In the zone

Mt. Shasta, from a nearby hilltop owned by Coca Cola

Before leaving on a trip to the West Coast, I copied my return flight information onto my Google calendar: SFO to BOS, 12:50 p.m. to 9:30 p.m. Now that I’m nearing the end of my visit to the Mythical State of Jefferson (see local landmark above), I’ve just checked the departure details by calling up the calendar on my cell phone. It tells me the flight departs at 9:50 a.m. and arrives at 6:30 p.m.

It could be worse, of course. As an eastbound traveler all that I risk is wasting three hours at the airport. If I were westbound, I might well miss my flight.

The developers of Google calendar would doubtless argue that automatic time-zone conversion is a feature, not a bug. If the event in question had been a conference call scheduled for 12:50 p.m. Eastern Daylight Time, then 9:50 a.m. Pacific Daylight Time would indeed be the moment to dial in. Or, if I had a series of pills to be taken at fixed intervals, it might be helpful to have the program remind me at the correct times as I wander across continents. But in the case of today’s flight home, the result is just plain wrong.

It’s not only a Google calendar problem. A few weeks ago a friend was entering a schedule of talks into a Drupal web page for a conference that begins November 7, 2010. All the times were mysteriously shifted by an hour. A 1:00 p.m. talk on the input form became a 2:00 p.m. talk on the displayed web page. The key to solving this mystery is knowing that November 7 is the date daylight saving time ends in (most of) the U.S.

I am certainly not the first to encounter such problems. Peter Neumann’s RISKS Digest reports hundreds of computational mishaps involving time zones or daylight saving time, going back over the past 25 years. The issue is known to Google. But what is the right fix?

Some other calendar software offers the option of specifying a time zone for an event. To handle the airline case correctly, the program needs to allow for different zones for the start and the end times. And the Drupal problem suggests we may also need some means of indicating whether or not to adjust for daylight saving time. It gets very messy. Note that an event scheduled for 1:30 a.m. on November 7, 2010, will happen twice. An event at 2:30 a.m. on March 13, 2011, will never take place.

For years my own makeshift solution to these complexities was to live in a single time zone, no matter where I was. I carried a laptop, but I never changed its time-zone setting, even when I was away from home for weeks or months. When conversion was needed, it happened in my head. Even now the Google calendar display on my laptop gives the correct departure time from SFO, because the laptop doesn’t know that it ever left Boston. But in our new world of location-aware devices, pretending to stay home is no longer an option.

I begin to wonder if the whole railroad-age concept of time zones hasn’t outlived its usefulness. But I haven’t time just now to consider the alternatives. Right now it’s time to leave for the airport. I think.

This entry was posted in computing, modern life.

6 Responses to In the zone

  1. John Cowan says:

    I begin to wonder if the whole railroad-age concept of time zones hasn’t outlived its usefulness.

    Only if you think the concept of working in the daytime and sleeping at night has lost its usefulness. Our bodily cycles are still synchronized by the Sun, despite working in windowless offices and staying up late with electric lights; even in a large city like New York, only about 6% of the population works the graveyard shift.

    At present the worst case is the city of Kashi (Kashgar) in extreme western China, which is on Beijing time (UTC+8) even though its solar time is more like UTC+5; it’s at about the same longitude as New Delhi. Office workers there begin work at about sunrise and go home in mid-afternoon, solarly speaking.

    The only fully general solution to the calendar problem is to allow all future dates to be specified in the local civil time (LCT) of a particular location. Financial instruments such as bonds, options, and futures, for example, are usually specified with respect to a certain date at 5 PM New York or London LCT. This cannot be directly converted to an unvarying standard such as UTC, because Congress or Parliament might alter either the DST hours or even the underlying standard time zone as a whole at that location before the event occurs.

  2. brian says:

    @John Cowan: I meant a world with no time zones, not a world with one time zone. Instead of adjusting the clock to the local cycle of day and night, we wake in daylight and sleep in the dark regardless of what the clock says. If the work day is 9 to 5 on the east coast of the U.S., it could be 6 to 2 out west.

    But I should emphasize again that I haven’t really thought through the consequences of this change. It eliminates some sources of confusion but would doubtless break other things. To me the big plus is that we can think of time as a monotonically increasing sequence of moments, essentially the value of a counter. The big minus is that the entire human population has to alter habits that have become pretty deeply ingrained in the few centuries since mechanical timekeeping was introduced.

    In any case, not to worry: No one is about to adopt such a plan on my say-so.

  3. Dan says:

    One big problem with everyone-use-UTC is that you would get day changes in the middle of a typical workday. I live in South Korea, which is UTC+9. If we simply used UTC for time here, what is now a 8:00 to 17:00 workday would become 23:00 to 8:00 and would cross a (calendar) day boundary. What if you leave a deposit at your bank right before going to work, but they don’t process it until lunchtime — on what day would/should the deposit be credited? Would banks need to recalibrate their systems to account for the local time?

    Of course, there are other time and calendar things that cross boundaries — the western school year starts in one calendar year and ends in the next, and there are people that work 11-7 graveyard shifts — but my guess is that having a calendar day match a solar day is too useful to give up.

    (BTW, John Cowan’s example of Kashi has a counterpart: Japan’s northern island of Hokkaido is in the same time zone as the rest of Japan and Korea, but is far enough east so that during the summer the sun sets quite early, even though Hokkaido is pretty far north and has quite long days during the summer. Personally, I think both Japan and Korea should permanently jump ahead an hour, but like Brian, no one’s going to do that on my say-so. :)

  4. @Dan, it’s the northern hemisphere’s school year that spans a year; here in the South things are more orderly for schoolchildren.

    @John, of course the problem lies in conflating “tomorrow” with “24 hours from now”: dates in the past but especially in the future should never be treated as intervals from the present. It’s not that there are two midnights on the day savings time goes into effect, it’s that there’s less than 24 hours between civil middays straddling such a change.

  5. Dan says:

    Matías: the northern hemisphere isn’t all in agreement — here in Korea (and Japan too, I think), the academic year starts in March and ends in December, so it matches the calendar year. I suppose I was thinking “North America and Europe” when I wrote “western”…

  6. Jim Ward says:

    With GPS becoming ubiquitous, there’s no reason we couldn’t go to a longitude based time system. Your local time = GPS UTC + corrected to your nearest longitude.