The Unfriendly Robot: Automatically flagging unwelcoming comments | Hacker News Reader

The Unfriendly Robot: Automatically flagging unwelcoming comments

stackoverflow.blog

27 pointsJasonPunyon23 comments

When Stack Overflow started, I tried to contribute but never succeeded, there was some kind of chicken-egg issue that prevented any answer to get through, I don't really remember what it was, just gave up since I wasn't interested in karmas enough to jump through those hoops.

So it became a read-only resource for me, fair enough. Now most of the useful information I find is decorated with comments from some moderator or contributor talking shit about the question or the answer. Somehow I feel the unfriendliness directed towards me, as if I should feel ashamed to want the wrong answer to the wrong question. Then I read the first sentence of this article:

We all want Stack Overflow to be a welcoming and friendly place.

I do, actually. But not sure about all.

Stack Overflow had a great thing for a long, long time. Once upon a time, the Stack Overflow karma score actually reflected one's domain knowledge and community spirit. But, as with anything that becomes successful and measures success comparatively, it attracted the attentions of people who are competitive, smart and a little ruthless. The score itself became the goal for these people, not the actual content.

Unfortunately, via the powers granted through its reputation system, it also rewarded these smart, competitive, a little ruthless people with tools to restrict and control content that others produce. So smart, competitive, a bit ruthless people control content for the rest of us, who just want to ask and answer questions about tech.

So, a sincere good luck with techno-solutions to a deep-seated, cultural problem over there. I sincerely hope it works. It would be nice to have the old Stack Overflow back.

Yep, plenty of ways to be sarcastic, hostile and smarmy without triggering some neural net.

I had the same thought, but I think this except from the post assuaged some of those concerns:

"We want to make clear here that even though the current version of the robot appears to be performing amazingly well, our human flaggers are just as important as ever. The only reason the robot is able to do as well as it’s doing is because of the 100,000+ flags that humans have raised and moderators have handled. The robot is a way to find things that look like things humans have flagged before, but we don’t think it can identify truly novel modes of unfriendliness. Only the humans in our system can do that. So if you see unfriendliness in the comments, please flag it. You’re helping to make Stack Overflow better."

Unfriendly comments are the least of their problems. The fundamental problem is their broken reputation system.

Questions are outright closed by people with a high enough karma even if it is outside of their knowledge domain. People who leave helpful feedback after a downvote will receive a retaliatory downvote in turn. People with massive karma will receive upvotes even if their answer is not a good one. To boost karma, it's better to be first than good, in general. New questions are closed and redirected to old questions with outdated answers. Non-native English writers get downvoted for grammatical errors, even if the question is fine. One downvote will result in more, as the mob piles on. Bringing up any of these issues in their "meta" results in downvotes.

So, Stack Overflow will focus on its moderation system and "unfriendly comments" while its entire system encourages a kind of arbitrariness that is itself unfriendly.

Exactly, the biggest problem is the culture of existing users, which was built up by stack overflow encouraging a competition for internet points which leads to all kinds of antisocial behaviour and attempts to game the system and discourage participation by rivals.

The worst part for me is when you ask a hard question and rather than admitting they don't know the answer they decide to close it because they think they have got you on a technicality about the validity of the question.

Or a similar variation, it's a hard problem, and they don't know, so they assume you are a complete idiot and ask you to verify things that were implicit in the question. Like, you said you did X and Y, and it's almost impossible to do those things without doing A and B first, but they give you this patronizing tone saying you should have done A and B first.

Here's another one: you write a decent answer to a question. Someone with a massive score comes along behind you an hour later and takes your answer and elaborates on it a bit. They get the upvote and check, you get a zero. The reason they got the points was because they spend all day gaming Stack Overflow and are happy to take advantage of any way to steal points and so have a high reputation.

I really feel like their data science team is not trying hard enough to find the dyed-in-the-wool pathological sons of bitches that are on Stack Overflow. There are plenty of them on there.

The problem is the pathological users ARE the active, participating userbase, the site is built on their input to cater to them and their desires.

Its the same reason that online communities almost inevitably degrade over time, they optimize for the most active, loudest users above everything else, which is not necessarily the path to a better community.

Still, there aren't alternatives on Stack Overflow. You either post there and hope your question isn't closed and it will get an insightful answer putting you on the good track, or you do some digging yourself until you get the answer, even if it's going to take a massive amount of time.

While I find Stack Overflow very useful, I find it's model very restrictive.

Sometimes you can get more insight on how to solve a problem from a discussion than from an answer. Why and when can be as important as how.

Also, if people dislike your question, even it obeys the rules, it gets down voted. They can dislike your question because they dislike the problem you are going to solve or the way you are trying to do it.

Some comments instead of being related to the question at hand, suggest that you shouldn't solve your problem but instead do things another way. Which can be fine, but doesn't fit well the site format because it doesn't allow discussions.

However, I don't see alternatives. You can ask on reddit but there are not many people willing to answer.

If you're lucky, your language/library/etc might have a fairly active IRC channel or similar. Popular languages often have a myriad of channels dedicated to different subtopics.

They're nice because like you said, discussion can be much more helpful than a single answer.

Great suggestion.

However, one issue I've run into with the rise of chat and SO, and removal of forums, is that if your question isn't the right fit for SO, and is a bit tricky/time consuming, it's easy for your query to be buried.

I reached out to the TypeScript community via Discord about assistance on how one would add TypeScript definitions to an existing JS repo. Got a little discussion, IIRC, and then because of other factors, was buried.

For math, I like math.stackexchange.com.

I know HN likes to berate on SO as if it’s some has-been that nobody uses anymore.

Just the other day I asked a fairly specific question, putting in some effort of writing code examples and add links to relevant documentation and within a day I had a four paragraph, in-depth answer.

No, I did not get any upvotes or smiley-stickers, but I got for free what I wouldn’t even know how to buy.

Moreover, I end up on SO daily when programming.

The utility of SO is second to none and it has been that way for over a decade.

That is pretty amazing in my book.

SO is still good, what they mean is that it has peaked or in the process of peaking. Just like previously-cool bars or festivals are still good but just 'not the same' and on the verge of becoming bad. It's a very specific mindset where you must be growing (not just growing, the growth rate must also be growing) and avant-garde all the time, and it's not really surprising to see it on a forum about startups.

I'll also add that although SO was of course invaluable to me as a reader, the few times I've had to actually ask questions because I couldn't find what I needed, other users were stumped as well. Just adding my little anecdote to yours

I feel they never truly recovered moderator goodwill after that fiasco months ago so they tried to promote more people into doing it, me included (1.3k to 2.1k reputation overnight)

Apparently it didn't work (I'll not edit people's questions!) so now they'll try automating it. Good luck to them but I'm not seeing them succeed in an area where nobody has.

And, as others said, SO isn't the same as before. It's mostly a homework finder and it doesn't encourage newer versions of code.

Stack Overflow karma system is broken. It turned into a playground for people who like to show off their power given by the points while becoming a searchable library of code snippets for the others. The code snippets tend to be old, since asking new questions is discouraged and nobody likes to update old questions with new answers.

That's a dark pattern.

They didn't like having to openly apply human judgement so they applied AI. AI that has some human judgement built in. And bias along with it. But now nobody is responsible because the machine did it.

TFA appears to address this, the automated system is used to help identify comments for human moderators. The human is the one responsible. And according to stack overflow they plan to use this system only to help the human moderators.

If you don't believe them and suspect they will have a fully AI flagging system in the future (in direct contradiction of this post) thats fine but it might be better stated as such, since human judgement is currently still paramount.

The main issue of Stack Overflow is the YX problem. The YX problem is, of course, the reverse (perverse?) of the XY problem, whereby, confronted by a question that stumps them, an SO user will, instead of admitting so, do one of the following:

-Attempt to second guess your use case for you. Are you sure you really want to do X? Don't you want to do Y instead? (No thanks, I really must do X and can't do Y because I have Z, A, B and C constraints.)

-Claim your use case is wrong, should not exist, or at the very least is exceedingly niche, and they couldn't possibly imagine why someone would have your use case

-Claim your use case is literally impossible to fulfill with the constraints you've given. While this may well be the case it is a way stronger claim than most people realize.

-If someone does provide an answer to your use case, will scramble to insist that this solution is clumsy and shouldn't have be written in the first place because it could give people wrong ideas.

-Insist you didn't do the research and if you did you'd find the solution is Y. (Yes I did, and it really isn't Y because my problem is X, thank you very much.)

-In the worst case, wrongly mark your question as a duplicate of Y

The strength of SO is that it is purely a question and answer website: I give question, you give answer. Forums always sucked because the question askers were often new and inexperienced and formulated questions the wrong way, and the entrenched users were patronizing and insisted a certain way of solving questions be used or say useless stuff like "I don't owe any of my time to you" (yeah well why are you bothering writing this post in the first place then?). With SO's system of questions, answers and comments, everything is clear or clarified, all the chitchat is dispensed with and no room is given to subjective stuff like 'is your question correct?' Because no matter how good of a hacker you are, your knowledge can't possibly cover 100% of all computer programmers' use cases.

I'm not pessimistic as other people as to the future of SO and I still think it is and will be an invaluable resource in the following years but if trends of the YX problem go for the worse it could mean a return to the forums that sucked.

More robo-flagging of human interaction based on the faulty axiom that you are somehow in control of how someone else perceives something. They even admit that malice isn't intended.

> The problem is the tone the reader experiences. Most of the time, it doesn’t appear that commenters are actively trying to make their comment condescending, dismissive, or any of the other subtle variations of unwelcoming we see. These are people earnestly trying to help others, even if their tone is off.

What tone the reader experiences and how they deal with it is entirely up to them. I am fine with on a site that abusive language is flagged, but something that is perceived to be unfriendly by a robot. This is just folly.

I certainly won't be contributing anymore. I don't want to be on a site that has a robot flagging human interaction.

EDIT: Abusive language filters usually don't work that well. I put in someone's real name into a system and the system flagged it for "abuse" their surname was "Cummings".

Also the term "faggots" (which in the UK is a meal you can buy in the supermarket) has been censored on social platforms.

https://www.rt.com/uk/468436-google-censors-faggots-peas/

Because guess what the context in which you are using a word is important. So most of theses places can't get abusive filtering right. They won't get this right either.

I'm just on math.stackexchange.com, and I greatly enjoy it. Maybe it's a good thing I am there instead of overflow.

No moderators left to ban, I see.

Reply on news.ycombinator.com