When the atomic clocks became avalable, events in their vicinity could be
measured very accurately. With the avalability of GPS now both atomic
precision time (well, about one us) and measurements in dm range are
available with relatively inexpensive receivers.
So why not take a well specified reference point (at the avalability of
sufficiently precise atomic clocks) and counting seconds (using the
SI metric system, which is quite reasonable) from then? I'd like to use
binary clustering, but the decadic number system is also quite adequate,
at least for us mehums. A time stamp made from heterogenous components
like seconds, minutes, days, weeks, months, seasons, years? Switching
seconds? Sweet Jesus.
The GPS coordinate system is less than fortunate, being geocentric. As our
machines are likely to expand into space in the next decades, using a
heliocentric coordinate system makes more sense. Relativistic corrections
are also much more important then, even though current GPS already must
use them to avoid large errors in position fixes.