This is a hopefully less complex alternative to this old request, directed mostly at the same problem. This request was inspired from Jeff Atwood's comment on this answer, quoted for quick reference:
[...] mass retagging should be done by moderators using tag rename and tag merge tools anyway. – Jeff Atwood♦
This mostly works, but moderators are lacking the necessary options to handle the largest (and consequently arguably most dangerous to leave alone) mass retag jobs. These are the jobs handling enormous, hand-sorted disambiguations, such as recent discussions for Stack Overflow on scope, float, and table. They have to be manually separated to organize which questions get which tags, a step which can be done by anyone in the community. But we can't just use our tools to be done with it in one fell swoop, so we have to do it the long-and-slow process of manual retagging. Which only leaves more time for the bad tag to continue to accumulate even more entries. Stack Overflow's speed makes the front-page damage much lower, but also means a higher accumulation rate and higher volume in the first place. Not a good combination.
I propose that moderators be given the option to do silent retagging in selective batches by specifying a list of posts. It should follow all the rules for our normal moderator retags, and should likewise be documented in our moderator history. The exception is that instead of just asking for the two tags in question, it also requires a list of QuestionIds or URLs that correspond to every question that needs the specific retag direction.
Since this is primarily for when there are tons of questions, I propose a minimum of 15 questions for any individual retag instance to function. Disambiguation jobs under 15 tags can usually be safely done by hand without any interference, and this allows the silent retagging system to continue its existence as solely for bulk jobs. While I trust every user that we have on the network with a ♦, we do not possess the ability to singularly retag any question silently (which I feel is not a problem). As such, I intend that this minimum post requirement respects the standards for vetting retags while still allowing proper handling of huge jobs.
It still requires all of the grunt work and research that we already need for disambiguation, and I'm fine with that because I feel the long research is a requirement for a successful disambiguation. With this, though, once that research is complete we will be able to clear out the most difficult portions with greater ease and haste.
