[uf-discuss] RE: Data drift vs. ISO dates WAS Benefits of Microformats

[MichaelMD]

> Is it a valid conclusion then that placing a date marker around a
> text-based date is a suitable solution, leaving the parsing of valid
> data to a parser of some nature?
> 

no...

People often write dates that are ambiguous even to other humans, let
alone machines.

Nobody is going to convince me that something like 12/09/08 could even
be considered human readable let alone machine readable.
Someone is Europe or Australia might see it as 12th September 2008
someone in USA might see it as December 9th, 2008

- and - 

using localisation cannot be considered reliable at the the server end
of things.
(the only place where it might be reliable in the real world is for a
translated display in a client based on user preferences)

I've often seen sites here in Australia showing dates the American way
because that is the default settings in their CMS
(and what about people posting on a site that isn't their own site? -
perhaps even in a different country?).

Humans are pretty good at guessing what date format is being used
because they take into account a lot of other factors like the type of
event, what date ranges such an event is more likely to be in,the type
of page they are looking at and memories of similar pages they have seen
before, and even then they can sometimes still get it wrong.

I'm not afraid to show humans something like 2008-09-12.
At least both people and machines usually get it right without needing
to know something about its context.

People who have never seen dates written like 2008-09-12 still usually
work it out correctly (even if they think it looks strange)

In my opinion its only when you start including the time and timezone in
an ISO date or using less common variations of it that it starts to get
confusing for humans (so I think there is a case for marking up times
and timezones separately to the dates.)

Regarding: screen readers - using words rather than numbers for the
month (with a full four digit year) may seem like a good option until
you start to consider how to handle those names in hundreds of different
languages.

...perhaps if there was a way to define such names in the actual markup?

...but specifying formats, languages, etc in the markup would make it
more complicated for authors.
Given how often I've seen 'Z' misused in timezones I think making it
simple for authoring is something worth considering too.