I'm not a big fan of the "problem" filter for titles. People have been discussing how to get around it. I think users with enough reputation should be able to use the word. Others think it should be removed entirely. If the first request was declined, then here's another one.
I know you're probably getting tired of it and decided to drop the science hammer on it. I'll do the same from a logical and statistical point of view.
The filter has been introduced on Sept. 28, at least that's when the last question with "problem" in the title was posted on Super User. Ever since then, I've seen people increasingly use synonyms to circumvent the filter, but leave the title ambiguous as-is.
That's my main issue there. If you just restrict the usage, people will either:
- Write a good title instead.
- Use a synonym and write an equally bad title (dubbed "benign").
Now you have a good outcome, which is nice, or you have a "benign" outcome. Still, this outcome is even worse than having "Problem" in the title. Because it can't be tracked. Before, every user could easily look for "Problem" and see where the problematic questions (…) were. We could issue a query that allowed us to look for bad posts. Now, we simply can't do that anymore, because the title might use a synonym or other bad wording.
I know the developers have a flag for a post that triggered the filter, but we reviewers don't see this. We can never find out which question triggered the filter, so we have to keep our fingers crossed that people actually write a better title, which they don't do, according to the "efficacy" stats. All these stats tell me is that we've now lost a way of finding bad quality content, because 62% of those title changes were "benign".
This is why it would make sense to just automatically flag posts that use "problem" in the title with the low quality flag, so they show up in the "low quality" queue, and in the flag queue. You could still restrict new users from using it, but because they triggered the filter, they'd get the automatic flag.
And now, to support this, here are some stats from Super User:
Jul. 15, 2009 → Sept. 28, 2011:
410 questions with "issue" in the title. An average of 0.51 per day.Sept. 28, 2011 → Dec. 09, 2011:
64 questions with "issue" in the title. An average of 0.88 per day.
Do you see the significant increase here?
Update: As @sth has pointed out, of course this is also explained by a general traffic increase on the site. This is definitely important to consider, but here's another smaller time frame, not taking into account the two years before:
- Jul. 1, 2011 → Sept. 28, 2011:
56 questions with "issue" in the title. An average of 0.63 per day.
This is just one synonym I can think of. And these are just some simple stats. This is not the main point of this post. The main argument is: There are probably many other patterns of escaping the filter too, but I have no idea on how we editors/reviewers could track this.
Therefore, just flag failed attempts.
I [was] too young here to know the importance of the issue it was supposed to solve and how efficiently it ended up solving it
, but Kevin's post looks right on the money. The filter indeed seems to be successful, because there is only a minority of users who will try to bypass it without improving their question's title. Sure, most of time (62%) their improvement is a simple workaround, but that still helps search. On the other hand, auto-flagging these 62% would add to the mod team's burden. We should ask them.helps search
, I mean that, for instance, everyone who typeswindows help system
orOpenGL texture problem on X11
into google can see the higher-quality questions about the winhelp system or OpenGL texturing on X11 first, instead of the myriads of questions that might haveOpenGL problem
orhelp windows
in their title, if that was allowed.