As the volume of moderator-only closures is quite important for this topic, I would like to share an observation that Cascabel and I made a few years ago, when playing with the SE data dump for similar purposes.
I think that there should be a differentiation between moderators closing for the "duplicate" reason (and maybe also for migration) on one side, and all other closing reasons on another side. There are two reasons for the division:
- closures as a duplicate are usually clear-cut and nobody disputes themrarely disputed.
- when the moderators are among the oldest users of the site, it is normal that they best remember having come across an older question of the same topic. This makes them naturally the ones who close the most duplicates.
- closing as a duplicate is less likely to reduce the information available on the site. If somebody sees a question they want to answer and the question is a duplicate, they can (and ideally, should) write up their answer under the duplicate target instead. This results in both more content than other closures, and less frustration among knowledgeable users.
On Cooking, we had been asking ourselves if we moderators overrule our users too much with hammers, and wanted to take a look at how much we close unilaterally. At first we were surprised at the high mod-closure rate, but once we removed the duplicate closures, we saw large changes in the numbers.