Confirmation if cursing is present

Question

I had no idea that cursing was not automatically filtered/flagged/censored on the site, until I had a user flip out on me for downvoting his answer.

Since the site has a strict no cursing policy, how come one of these clean-up methods is not applied?

The words the user used in his comments were clear curse words, and there do not appear to be any other meanings they could carry. Shouldn't the system automatically take care of such situations?

Perhaps a user can receive a warning before posting: "hey are you sure you want to post this?" with a link to the rule. Then the question could either get auto flagged or put into review. Granted, I do think that most visitors would reword their post if they received such warning.

Edit:

After a few answers I see the issues with making an auto filter. I never meant prohibiting such posts entirely, simply giving a user a form of "heads up." If someone is angry, they might not be thinking clearly, thus leading themselves into trouble.

Also, those who are angry will most likely not be the same people who are trying to find creative ways to outsmart the filter.

Similarly to how automated phone systems now attempt to predict the emotional state of the caller, based on the way they respond to the automated system.

Intent is not to block, but to make think twice before hitting submit.

Overcoming the filter will not do any good, as the user can post regardless, just like they can right now. The post will not be auto censored. As others have said, the community can take care of flagging if it is needed.

Answer 1

Banning curse words doesn't ban the intent and meaning of those words. Rather than trying to deal with the creative ways people would use to get their point across if we had a ban, we can instead simply search for the curse words and find most cases where that sort of abusive comment is being made. In effect, a non-ban is more effective at banning the abuse since it's easier to find and shut down. This is the same reason I am against the ban of duplicate titles, since all it does is make it harder to find crap questions, but that's a bit tangential.

And, as already noted, there are appropriate uses: Questions on the English site, quotations, the name of a certain programming language, etc.

Edit: A confirmation step is better than an outright filter, but I think it would be annoying to all the users participating constructively if they were using a filtered word. I don't really have much else against it; the devs might have greater insight.

Answer 2

Apart from the "it can't be done easily/well/correctly" of the Zypher's link in the comment, it's simply not a big enough problem to worry about. If users get out of line they are flagged and the mods take care of it - including banning repeat offenders.

If people started using inappropriate language on a regular basis then it might be worthwhile looking into implementing something. Otherwise it's better to treat the special cases by allowing the community to report and the leaders to react.

If the users expect a high standard from each other, and the mods personally reprimand users when they go out of bounds, then, so far, we generally get good results when it comes to inappropriate language.

If we "automatically" took care of it, some people would get in the habit of posting inappropriate material expecting the system to take care of it (Meaning we'd have to cover every text input field with the filter, not just a few), or worse - seeing what they can get away with and testing the filter. There are an infinite number of ways to put a bad word on the screen, given unicode.

Answer 3

I had no idea that cursing was not automatically filtered/flagged/censored on the site, until I had a user flip out on me for downvoting his answer.

Since the site has a strict no cursing policy, how come one of these clean-up methods is not applied?

What is the real problem here, cursing, or abuse?

Abuse

Abuse is mentioned explicitly in the FAQ:

We actively moderate our community, but we need your help to do so. Anything that is getting consistently flagged by our community members will be investigated and followed up on. And of course you can always email us directly if you feel the matter is urgent.

Most importantly, don't feed the trolls! Replying to abusive, off-topic, or inappropriate content only encourages it – whereas flagging allows removal without providing undue attention.

There is already a mechanism present, and in my experience it is highly effective.

If this isn't enough for you, how would you automatically detect abuse in cases where there was no cursing present? Word filters won't work (I'll get to that point later).

Cursing

I don't see cursing mentioned in the FAQ at all, except implicitly as "anything that is getting consistently flagged by our community".

Edit: Cursing isn't allowed, though that doesn't change my final point.

The words the user used in his comments were clear curse words, and there do not appear to be any other meanings they could carry.

You're conflating abuse with cursing. Cursing really isn't bad in a non-abusive context - it is just another way to express one's self.

If someone is angry, they might not be thinking clearly, thus leading themselves into trouble.

As you've pointed out, words don't matter as much as intentions.

But perfect word filters still cannot gather any information on intent. Intent requires AI, and implementation will not be cheap.

Implementation Cost

As detailed in the article Zypher linked in a comment on the OP, word filters suck. Any work on them would be inaccurate and open-ended, and still wouldn't be robust in any sense.

Any word filter would impact site perf. It doesn't matter if it is server side or client side. The server still has to serve up the script, and the processing still will impact end users.

As others have said, the community can take care of flagging if it is needed.

As you point out, there are already mechanisms in place that cover abuse, and they work quite well. Any work in this area would be duplicated effort.

Return on Investment

Anecdotally, I don't see much abuse in general, and very little cursing. Mostly just flame wars and people feeding trolls rather than flagging - and still quite little of that. Our community rocks :)

I don't think we'd get much of any return on the investment, especially since the flagging mechanism already covers most of the cases and already works fairly well.

The Specific Scenario: Our Community

Similarly to how automated phone systems now attempt to predict the emotional state of the caller, based on the way they respond to the automated system.

It might be different if this community were gmail, and were highly prone to drunk-messaging, or were tech support, and people were paying for products. This community is neither of those, so we require different optimizations.

While this suggestion is good food for thought, I don't think our community matches those scenarios very well, so those solutions wouldn't be as effective here.

Answer 4

Quite apart from the problem being comparatively small, as suggested by others, your question does make a lot of assumptions. When dealing with inappropriate language, context is everything. A few hours before seeing this question, I read an article in the Independent about a proud resident of Effin (in Eire) battling with Facebook to get them to accept the name of their town as part of their details. You say:

The words the user used in his comments were clear curse words, and there do not appear to be any other meanings they could carry.

In the particular case you may be right, but all an automatic filter can reasonably do is recognize particular words, and not the context and intent of the author.

Further you suggest that

the site has a strict no cursing policy

I thought the policy was effectively "be kind" - where is the strict policy of no cursing.

Finally, your description of the policy does not take into account cultural differences - as a Brit, I often find Americans' approach to cursing, rather "prissy"; and I have no doubt that similar or reverse views may be held in Americans' view of English.