Oh I'm saying the reverse, I think. Future timestamp for "time at location" is impossible because you don't know within which time zone a location will be in the future. But for the right time zone database structure you can have indeterminate time zones - so you can know future timestamps for a time zone, but you don't know if any particular location (or any location at all) is using that time zone in the future.