Do we need Timezones

Created by: Lester Caine, Last modification: 08 Aug 2014 (15:25 UTC)

Some links posted by Paul Eggert on the tz list ...

In Vox yesterday Matthew Yglesias argued that time zones ought to be abolished, writing "They were a good idea at the time, but in the modern world they cause more trouble than they are worth." It's tempting to agree. See: Yglesias M. The case against time zones: They're impractical & outdated. Vox 2014-08-05. Yglesias's underlying justification is essentially the same one that McCarthy and Klepczynski used in their proposal to discontinue leap seconds; see Steve Allen's summary of the resulting controversy in <>.

It seems appropriate to comment on these proposals as both seem to miss the fact that abolition whould simply mean some has to be used in their place. I posted some comments on the tz list, but I'm copying them here so I can expand on them as things evolve.

Abolish Timezones

It seems to miss one fundamental fact ...
That is currently how we work anyway. Any decent computer system is running everything on UTC and I've run the information systems that way for 20+ years. Timezones ARE the local schedules and do vary according to local dictates. They just provide a convenient way that those of us who need to check a train time across Europe can have some idea to the local time. If you scrap them then you need some means of managing a central repository of data on the replacement anyway? It's just called something else?

Since this link was posted I've had some further thoughts about this

Discontinue leap seconds

Now that is a different matter altogether. Just because some idiot decided that since time was based on seconds then we should count time from some arbitrary point was the mistake here. I've said before that I work with 'days' as the base unit and time is then fraction of a day ( TIMESTAMP on Firebird ). If one day just happens to be a second longer I can observe or ignore that fact. My own historic data only needs to align with midday Greenwich hunderds of years ago.  

We only need the tz database because we are working with a fixed clock. What that clock is locked to is nominally noon over Greenwich and because the rotation of the earth around the sun is not a constant, it's only an approximation anyway. That the approximation currently used only drifts a second every so often is probably testament to the achievements of the clock-makers of the past, and do we really know how today accurate that will be in 1000 years time? Just as we have leap years for dates, we need leap some-things for time and for the next 100 years seconds are as good as anything? It's only now we have a stable time source that we see the problem ... and that is perhaps because someone got the duration of a second wrong anyway? Adding a few more cycles to the definition of a second would be the sensible fix?