7

"Spam" is defined right there in the flag popup as "Exists only to promote a product or service":

enter image description here

but most of the spam flags I process have no advertising, promotion etc. They are just VLQ, NAA, gibberish, etc.

In an effort to dissuade the spam flag's misuse, I decline incorrect spam flags with a "a moderator reviewed your flag but found no evidence to support it" message, or sometimes a custom message like "I see no advertising here". But it's a losing battle.


Can we add a field to the popup that asks the user to enter the url/name of the thing being advertised?

The idea being that if people are too lazy to read (as evidenced by the statistics), they are also too lazy to enter anything and so won't use the spam flag incorrectly.

It would also help speed the review process, making it faster to identify multiple spam posts from a user or about a product.


Alternate, simpler, proposal... remove the short names!

enter image description here


Admittedly, the amount of spam per day isn't that large, and perhaps this question should be tagged .

migrated from meta.stackoverflow.com Nov 23 '16 at 18:43

This question came from our discussion, support, and feature requests site for professional and enthusiast programmers.

  • 2
    Perhaps users with high decline rates of spam flags should have their flags rate limited, although if it is not many this may be too heavy handed. – Travis J Nov 22 '16 at 23:38
  • 27
    I don't really like this idea. I've spend a lot of time flagging spam that is caught by Smoke Detector. I have over a hundred helpful spam flags, and many users have waaaay more than that. I think it would be really obnoxious if I had to do this every single time I come across spam, which since it's cleaned up so quickly, is more often than you'd think. – DJMcMayhem Nov 22 '16 at 23:43
  • 4
    I assume this field would then accept phone numbers, email addresses etc. in cases where the thing being advertised is not a URL? – John Dvorak Nov 22 '16 at 23:44
  • 2
    @TravisJ that happens already. Decline rate > 25% and at least 10 flags in the past 7 days prevents you from flagging, plus declined flags reduce your flag/day count – John Dvorak Nov 22 '16 at 23:49
  • 3
    @Bohemian The alternate proposal is fine. I'm not really for or against it, but if you think it'll help cut down on improper flags, I'm fine with it. – DJMcMayhem Nov 23 '16 at 0:12
  • 10
    FWIW, I've previously custom flagged gibberish, only to have the message come back saying that a spam flag would've worked, so I gotta say the guidelines have been a little hazy, for me, anyway. Also, what's carthesis? Catharsis? – Mike M. Nov 23 '16 at 0:40
  • 1
    (OT) What's "carthesis"? What I found is status ailment from some obscure RPG and mistyping of "catharsis". – ivan_pozdeev Nov 23 '16 at 1:44
  • 2
    @ivan_pozdeev can't spel :/ – Bohemian Nov 23 '16 at 1:54
  • 3
    +one million kajillion for removing the short names. It makes so much more sense like that. I wholeheartedly believe that would drastically reduce the number of bad flags – Diminutive Colossus Nov 23 '16 at 2:03
  • 1
    It is also worth noting that excessive self promotion is different from spam. If a user keeps adding his link to several answers but if the link is helpful and in someway related to the question, then its a case of excessive self promotion. We handle those differently. – TheLostMind Nov 23 '16 at 5:53
  • 8
    I want to echo @DJMcMayhem 's sentiment here. I have over 2000 network wide spam flags myself and lots of times, the service isn't advertised via URL, maybe just name, maybe just email adress. That would add a devastating amount of extra paperwork to something that is currently simple and should remain simple. – Magisch Nov 23 '16 at 7:49
  • 5
    Hmya, you stop users from using the wrong flag by educating them. Conveniently done by rejecting their flag. Sure, a job that's never done, tedious. Resembles having 12,000 questions every day on a web site that already has 13 million of them, that job seems to never get done either. – Uphill Luge Nov 23 '16 at 11:11
  • 1
    ... And shuffle the options every time you open the dialog. – Won't Nov 23 '16 at 13:53
  • 2
    I think this be migrated to meta.SE - I just realised there is nothing SO'ish about this, except for the volume of flags. Does anyone think otherwise? – Bohemian Nov 23 '16 at 15:51
  • 1
    @brandonscript spam is treated differently in terms of penalizing the spamming account, so expanding it to include gibberish wouldn't be sensible. Gibberish falls under VLQ when it doesn't include advertising. – hobbs Nov 23 '16 at 18:27
45

The moderator flag queue is biased towards bad spam flags.

Posts that are deleted by 6 spam flags from the community have their flags marked helpful. This removes the flags from the moderator queue. (I've just tested that this is the case.)

That biases the queue towards unhelpful spam flags.

In reality, there are probably far more helpful spam flags than there are unhelpful spam flags, you're just not seeing them because they're removed from the moderator queue. Given that, I'm not sure it's necessary to do anything extra to spam flags - it would increase the amount users have to do to nuke actual spam because of a likely small proportion of bad flags.

I'm also not sure that requiring a reason would actually make people think - you'll just end up with lots of reasons that just say 'spam'.

  • Though I believe that spam flags on posts that accumulated 6 of them are still reviewed (albeit at a lower priority) by moderators, to prevent, for example, using sockpuppets to delete posts and deliver massive rep penalties to users that they don't like. – Stephen Leppik Nov 23 '16 at 0:56
  • @StephenLeppik The only way for moderators to review finished spam flags is to go and look at the post - which is unlikely, because it won't appear anywhere in the queue. – ArtOfCode Nov 23 '16 at 1:00
  • 4
    I sort of doubt that; the idea of a user making 6 accounts so that they can delete any post on a whim spam/offensive flags sounds exactly like something that a sufficiently butthurt/annoyed user would do. – Stephen Leppik Nov 23 '16 at 1:07
  • 5
    @StephenLeppik I am a moderator. There is nowhere that a destroyed spam-flagged post will show up in the moderators' queue; it'll only show up in the lists of locked/deleted posts, etc. – ArtOfCode Nov 23 '16 at 1:10
  • 2
    @StephenLeppik there is a minimum rep to flag, so someone can't just create 6 accounts and start spam deleting anything he wants. – psubsee2003 Nov 23 '16 at 12:38
  • 1
    @psubsee2003: That's only 15 rep, though, so it's a really low bar to reach. – user2357112 Nov 23 '16 at 18:23
17

Stack Exchange has very strict definition of spam. I am an experienced flagger with several thousands helpful flags cast over few years at multiple sites but even to me it took quite a bit of effort and messing on meta and in Charcoal HQ to realise how strict it is.

(I am not happy with that but oh well. I guess it will be that way no matter how I feel.)

From what you describe it looks like many flaggers feel like I do. And based on my experience I can tell that this is unlikely to stop, you will keep getting this kind of troublesome spam flags no matter what.

And I can tell that most flaggers will be unhappy with your declines (I won't anymore but this is exception not a rule). Getting a bland "no evidence" decline for a blatant job ad or research survey that didn't qualify only because poster didn't add link to job site / survey service, give me a break. Give me a f#cking break!


Given above, I think that best we can do is just decrease tension. I think having a dedicated canned decline reason would help in that. (Canned because there are and there will be many flags of that kind.) Reason text could be something like that:

We are sorry but Stack Exchange has a very strict definition of spam and post you flagged doesn't qualify. Strict handling of spam flags is necessary to maintain our automated anti-spam system. In the future please use spam flags only on posts that strictly match criteria used at Stack Exchange.

To make things really convenient this decline dialog would also better have something like a checkbox "delete the flagged post". Because let's face it, vast majority of these posts are delete worthy even though they can't feed into anti-spam system.

  • 5
    This is the best approach I've heard so far. It addresses the education aspect because you accurately explain why the flag was declined and provide background on the actual defintion. – psubsee2003 Nov 23 '16 at 12:17
  • 3
    Related: How is this "question" not spam? – Josh Caswell Nov 23 '16 at 14:29
  • 4
    I don't always agree with you, gnat, but this answer literally takes the words right out of my mouth, and offers even more insight than I would have been able to come up with. The strict definition of spam that we employ is, in my opinion, nonsense, but certainly an endless source of friction. Most of the justification for it seems like the harsh penalties that we automatically apply to validated spam flags, but that's a classic case of implementation details creeping into the interface. If something looks like spam, it should be flagged as such. Doling out penalties is someone else's problem. – Cody Gray Nov 23 '16 at 18:53
  • 1
    Unfortunately, even the "unsolicited advertisement" definition is insufficient. Josh has already pointed to a case where something that was both unsolicited (no one asked for it, nor was it relevant to anything) and an advertisement (of a product launch), yet when it was flagged as spam, the flag was marked as invalid. I've already thrown my fit in comments over there, but this system is horribly broken, and tweaking the definition is not going to solve the problem. It is never justifiable to spend this much time worry about obvious crap. – Cody Gray Nov 23 '16 at 18:56
  • 1
    I forgot about this one: meta.stackoverflow.com/questions/336742/… Another great example of condoning a crap post because it doesn't exactly fit the perfect, precisely codified definition that has accreted around the flag. – Josh Caswell Dec 11 '16 at 22:06
16

I have to disagree vehemently with the proposal of requiring an additional reason.

Since I joined this site, I've accumulated over 645 helpful spam flags on SO alone, only 3 of those have been declined. I suspect most heavy users of the spam flag have very similar ratios.

The flag queue seems to show only an excerpt of spam flags (those that don't get deleted instantly via 6 flags / post) and is thus weighed towards the more controversial spam flags.

Adding an extra step to this process would just make it slower to flag spam, and it would also serve no purpose other then increasing the time it takes to flag something as spam.

A valid spam flag is very obvious. For anything that you can't identify as spam quickly, a moderator flag is more appropiate anyways. So in essence by requiring extra reasoning you are punishing everyone who casts valid spam flags for no reason other then to discourage some people who misuse them.

14

I think the reason you're seeing this could be because there seems to be a second internet definition of spam, and that is "posting a lot of not-useful stuff in a short amount of time" (example of a site that uses this definition).

One possible solution would be to have a dialogue when users have a decent number of declined spam flags, saying something like this:

"Spam" has a specific definition on {{site}}—consider other flags such as "Not an Answer", "Rude or Abusive", or "Very Low Quality" before flagging.

Remember: the text defining it is smaller and lower-contrast; you can definitely count on people not reading it and saying, "SPAM—that sounds like the flag that I'm looking for!"

  • 8
    Maybe we should get rid of the word "Spam" and just name the radio button "Exists only to promote a product or service"!? – Bohemian Nov 22 '16 at 23:59
  • @Bohemian maybe, but that's more the gist of the other answer. – Stephen Leppik Nov 23 '16 at 0:00
  • We need both spam and undesirable content. Separate them maybe? Or combine them? ¯_(ツ)_/¯ – brandonscript Nov 23 '16 at 18:00
  • Maybe we should worry less about the wording and precise constitutional meaning of the flag description and worry more about handling clearly bad content. – Josh Caswell Nov 27 '16 at 16:35
5

What if we renamed the spam flag?

Users flag posts as spam because they think those posts are spam. They think "spam" is just another word for "junk." So instead of trying to convince them "that's not spam here," why don't we use a different name for unsolicited advertising?

  • 2
    But that why God invented downvoting. Rubbish does not require moderator intervention. – Bohemian Nov 22 '16 at 23:58
  • 1
    @Bohemian I'm not sure what you mean. – NobodyNada Nov 23 '16 at 0:00
  • 4
    I don't particularly like this idea, it should be named spam in my opinion. I would downvote, but at least proposing the idea seems constructive so instead I left a slightly dissenting comment ;) -0 – Travis J Nov 23 '16 at 0:03
5

First of all, some research might be needed in declined/helpful/auto-helpful spam-flags vs user's moderation experience (e.g. total flags, perhaps rep, etc). I am guessing that experienced users are less likely to get declined spam flags.


A new (or relatively new) user, will definitely not go through pages of SO policy, and will probably not bother much with the explanation below the word "spam":

spam
Exists only to promote a product or service, does not disclose the author's affiliation.

They will instead use their own definition of spam. Trying to educate them on spam's definition on SO would be ineffective for the reasons mentioned above, therefor we should remove completely the word from the flagging process:

Exists only to promote a product or service, does not disclose the author's affiliation.

Other short names should be kept though (perhaps some could be changed as well). It's much easier to have abbreviations for flags when discussing in Meta.


Additionally, the flagging layout could be changed:

  • Descending order. Flag options should be in a "most likely" to "less likely" ordering. That way if someone semi-randomly flags something the odds of getting it wrong decrease. Or from "less side_effects" to "most side_effects", where side_effects == moderation_work_involved or community_work_involved.
  • Coloring. E.g. you can use small colored boxes or letters or whatever, to indicate that the spam/rude+offensive flag is red and scary, and should be used with caution, while the "off-topic" flag is yellow and not so scary.
  • Spacing. Keeping spam/rude and custom flags slightly away from the rest could help avoid accidental flagging with adjacent flags.

Making spam flagging harder by demanding links is a bad idea as mentioned in other answers.

3

While I believe education is the best form of prevention, have we ever considered that maybe the spam flag is used somewhat as a "catch-all" (gibberish/VLQ/rants/etc.) flag by inexperienced users because it is the first flag on the list? IIRC, the point to the ordering as it exists today is to make it as easy to find the spam flag as possible, but possibly that is part of the problem since it is so easy to find, and the term "spam" is misleading to users unfamiliar with the specfic defintion on Stack Exchange sites.

You would have to test this with some A/B testing, but what about reordering the flag window? Move Spam to the middle somewhere, put Offensive above it, and possibly even the close flag buttons above it. Users may use it less because they find a more fitting one first.

Flag dialog reordered according to this suggestion

After the A/B testing, you could then look into whether or not:

  • Spam flagging increased or decreased
  • The moderators rejected more or fewer Spam flags
  • Flagging as a whole suffered (fewer flags, more rejections, etc)

If spam flagging did not suffer, with mod rejected spam flags (and aged away spam flags) decreasing, you could probably try this for an extended period.

  • This is by design, I have to dig it up (I think it was Shog that wrote that), but the spam/rude flags are there because SE wants you to consider those first. – Braiam Nov 23 '16 at 15:55
  • 1
    @Braiam I know it is by design, and I think I remember the post you mentioned, or at least some discussion on that topic. I'm offering it as an alternative to address the concern raised here. A/B testing would certainly show whether (1) correct use of the SPAM flag decreased and (2) incorrect use of the SPAM flag decreased, and if (3) flagging overall suffered or benefited from the change. If you see any negative impact, you leave it as-is – psubsee2003 Nov 23 '16 at 16:01

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .