83

Spam posts, without proper context, especially those tricky ones, can be extremely difficult to assess; and even when you successfully identify it as "spam" the review queue doesn't allow you to flag it as such, hence there isn't a correct action in the Low Quality Review Queue when seeing spam other than going outside the queue and flagging it as such, so this shouldn't be used as an audit test.

Audits are meant to teach reviewers how to correctly review posts, but in this case the correct action is doing nothing with the review, so as audit its usefulness is null.

Can we please not use spam-deleted posts as audits in the Low Quality Review Queue?

  • 1
    actually, lot of spam is quite obvious. Also, we really want people to be well trained in detecting spam. – John Dvorak Aug 27 '14 at 6:56
  • here's compelling reasoning in favor of using them – gnat Aug 27 '14 at 7:41
  • @gnat and my answer to that compelling reason – Braiam Aug 27 '14 at 11:08
  • @Braiam I saw this; your considerations also make sense - would be great to have this sorted out here (I for one still can't make up my mind) – gnat Aug 27 '14 at 11:11
  • 10
    @JanDvorak yeah, but the LQRQ is not the correct queue to teach that, since to pass the audit they need to do the incorrect action, mark it for deletion instead of flagging them as spam. – Braiam Aug 27 '14 at 11:11
  • 3
    actually, wouldn't it be better to add a flag option to the LQRQ? – John Dvorak Aug 27 '14 at 18:12
  • 2
    @JanDvorak poke Shog about it – Braiam Aug 27 '14 at 18:18
  • Apparently offensive/abusive flagged posts are also used. – Braiam Jun 5 '17 at 18:34
  • 1
    And they are not used anymore – Braiam Feb 1 at 15:28
46

I fully support this proposal. The main reasons why it makes sense to me are:

  1. Spam does not belong in the LQ review queue in the first place. As far as I understand, spam flags are handled by diamond mods, and do not send posts to a review queue. Therefore, spam vs. not spam decisions are not a normal part of working the LQ review queue.
  2. Related to the previous item, there is no good way to handle spam answers directly from the LQ review queue. They should be flagged as spam, which is not supported directly from the review queue interface.
  3. The way spam is defined on the SE sites, it is often not possible to apply the criteria when looking at one post in isolation. It should be possible to make decisions on review queue items without extensive background research.

The biggest problem for me is item 3 from this list. Based on various discussions and answers I have read on MSE and MSO, my interpretation is that the decision if a reference/link is spam or not has a lot to do with:

  1. Is somebody promoting their own work or business?
  2. Is it repetitive?

I don't really have a good way to evaluate either of these two criteria:

  • On item 1, most users do not use their real name, so I couldn't research it even if I tried. The best I can reasonably do is look at the "About Me" box in their user profile, and see if they disclose any association with the product/site in question. But that's often blank or not useful.
  • On item 2, there's no good way to evaluate that either. Even with 10k privileges, I can't see deleted posts in the user profile. So I can't tell if they might have posted the same thing before, and it was deleted. Also, users can do LQ reviews before they reach the 10k level, so they can't see deleted posts at all.

As a case in point, let's try a little exercise. Can you tell if the following post is spam, without reading ahead to the answer below?

Twilio has the best pricings I've seen around, SMS messages around $0.01.

where the product name is a link to pricing information for the product on a commercial web site.

Answer: Not spam. Came back as disputed when I flagged it. No reasoning given, but my interpretation is that it's based on the criteria above. This was indeed the only time the user posted about this service. And there is no information I could find suggesting that the user might be associated with the company. So not considering it spam was most likely correct.

On the other hand, if somebody includes a link to some jQuery plugin in an answer, I might have to recognize it as spam when it shows up as an audit. The difference to the above is most likely that a user posted answers with those links all over the place, and a moderator deleted them. But just from looking at the answer, it looks much less like spam than the example I quoted above.

Based on this, I do not think spam should be used for audits in the LQ or Close review queues.

Disclosure: You could consider me biased because I got myself two review bans after stumbling over this type of audit, combined with one or two clearly bogus audits of different types.

You must log in to answer this question.

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