> > A machine-readable value would make it easier to retrieve them with
> > high recall, for example. Maybe there are different LCSH (or other)
> > URI's, such as http://id.loc.gov/authorities/sh20something that could
> > be helpful for such kinds of dates.
> But why on earth would you need to search for "first Thursday of
I don't think it is a matter of searching, I think Saašha is suggesting that there is value in being able to represent it in authoritative form, so for example, let's say there is a controlled vocabulary, including the term 'firstThursdayOfFebruary'. Then a machine encountering this term would know what it means, whereas if the machine encountered '1stThursInFebruary' it wouldn't.
However, I don't see what value there is in the machine understanding "first Thursday of February". There may be applications where this is useful, but I would think that for our purposes a human-readable expression is sufficient. I think we would need Ed to elaborate the requirement.
I want to point out further, if the suggestion is that there would be authority records for these expressions analogous to those I described in a different thread on temporals: I have no idea where they would come from but I can pretty much guarantee not from LC.