Closing as dupe is one of the harder things high-rep users and moderators do. It involves a lot of judgement in order to be done right.
We have to determine how close of a duplicate a question is. Sometimes its a 100% duplication of the original question (noobs). This is the easy case--close as dupe and merge. Sometimes a question kinda sorta covers the same area as a previous question, and we should leave it alone. There is no way to determine where this line is; it is completely arbitrary.
In addition, there is the desire, expressed by many people [citation needed], to codify the best particular question on an identifiable subject as "The Question" of which all other similar questions are duplicates.
I would suggest that we can combine both of these into a single change to StackExchange. I suggest that the number of times a question is used as a "close as duplicate" reason be kept and displayed when the closs-as-dupe dialog is shown.
This is not a perfect solution. While chances are that if an answer is suggested as the "best duplicate" it actually is, there is still a chance that it is suggested in error. Also, just because a question has been chosen as the "best duplicate" in the past, it does not mean that it actually is the best question for its subject. And, lastly, there is no guarantee that the true, current canonical answer is known by anyone who votes to close as dupe, thus resulting in a less than optimal selection.
However, I do believe that over time, as this statistic is gathered, it would allow for a number of improvements to the process:
- For a common set of tags, canonical question candidates can be reviewed and ranked by close count. Users with high enough rep would have the option to merge "best duplicate" choices together, replacing the "best duplicate" of questions closed as duplicate of A with the better canonical question B.
- When voting to close as duplicate, choices can be automatically provided to users (possibly of a certain rep to avoid lazy choices) based on tags and close count
- "Best duplicate" choices which have a low close count can be used as a heuristic to determine if the best choice has been made; questionable choices can be listed on the Tools page so high-rep users can review these.
At some point, a [Canonical] tag may be implemented and bestowed on questions with a very high close count. Questions with this tag would be great candidates for improvement over time through user edits. This tag could then be used as a better alternative to the Faq tab, providing a true place where the cream floats to the top.
Dunno. Something I thought about in between builds.