In the course of a slightly-related discussion on Thurstan's talk page, he said:

We will have to have a discussion of the data model for the dates: I think too few people will have documentation of when they arrived and left residences, so we will want more of a "address was X when daughter married" sort of model. Thurstan (talk) 19:22, November 2, 2014 (UTC)
Too pessimistic? Births etc also have some unknowns, which we handle moderately well. If a "when daughter married" residence is all that is known, a date range can have a "bef" and an "aft". -- Robin Patterson (Talk) 23:58, November 2, 2014 (UTC)
I don't think it is too pessimistic: births and deaths don't have 99% unknowns, but I think residence start-dates and end-dates do. Perhaps we have to give our contributors a choice for each residence of using a time-interval or an event-based representation. Thurstan (talk) 01:00, November 3, 2014 (UTC)
You seem to be saying "No" to the question but otherwise not actually disagreeing. I've now done some more idea-floating at Familypedia:Residences template. It might have been better to use the forum or to put more on its talk page, and we can still do one or both of those. Please give a preliminary opinion on what's there. -- Robin Patterson (Talk) 11:28, November 3, 2014 (UTC)
Sorry I have not been clearer: I don't think your approach is worth the effort, and I can't see myself having anything more to say about it. Thurstan (talk) 21:07, November 3, 2014 (UTC)