I have been surprised by the questions featured as the "greatest hits" in the Stack Overflow newsletter.


January 3


November 29

November 15

November 1

September 20

September 13

August 30

August 23

August 16

Do we really want the newsletter to feature questions like these? I understand that they exist for historical reasons, but given that moderators would probably close any new questions like these without hesitation I don't think that we should be actively promoting them, either.

Would it be possible for someone to give the newsletter a quick, manual check before it goes out, or is Stack Exchange not really set up for that? How difficult would it be to catch off-topic posts like these before the newsletter is actually sent?

Interestingly enough, I don't notice an overwhelming number of these "inappropriate" questions on Stack Overflow's Greatest Hits page, which is probably used to populate the newsletter. Any idea why?


4 Answers 4


Newsletters now use the same query to fetch greatest hits as their respective sites' Greatest Hits pages. This version of the query takes several extra factors into account, such as post feedback and number of views, resulting in higher-quality "hits".

Also, from now on, a newsletter will only feature greatest hits if the site is at least three months old.


I think those questions should just be put in protected status, and then have the algorithm that chooses "greatest hits" ignore protected posts (along with closed ones).

  • 14
    Lots of great, appropriate questions are protected. I don't think protection is a valid flag, but I agree with the idea that such questions should be flagged and prevented from being placed in the newsletters. I think they should be closed and locked, and those questions which are closed should be excluded from the newsletter.
    – Pollyanna
    Commented Aug 17, 2011 at 19:20
  • That's a good compromise, at least. I would assume that it already ignores closed questions, though -- wouldn't it be better to just close them, if we don't want to delete outright? One would think the ability to add new answers on old off-topic questions is not important, and you gain by actually marking them off-topic so you have examples.
    – user154510
    Commented Aug 17, 2011 at 19:22
  • @Adam, I wasn't sure we really wanted to lock all of them, for example, maybe a new Python IDE tool will come out, and someone should add it. Commented Aug 17, 2011 at 19:27
  • 4
    @Lance if the question is unacceptable, we shouldn't be leaving it open for new posts. IMO, of course...
    – Pollyanna
    Commented Aug 17, 2011 at 19:33
  • 3
    @Adam I don't think that you could realistically find all of the questions that need to be prevented from being placed in the newsletters. Perhaps it would just be best for someone to preview the newsletter before it goes out and take action if any questions require it. Commented Aug 17, 2011 at 19:35
  • @Adam, well, it's is acceptable enough to not delete or migrate, then it might be worthy of more answers. I think it's case-by-case, so for the Python IDE a new answer might be useful, while I think the "Next Big Thing" is completely useless. Commented Aug 17, 2011 at 19:36
  • @Chris, I was thinking that these questions all needed status reviews anyway, so this would be more motivation to set those statuses right. Commented Aug 17, 2011 at 19:37
  • @Lance How would you go about finding all of the questions that need status reviews? I'm sure that there are thousands where these came from. Commented Aug 17, 2011 at 20:02
  • @Chris, most of the ones that matter have crazy upvotes and views, so just start at the top and work your way down. Though SO could make it easier to do by labeling protected and/or locked questions on the questions pages with a [protected] and/or [locked]. Commented Aug 17, 2011 at 20:05
  • 6
    we are adding some metadata annotations for posts soon, so this approach does make sense @adam Commented Aug 17, 2011 at 23:27
  • 3
    @Jeff I didn't want to suggest meta data (Hey guys, all we need to do is add another table or two to the database, and modify a few existing tables! Guys...? Hello?) but it's probably the correct solution to this particular problem, and will allow other meta data to be added later for other purposes.
    – Pollyanna
    Commented Aug 18, 2011 at 4:36

This is exactly the problem with keeping these questions around. Broken windows encourage more damage (more questions of the same ilk) even when not publicized in the newsletter. Though the newsletter exacerbates the problem or brings it further to light, it's not a problem with the newsletter; having these questions around is the problem.

I would like to see these questions killed with fire, historical or not.

  • 16
    Respectfully disagree: these questions were tolerated at the time, and people put effort into asking and answering them. Nuking that content just because Things Have Changed would be disrespectful to these people and prevent us from learning from the past, which is always a bad idea IMHO. Changing the rules during the game is cheating, and I don't think you'd keep the same reasoning if, say, questions about rooting became discouraged on android and users started to point their pitchforks and torches at your content. Commented Aug 17, 2011 at 19:10
  • 5
    Another interesting wrinkle that this issue brings to light is that these questions remain very popular: presumably the "greatest hits" section of the newsletter is automatically generated based on page views. Would it be prudent to delete this content even though it brings traffic—and by extension, new users and content—to Stack Overflow? Commented Aug 17, 2011 at 19:14
  • 4
    @Frédéric It wouldn't prevent learning at all, the learning has already occurred and is recorded in policy/discussions. It's kind of like saying you need to keep a list of all your past relationships on your Facebook profile in order to learn from them. If we had a good reason for eliminating a subset of Android questions, I would eliminate them. I recently closed and deleted all the now-offtopic questions over a year old on Android, and I always close old bad questions when I come across them. It's not a game, and iteratively developing better rules is important to developing a great site.
    – user154510
    Commented Aug 17, 2011 at 19:15
  • Those questions are more the kind people just throw things onto the fire rather than any concerted effort or thought. No real boner for actual problem solving by people sitting around drinking coffee. Different than when subset of content is off topic. @fre
    – random
    Commented Aug 17, 2011 at 19:21
  • 1
    @Matthew, good point about learning, even though I still think that linking to real questions from policies/discussions works better than enumerating characteristics of deleted content. But how about the hard work of the users who asked and answered those questions? Should it really be ripped from them? Commented Aug 17, 2011 at 19:23
  • 4
    Could we just have a better filter on which questions are chosen. Perhaps favoring questions which are less than 6 months old?
    – M. Tibbits
    Commented Aug 23, 2011 at 17:35
  • Showing these questions under the "Historical Relevance Only" banner seems like the best compromise. It's a shame we don't do that anymore. Commented Nov 29, 2011 at 18:58

Given that the policies for on-topic questions may alter as new (and intriguing) sites are added, I would suggest that these 'Greatest Hits' selections have a filter to feature only high view questions which were asked less than say six months ago. I would also suggest avoiding questions which have been closed & re-opened more than once. As for the previously featured questions:

I believe the current modus operandi of Stack Exchange Moderators is to not migrate a question which pre-dates the existence of a site where it would now be considered on-topic, except on a highly selective basis. Within the past year, we now have added sites where some of the above questions would be on-topic (the links below are to their Area 51 proposals with launch dates):

and many more. In contrast, there have been about 200 questions, asked within the past six months and have more than 5,000 views many of which I think would make excellent candidates for the 'Greatest Hits' category.

Here's one example with 47,277 views & 259 upvotes where Sir Skeet-a-lot has already earned 599 up-votes on a question which was asked on July 27th, 2011 at 8:15pm.

You must log in to answer this question.

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