On Tue, 3 Oct 2017, Owen Jacobson wrote: > The “technical domain” precedent has never, to my knowledge, been overturned. > The tooling I’m working on right now treats the Date: header as authoritative > by > default, on the presumption that (a) we can amend a recorded message that’s > got > the wrong date, and (b) people won’t lie in their headers very often, but this > is how that assumption can fail. I also use Date: when figuring out event > times > for reports.
So the TDOC precedent was set long ago in a very different ruleset. I've always been of the opinion that we should go with the Date: header, and the knowledge that it can be forged for tiny advantage be dealt with by some kind of crime (e.g. "if the date-headers show discrepancy, and that discrepancy would gain a material advantage...") and say "not messing with the headers" is the same as "not playing as two people from different accounts": strong social pressure not to do so, but not worth our time to be paranoid about. The convenience of the Date header all around is worth accepting that it's *possible* to fake.
