I noted that the old duplicates and the new duplicates have different formatting.

The questions that are now getting closed as duplicate get a bluish plaque on top saying "This question already has an answer here: ..." But the questions that were closed as duplicate long time ago have a grey plaque on top saying "Possible duplicate: ..."

If you look at all the duplicates sorted by date, then the borderline between these two formats seems to pass around January 2013, currently page 19. On page 20 duplicates are already mostly of the old format (even though it seems to depend on the closing date, not on the posting date).

The old format looks quite obviously inferior: it has strange wording ("possible"?), does not display the number of answers in the linked question, and does not update the title of the linked question if it gets edited (the link seems to be "frozen").

I searched the meta.stackexchange, but could not find any reference to this behaviour. I assume that the format was changed at certain moment, but why are the old duplicates not displayed in the new format? This seems weird. Any explanations?


Here is one of the old duplicates:

If you start editing this post, you will see that it starts with the following text:

> **Possible Duplicate:**  
> [Replacement of NA values for PCA analysis](https://stats.stackexchange.com/questions/35561/replacement-of-na-values-for-pca-analysis)  

<!-- End of automatically inserted text -->

so the link to the dup is actually hard-coded into the question text. Well, I thought if I edit this question (and remove this automatically inserted plaque), it will trigger automatic re-generating of the HTML page and it will get the "new" blue duplicate plaque. This did not happen! I was able to erase the old hard-coded plaque, but the new one did not appear. The post was marked as duplicate without any link to the dup. I edited again and inserted the hard-coded plaque back.

This is weird.

  • 2
    $\begingroup$ Some of the default "messages" (and available closure reasons) have changed several times in my recollection. $\endgroup$
    – Glen_b
    Commented Feb 2, 2015 at 23:52
  • $\begingroup$ No idea. Nice bit of research, though. $\endgroup$ Commented Feb 3, 2015 at 2:29
  • 1
    $\begingroup$ Note that the change is somewhere near the 19th page if you are viewing 50 threads per page. At any rate, we can narrow the change point to somewhere between Feb 2, 2013 & Feb 6, 2013. $\endgroup$ Commented Feb 3, 2015 at 5:04
  • $\begingroup$ What I don't understand, @gung, is how it even can be different. Questions are stored in the database, as well as the information about which questions are duplicate of which ones. Then there is (presumably) a script that queries the database and renders the question's HTML page. Why would this script work differently for questions closed before and after February ~4th, 2013? $\endgroup$
    – amoeba
    Commented Feb 3, 2015 at 12:53
  • $\begingroup$ I don't know too much about HTML, but I wouldn't have thought that everything was stored in a database & pages were recreated every time someone navigated to one. That strikes me as too cumbersome. I'm aware that there is a database & some info is read from it (rep, eg), but I always thought much of the page was preexisting. $\endgroup$ Commented Feb 3, 2015 at 14:32
  • $\begingroup$ @gung, you are right: generating all the pages on the fly would (usually) take too much resources, so big websites never do that; instead pages are "cached", i.e. are generated from the database once in a while and remain stored for some time to be delivered upon the request. However, I would be surprised if they are never regenerated. Actually, there is a way to test it: we can edit one of the "old" duplicates, and see if it changes format; presumably an edit would trigger updating of the cached copy. $\endgroup$
    – amoeba
    Commented Feb 3, 2015 at 14:38
  • $\begingroup$ @gung, I tried doing it, and results surprised me. See the update. $\endgroup$
    – amoeba
    Commented Feb 3, 2015 at 15:43

1 Answer 1


I have finally managed to find some relevant meta.SE threads.

Before February 6, 2013, closing a question as a duplicate caused the duplicate text header to be inserted into the question text. It was static and forever, though it can be manually updated.

On February 6, 2013, this process changed and the duplicates are no longer stored as text in the question itself [...]

Apparently few people think it's worth implementing: these suggestions remain un-implemented and poorly upvoted. Pity.


You must log in to answer this question.

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