Stupid Use Of Profanity Filter Makes A Mess Of Virtual Paleontologist Conference
We've known for some time that the sorts of automated filters that get applied to various internet-y things are flawed in the extreme. But of all the filters that annoy me the most, profanity filters are the worst. And, no, it's not just because I use curse words like commas. Rather, it's the combination of just how badly this is used, such as how Google thought for years that "bisexual" was a naughty word, along side how nefarious actors can block all sorts of non-profane language just by calling it profane. Add to all of this that a total lack of nuance for identifying so-called "naughty words" regularly causes perfectly non-profane content to be blocked or censored and this all begins to look like an exercise worth giving up.For a great example of that last bit, we need only to look at a recent remote conference conducted by paleontologists that went awry due to a profanity filter.
Participants in a virtual paleontology session found themselves caught between a rock and a hard place last week, when a profanity filter prevented them from using certain words – such as bone, pubic, stream and, er, beaver – during an online conference.“Words like ‘bone’, ‘pubic’, and ‘stream’ are frankly ridiculous to ban in a field where we regularly find pubic bones in streams,” said Brigid Christison, a master’s student in biology attending the event, in an interview with Vice.Why, yes, that is really stupid. If your profanity filter is filtering out words you need to use for your field, then your profanity filter sucks and should be done away with. And, really, are members of the Society of Vertebrate Paleontology really so sensitive that any profanity filter need be in place at all. These people are adults and can be trusted, not journalists for The New Yorker.Now, much of the blame for this comes from the organizers of the event for some reason including a filter setup for typical business meetings.
“Apparently it came with a pre-packaged naughty-word filter. After getting a good belly laugh out of the way on the first day and some creative wording (my personal favorite was Heck Creek for Hell Creek), some of us reached out to the business office, and they’ve been un-banning words as we stumble across them,” an SVP member explained to Reddit users.I'm not entirely sure why any of this is funny, to be honest. It's just annoying. Especially, as the article notes, when there are some curious choices made in the stock filter as to what words to filter out as profane or not. "Wang" is filtered for instance, despite it being a common last name, but "Johnson," which has the same slang meaning, is totally allowed.Again, all of this is simply annoying and unnecessary. Trust adults to be adults and either not use profane words, or else be able to handle it if someone else occasionally does. These filters aren't working.
Read more here
posted at: 12:00am on 22-Oct-2020
path: /Policy | permalink | edit (requires password)
Content Moderation Case Study: Handling Trolls Invading A Community (1993)
Summary: Before even the World Wide Web existed, Usenet was a popular gathering place for various niche communities to congregate. In many ways it was similar to what Reddit has become today, except that it was not controlled by any single company. Instead there were a number of newsgroups (like subreddits) and various news servers that could choose to carry whichever news groups they wanted.Each news group was a specialized topic area, so there could be newsgroups for Bob Dylan's music (rec.music.dylan) or computer languages (comp.lang.lisp). One newsgroup, started in 1990, was alt.tasteless which, as the name suggests, was a newsgroup for sharing offensive content. As its Frequently Asked Questions (FAQ) noted:
A newsgroup devoted to tasteless phenomena in all its forms.A place for people with a twisted and sick sense of humour.In alt.tasteless we like to get into the details. We wantthe feel of it, the smell of it, the stench of it, everylittle rotten and puss-oozing detail. And then, of course,some rough pictures of it in alt.binaries.pictures.tastelessor alt.tasteless.pictures.One of the founding members of alt.tasteless, Steven Snedker, told Wired Magazine in 1994 that one of the reasons it was created was to keep other parts of Usenet clean. Except, as a classic Wired article notes, that's not what happened. Instead, alt.tasteless decided to invade another newsgroup: rec.pets.cats, a newsgroup discussing cats. Today, having trolls from a tasteless part of the internet invading a more wholesome spot is a known phenomena. But the alt.tasteless invasion of rec.pets.cats in August of 1993 may have been the first case.It began with a fake anonymous posting to rec.pets.cats by a member of alt.tasteless. The subject was Cats & Dates and it told a slightly off-color story about a pet cat in heat scaring off dates. Many participants in rec.pets.cats tried providing useful advice, but then others from alt.tasteless started responding with tasteless responses.This only began an escalating war between newsgroups in which one side (alt.tasteless) clearly held the upper hand. As summarized by Wired, the leader of the alt.tasteless effort was mostly a college student named Constantino Tobio Jr., who went by the evocative name Traschcan Man on alt.tasteless.
During the ensuing months, it is safe to say that hundreds of messages passed back and forth. Whenever life began to return to normal at rec.pets.cats, someone from alt.tasteless would post an article there looking for, say, a good recipe for Polynesian cat. Occasionally, someone from rec.pets.cats would try to fight back by posting an article about his or her cute kitty on alt.tasteless. But that only led people like Trashcan Man to respond with their own articles about topics such as vivisecting the cat and having sex with its innards.Yes, Trashcan Man was right there on the front lines of the war.One day, a (real) rec.pets.cat person posted an article to his own news group asking whether anyone could suggest a way of keeping a neighborhood dog from harassing his cat. Trashcan Man had a suggestion: Spray it in the face with muriatic acid, a form of hydrochloric acid. Of course this is dangerous - about as sensible as warming a cat by putting it in a microwave oven (which someone had previously recommended).This situation created something of a crisis on rec.pets.cats, where people were (a) just wanting to talk about cats and (b) worried that someone might actually fall for the dangerous suggestions from the invading users.Decisions to be made by participants in rec.pets.cats:
Questions and policy implications to consider:
- How do you stop users who are posting to the community for clearly malicious reasons?
- How can you distinguish bad actors from those who legitimately want to be on rec.pets.cats?
- Is there any way to prevent the invasion from continuing?
- Are there ways to effectively punish those who began the trolling behavior?
Resolution: One user on rec.pets.cats, Karen Kolling, decided to take matters into her own hands, first teaching the (not always tech savvy) users of the newsgroup how to use kill files to block the trolls from alt.tasteless, in an effort to lessen their impact.
- On a system without central control, who handles content moderation decisions?
- Is it reasonable to leave such decisions up to the users in the community?
- Can there be open niche communities that will not be ruined by malicious users?
Kolling began teaching the cat people how to write "kill files."On Usenet, kill files are a way to filter out messages from people you don't want to see. Some places refer to them as "bozo filters." With a kill file, you can, say, screen out any messages posted by someone named Trashcan Man, so that when you go through the day's articles none of his will show up. It will be as if they were never written."Let's say Joe Smith posted a message to rec.pets.cats, full of descriptions of how he likes to mutilate cats," Kolling explains. "You can set the kill file to get rid of the messages of Joe Smith." And that is precisely what she taught her friends to do. It worked for a while, until alt.tasteless people began "counterfeiting names so you wouldn't know it was Joe Smith," she says.The battle between the two newsgroups escalated, including personal attacks on Kolling. Eventually, she realized that she needed to get the leader of the attack -- Trashcan Man -- to stop. And she did that by getting his internet service provider to threaten to disconnect him:
Kolling and a few of her Net-savvy friends began contacting the people who provide Net access to some of the more flagrant abusers. These system administrators were at universities in some cases; in others, they ran commercial gateways to the Internet."What I was hoping was that the system administrators would just tell them, 'Hey, grow up,' and that would be enough," she says.Maybe that's what did happen in some cases. In other cases, though - cases like Trashcan Man's - the reaction was more pointed. At the time, Trashcan Man was reduced to buying net access from Panix, a popular provider in New York City. He had had a free account at Columbia, but that was suspended after Trashcan Man planted a thing called a fork bomb on the system, which caused it to crash.One morning late in the fall, Trashcan Man connected to Panix and got a stern warning from the owner, Alexis Rosen. In e-mail Rosen noted that there had been complaints - again - about Trashcan Man. The tone of the message was clear: Knock it off, or else."So I knocked it off," Trashcan Man says, "because I value my Net access."The Wired story ends by saying things calmed down somewhat after that, with Trashcan Man saying that he'd learned his lesson (though he said he was still flaming people -- but just those he believed deserved it). Soon after the article ran in 1994, Trashcan Man graduated from Columbia University. According to LinkedIn he's now been an engineer at a tech company for more than two decades.
Read more here
posted at: 12:00am on 22-Oct-2020
path: /Policy | permalink | edit (requires password)