, I was thinking F-word woukd be fuck, and all the other blew over my head. I couldn't understand why they would censor fuck in a nickname.

Slur blacklists have a problem with different languages.

Was actually thinking Dyke, Seppo, and Mick to keep with the theme of slurs against groups of people. (Slut is obviously a slur as well but I think dick and motherfucker are just insults...

Surprised they didn't do it sooner... this is like step 1 of automated moderation. If you've got an input, someone will put a slur in it.

I still can't sign up and see what it's all about anyway. I'm not losing sleep over it, but I'd still like to poke around.

That's gonna end up being a HUGE array.

Give me five good men and fifteen minutes, we'll compose a far more comprehensive list of cuss words. lol

I'm torn between wanting to add additional slurs and not wanting such a commit on my github history.

Must be a fun morning for the devs to come up with a list. Not very complete as of yet.

My grandmother could have made a killing as a consultant on this kind of thing with her extensive domain expertise, although I’m not sure how much Bluesky is interested in slurs for those English people across the river.

what a joke. lol

Seriously. My single-dev web service has a better list.

Reply on news.ycombinator.com