Hacker News Reader

Comment by ufmace | Hacker News Reader

I know they've removed it since the time of the comments I will refer, but I still feel like I would not choose Lemmy to host a forum due to the conduct of the maintainers on the threads regarding the hard-coded slur_filter. See https://github.com/LemmyNet/lemmy/issues/622. When the conversation with the project maintainers goes like:

> Hey I don't think a hardcoded slur filter is a good idea due to good reasons a, b, and c that don't relate to actually being a troll or calling people slurs in a hostile way

> Screw you racist, we're never going to remove it

I feel like their priority is no longer to provide a flexible tool other people can use as a communication platform in the way that best suits their community, but instead to impose a narrow-minded agenda on everybody. After that, even if they change their mind later, who's to say they won't change it again at some point?

View on HN

I've seen a few recent projects hard coding things like this. I am not a fan. Not because I want to use the words but rather I want to be in control of the instance I am running. UnrealIRCD for example has an optional module for word filtering that is quite powerful and can be left entirely blank, or use regex to replace words, block words or kick someone off. I just prefer having flexibility and multiple options. One of the goals of open source was to give people freedom of choice.

It's too bad as I really like the idea. I suspect and hope more things like this will pop up with recent events. Ultimately it is their project and their choice. I think I will stick with phpBB [1] as a forum, UnrealIRCD for the deep dark places [2], uMurmur for real time text and voice and the Mumla app to access uMurmur from my phone but that is just my personal preference.

[1] - https://www.phpbb.com/

[2] - https://www.unrealircd.org/

[3] - https://github.com/umurmur/umurmur

I should add to this for completeness sake that if one finds UnrealIRCD to be too complicated, the simpler alternative is NGIRCD [1]. It takes all of maybe 4 or 5 minutes to get it set up and running with TLS from LetsEncrypt. It's dirt simple but sometimes quick and simple is useful in a pinch.

[1] - https://ngircd.barton.de/

I just read the link. I don't think your characterization of the conversation is fair. Either way, it seems like the people behind the project have a CoC about what language is allowed. Further, the first response seems pretty reasonable to me from both a technical and ethical standpoint.

I would add, lemmy doesn't seem to be about pure free speech, or free speech above all else. They've made it clear that slurs are not acceptable, and anyone who doesn't agree is free to go somewhere else.

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

>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....

https://github.com/LemmyNet/lemmy/issues/1481

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.

> I don't think your characterization of the conversation is fair.

You're free to think so. I for one don't care to censor those who disagree with me. Note that there's a number of deleted comments, and neither of us know what those say.

> anyone who doesn't agree is free to go somewhere else

Fair enough, and I just said above that that's exactly what I intend to do.

I think it needs to be remembered here that their project is attempting to build a forum hosting tool, not a particular forum. If a particular forum intends to censor any slur list they choose, I have no issue with that, that's entirely their right, and I'll post on it or not. But I feel it's highly inappropriate for a team building a hosting tool to impose a particular chosen slur list on anyone using their tool to host a forum, with no regard to what the purpose or character of that forum is intended to be. I shall indeed choose not to use such a tool.

>Note that there's a number of deleted comments, and neither of us know what those say

Does that mean you get to just create a narrative? Dessalines is the closest in that thread with telling people they can use voat or gab if they want to use slurs, which at least implies some racism but that other dev Nutomic has a pretty inoffensive stance when he says

>If you dont like it, fork it. Stop bothering us about it, we will never fully remove the slur filter.

>Does that mean you get to just create a narrative?

Eh? GP was pretty clear in showing that the developers of the project are pro-censorship, not only in hard-coding in naive badword filters, but also in how they handle responses to requests on their own project

>Dessalines is the closest in that thread with telling people they can use voat or gab if they want to use slurs, which at least implies some racism

There is no "some" in their statement. They are saying that if you have a problem with blanket-filtering language, then that means that you want to use words that are designated as bad, and that only bad people use those words. It's purely black-and-white. Nobody could ever want to use those words in any other context, according to them

>that other dev Nutomic has a pretty inoffensive stance when he says "If you dont like it, fork it. Stop bothering us about it, we will never fully remove the slur filter."

On the contrary, saying that anyone who disagrees with you is a "bother" is offensive, especially if you flip-flop on your position and plug the change in anyways, which is what they did

The user I was replying to dropped this as a quote when describing what the devs were saying.

> Screw you racist, we're never going to remove it

That's creating a narrative, it was never said.

>On the contrary, saying that anyone who disagrees with you is a "bother" is offensive, especially if you flip-flop on your position and plug the change in anyways, which is what they did

Literally anytime you talk to someone and they don't want you to, that's bothering them. There's a discussion to be had if its reasonable to be bothered by the behavior, but whether or not its is bothering someone is literally defined by that person's belief.

On that note, if you find what Nutomic said offensive, I think thats a little too trigger happy on being offended.

>Nobody could ever want to use those words in any other context, according to them

In what context would you want to write out a slur against someone as part of some online discussion? Even if you were quoting some external source, you could just censor the slur/hate speech, and still carry on with the discussion.

> In what context would you want to write out a slur against someone

There are other contexts than using a slur against someone. For example, describing an instance where a slur was used against you personally, explaining historical contexts of the slur, and so on. The existence of a slur does not mean that it's only ever going to be used in cases where it's being thrown at someone else

Right but in the example you use you could write sl*r or sl-r instead of slur and anyone reading would understand you and you would get past the filter.

No, you couldn't. For example, if you were teaching someone who's learning English about which words to avoid, they wouldn't know which words you're talking about since you're not actually using them

cocs are supposed to police the contributors, no?

who will you complain to when a user violates Corey Ehmke's Ten Commandments? the cyber police?

> cocs are supposed to police the contributors, no?

its a generic phrase. I definetly would have thought that if I was reading a github page; but not necessarily when I have to read a COC for joining a Discord.

Apparently it's getting tough to operate outside of everyone's random (usually poorly written) CoC on the net.

Discord and the other services have ToS/EULA. a legal agreement that has actual power, unlike github cocs

I read the whole thread, and the developers don't come off as principled, it feels more like they're just wedded to a particular implementation and are grasping at all straws to defend it. A singular regular expression as a slur filter is pretty terrible by any modern coding standard. I'd avoid the project on the grounds that the rest of their code is probably just as amateurish.

>I just read the link. I don't think your characterization of the conversation is fair.

"If you dont like it, fork it. Stop bothering us about it, we will never fully remove the slur filter."

"stop bothering us about it" is , in my experience, a huge red-flag to run into while perusing issues and commits.

a 'code of conduct' that dictates all use cases for software for the forever-future is a cute concept -- but 1) it has never worked, and 2) it's basically untenable in the open-source environment.

If the software IS workable, it WILL be forked, and your power-from-on-high to force behaviour on people crumbles. If the goal really is to force your political/ethical position on people who use your software it's probably better to do so more transparently so as to not scare the herd into another avenue of control from another actor.

As with any software tool in my preferred catalog of tools I prefer one that doesn't foist it's own methods and standards onto me; furthermore I prefer my tools don't foist their politics and societal standards onto me. I don't think that's the purpose of a tool.

>I would add, lemmy doesn't seem to be about pure free speech, or free speech above all else.

that moral absolutism doesn't need to exist within software. We know that hard-coded lists of restrictions hidden deep within software are a Bad Thing for a long variety of actual technical reasons -- there is nothing stopping the developers of Lemmy from shipping a filter-list that reflects their values right along side their software; the choice towards inflexibility and zero-tolerance just makes the software less reliable, less maintainable, more opinionated, and less popular.

well if people maintain/own a project, and say no to a request multiple times, eventually they will just say, please stop bothering us.

In regards to the technical side, the original linked github issue was resolved. See:

https://github.com/LemmyNet/lemmy/issues/622 https://join-lemmy.org/docs/en/administration/configuration....

Nah, they definitely overreacted to a reasonable request. Different languages, different cultures and different groups will have different words they consider as slurs or not acceptable.

Then there are edge cases like history forums where they have a valid use case for posting transcriptions with words that are currently deemed not appropriate.

Their response can drive possible adopters away because they could only think in US centric politics.

What is US centric about not wanting to allow words that are commonly understood in English to be slurs, on a site that is predominately English speaking?

Also, why is it wrong to expect non native speakers to eventually work out what is and isn't acceptable in day to day conversations with native speakers?

Isn't the idea of Lemmy to run your own instance that could be used to host communities that don't speak English? In their instances list they have servers that speak German, Spanish, Portuguese, Finnish, Euskera, French...

Why would some non English speaking person be forced to change their language because someone in the USA decided that an innocuous word in their language is now a slur?

Also USA is not the only country speaking English and some words considered slurs or offensive in the US are not in other countries that also speak English.

They did in the end made it configurable and that's good, but their initial response is a major turnoff for anyone thinking on seriously running the software without planning to maintain a fork in the future.

Ah yes, I should have thought of all the words in Euskera that overlap with racial/homophobic/transphobic slurs in English

Also note that slur filter is now configurable [1] and seems to be completely optional [2].

[1]: https://github.com/LemmyNet/lemmy/issues/622#issuecomment-89...

[2]: https://github.com/LemmyNet/lemmy/blob/235cc8b22897bfb3e71ba...

I agree with you, the dev team for Lemmy are hard line ideologues and it has hurt their original stated mission to take online communities out of the hands of corporations. As you noted, all that slur filter stuff was removed from the codebase some time ago, they've walked it back a bit, though they're still the same people of course. The software works.

If you're interested in a different federating link aggregator there's also lotide https://sr.ht/~vpzom/lotide/ which looks promising as well, and littr.me/littr.go/brutalinks https://sr.ht/~mariusor/brutalinks/ which is single community oriented like HN.

Yeah, their responses to legitimate discourse make this project a non-starter for anyone serious about relying on open source software.

It looks like this has been merged 6 days ago.

https://github.com/LemmyNet/lemmy/pull/2492

In case anyone was wondering, this was (is) the slur filter in question: https://github.com/LemmyNet/lemmy/blob/b18ea3e0cc620c3f97f98...

> this was (is) the slur filter

Was (not is) is correct: https://github.com/LemmyNet/lemmy/blob/71aed94a000c951ddfdb4...

So it only took them two years to decide that it should be a configurable option and not hardcoded? Doesn't really solve the terrible implementation, but it's a step.

Wow, many of those words have legitimate non-slur uses. I can't imagine an automotive forum where users cannot discuss how to retard the timing on a distributor.

> For example, there was someone who tried to talk about Stardew Valley and only managed to talk about Sremovedew Valley.

Even beyond whole legitimate words, it filtered inside legitimate words-- you'd think people would consider the Sremovedhorpe problem by now.

Just hopping in to say : it's laughable to claim that open source projects have "hard-coded" anything. It's literally impossible to hard code things when anyone can download the source, edit it, compile it and run it on any machine they own

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

> Hard coding (also hard-coding or hardcoding) is the software development practice of embedding data directly into the source code of a program or other executable object, as opposed to obtaining the data from external sources or generating it at runtime.

The real Lemmy wouldn't allow a slur filter. He'd tell you to use all the slurs you want and accept the consequences, and if that's a punch on the nose then it ought to be a lesson to you.

Be more Lemmy.

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

The slur filter is optional. You don't have to use it.

It is now.

But I see no point in using software that has such disdain for downstream users. I'd rather use software that treats me like an adult.

Adults are largely why software needs slur filters, so that's not really the high bar you'd expect it to be.

They’ve already said you are free to fork it.

Why do you feel they are obligated to cater to your needs?

>but instead to impose a narrow-minded agenda on everybody

Having a slur filter is not "narrow-minded agenda", it's mainstream. And, well, it's open source, so anyone can maintain a fork that removes it with not much work.

I'm LGBT and slurs are used socially within the in-group.

Nevermind the fact that art such as film frequently relies on the use of slurs when dealing with this subject matter.

We aren't protected in real life from being called words. I've certainly had my fair share hurled at me. By having to deal with it both at a young age and in adulthood, I've grown a thick skin and strong defense system.

I worry that this attitude is more dangerous that what is trying to be accomplished.

I'm a black American and I'm honestly sick of white people telling me when I can use the "n-word." I guess if they can't have it, no one can.

Good points from you and echelon. This (by which I mean hardcoded slur filtering in Lemmy/LemmyBB) is not a good thing. On the other hand, it's obviously within their rights to code such a thing into their own thing.

> I guess if they can't have it, no one can.

I'm a white american and I think I agree with this. Making sure that only one race of people can say bad words on the internet doesn't seem practical to me. If we don't want some races saying a word, I don't see a way of preventing that which doesn't sacrifice either anonymity or black people being able to say that.

There's only one implementation that I can imagine, and it's pretty impractical and ethically dubious. The idea is to have some kind of verification process to prove that you're a certain race, and somehow publicly store the public keys of people in each category. Then, when someone want to post a word that only some races can say, they submit a zero-knowledge proof that the message was written by someone who controls the private key of one of the public keys in the appropriate category. But I'm pretty sure that a registry of people organized by race is neither practical or desirable.

The only thing this does is make people invent new slurs.

Says the thromrod!

If you read the GitHub issue, the debate was not over its existence, just that instance operators should be able to configure it, which it sounds like they now can anyway (last comment on issue, he said it was added).

There was no need, and is quite malicious, to imply and talk down to their own users/supporters of their software as anything"-ist" in their motivations, which in many cases is considered a more corrosive/ruinous label (court of public opinion) than "convicted criminal" (court of law) in the West.

Having a slur filter on a particular forum instance designed and chosen by the admins of that site is a perfectly good idea. They know the character of their forum, they can choose what is and is not appropriate to say.

I think it does display a narrow-minded agenda for the authors of hosting software intended to potentially host a wide variety of forums to declare unilaterally that all sites hosted using their software must use a particular slur filter declared by them, or else create and maintain a fork.

Reply on news.ycombinator.com