[tz] Stability versus correctness

[Paul Eggert]

Lester Caine wrote:
> I had thought there was an agreement that accurate historic data would
> be retained when the previous cull was proposed even though the 1970
> limit is used?

Yes, we want to keep data that we have a reasonable amount of confidence 
in.  The recent proposal attempted to discard data for which the only 
source (Shanks) has proven to be very unreliable in the past, when 
applied to out-of-the-way locations before 1970.  (I clearly went too 
far in that proposal, and plan to fix that soon.)  The general rule of 
thumb Shanks used seems to have been that when there was a gap in his 
database, he made something up.  The tz database does not need to 
perpetuate this tradition everywhere.