How to Eliminate Daylight Savings Time?
August 31, 2007 12:46 PM   Subscribe

I'd like to build a clock that makes shifting to daylight savings time unnecessary by instead gaining or losing a bit of time each day as appropriate for the season so that when the second Sunday in March and first Sunday in November roll around it's already in sync. Setting aside the logistics of making this work from a societial point of view, how would you do it technically? Something like leap seconds or leap minutes? Or smaller changes on a more constant interval?

Now, consider the societal impact. We like to think of time as this holy construct, despite the fact that global synchronized time is a modern invention and Daylight Savings Time is even newer. Since the US government recently changed the rules for when Daylight Savings Time is in force and several communities apparently survive without observing it, the time seems ripe for making the argument that it's possible to do without a semi-yearly shift.

Many people get their time through their cellphone instead of wristwatches, and computer clocks are often synchronized to a remote source, so it seems like it would be possible to distribute post-industrial time to a wide population by getting a few carriers on board. Plus, there's historical precedent for clocks losing time. Many people grew up winding a clock each day, or adjusting the weights in their Grandfather clock. It seems like the shifts would be subtle enough that people could simply adjust any disconnected clocks as necessary every week or so.

I'd like you to help me come up with some mathematical strategies for how to build this clock, then explain to me why it'll never work.
posted by Jeff Howard to Society & Culture (23 answers total) 4 users marked this as a favorite
So the day before daylight savings time your clock would be 59 minutes off? Wouldn't you be late to work?
posted by JJ86 at 12:58 PM on August 31, 2007 [1 favorite]

It's a good idea in theory, but assuming you wanted to keep to 60 seconds in a minute, 60 minutes in an hour, 24 hours in a day, etc, it would involve a weird situation where you had two kinds of seconds of differing lengths: "scientific" seconds as defined by the current definition, and "everyday" seconds, as defined by your new time system. Either you'd have to convert all scientific data ever produced that involves time somewhere in the calculations to the new second measurements (and risk confusion even easier to create than obvious stuff like metric/imperial measurements that even NASA have fucked up), or you'd have to deal with 2 types of second which would add needless complexity to scientific research/education and the use of time in peoples' lives.

All for the benefit of not having to change your clock twice a year (which is done automatically now anyway on computer clocks and radio-controlled clocks/watches).
posted by EndsOfInvention at 12:59 PM on August 31, 2007

You can rotate the whole clock mechanism by 30% backwards and forwards over the year. Then you don't have to mess with the existing time circuits. I think you could use a cam to give you the sinusoidal motion you want.

Assuming analog clock of course.
posted by smackfu at 1:00 PM on August 31, 2007 [1 favorite]

Best answer: Continuous Daylight Savings Time.
posted by vacapinta at 1:00 PM on August 31, 2007

So the day before daylight savings time your clock would be 59 minutes off? Wouldn't you be late to work?

I think the idea is to have everyone convert to the new time system so everyone's synchronised anyway.
posted by EndsOfInvention at 1:01 PM on August 31, 2007

Best answer: Well, consider a conventional clock, which must move at a constant rate. That is, it moves 1 hour per hour, or for an analog clock, 1 rpm for the minute hand. (and 1/60 for the hour hand).

Instead, say that half the year, it must move slightly faster (so that over the course of 6 month, it gains an hour), or a rate of 1.000228 hr/hr (or 1.000228 rpm for the minute hand). The other half of the year, it would go that much slower, or 0.99977 rpm. Of course, there are infinite alternative approaches; a more mathematically elegant approach would be to use a sinusoidal function, so that the speed of the clock would be continuously changing. That is, the minute hand would move at 1 + 0.000228*(sin date/C + offset). Determining appropriate values for C and offset is left as an exercise for the reader.

The infinite number of implementations, combined with the difficulty of each implementation for some types of timepieces (what works well for a computer might be exceptionally difficult for a mechanical watch, and vice versa) would make this a logistical nightmare. Add that the change would be within the error of most clocks, and it's clear that this would be very difficult - although not impossible - to implement. As long as you could convince people to consult networked clocks extremely frequently and abandon all existing timepieces.

It seems like eliminating DST is a much better approach if you wanted to make a change. Your approach adds a lot of difficulty for its own sake.
posted by JMOZ at 1:07 PM on August 31, 2007

It's not practical. Suddenly, most of the clocks in the world have to be much more complicated.

Instead of swinging a pendulum, counting the oscillations in quartz, or metering the number of electrons coming out of a hunk of Cesium, now they also need to know the calendar date and a enough smarts to apply the date to a function that calculates the length of a tock.

Say goodbye to grandfather clocks everywhere.
posted by cmiller at 1:15 PM on August 31, 2007

I'm so excited about this concept that I can't slow down enough to wrap my brain around it before contributing.

Could this same clock ensure that sunrise occurs at (for example) 6am and sunset occurs at 8pm every day, all year round? So throughout the course of the year, some workdays would literally go by faster than others?

So in the winter, we'd sleep a lot longer and work a lot less. I could get behind that. Unless I have it backwards.

I think *that* would be fascinating. Might not be what the OP had in mind.
posted by iguanapolitico at 1:32 PM on August 31, 2007

Could this same clock ensure that sunrise occurs at (for example) 6am and sunset occurs at 8pm every day, all year round?

That clock would have to know your latitude, and time would be different for every lattitude, and what the heck would they do in the land of the midnight sun? That would be one very long workday in the summer.
posted by teg at 1:38 PM on August 31, 2007

Response by poster: I'd probably build this as an electronic clock to begin with. Maybe a Widget on my Macbook desktop.

Iguanapolitico, I hadn't considered that, but sychronizing it to the sun rather than to DST is an interesting idea. When the US implemented a national time, critics said that God's time was being replaced by Vanderbilt's time. Vanderbilt has had his day in the sun. Synchronizing a clock to sunrise and sunset might be interesting.

I thought about the problem of having two different definitions for seconds. Maybe you have to call them something else, like the difference between imperial and metric units. Both units can coexist, it's just a different way of measuring.
posted by Jeff Howard at 1:46 PM on August 31, 2007

Response by poster: So the day before daylight savings time your clock would be 59 minutes off? Wouldn't you be late to work?

I think of it more that everyone else's clock is off by 59 minutes the day before. That's why they have to shift them--the illusion becomes unsustainable.
posted by Jeff Howard at 1:50 PM on August 31, 2007

Best answer: Smackfu has it right for a mechanical clock. For a digital clock, you could insert/subtract a leap-second roughly every 72 minutes (assuming the DST and standard time were each 6 months long).

EndsOfInvention is more or less correct, but the fact is that astronomers are already doing this, since they don't observe DST, and use Julian dates instead of (or in addition to) the Gregorian calendar. So, yes, an added burden, but one we already bear.
posted by adamrice at 1:51 PM on August 31, 2007

As I understand it, the Romans kept time more or less as iguanapolitico described. Each day was divided into ten hours from sunrise to sunset; in the winter, hours were necessarily shorter. This makes a lot of sense if your workday depends on available natural light. Less compelling in the modern artificially-lit world, though.

The World Famous, that's not how sundials work— a sundial will show sunrise occurring at different times throughout the season, exactly like normal civil time without DST (except with a small constant offset according to your longitude).

Going back to the original question. I think messing with the length of the second would be a real can of worms.

I think what I would do is just create a notion of dawn- or dusk-referenced time. Right now, our timekeeping is referenced to noon/midnight (mean solar noon, plus adjustments): that is, 9AM is always three hours before noon. On the other hand, if you had a dawn-referenced time scale, 3:15 o'dawn would always be 3 hours fifteen minutes after sunrise; 21:00 o'dusk would be three hours before sunset. The three timescales would slide across each other throughout the year. You'll have some problems above the Arctic circle, but maybe you could define civil sunrise/sunset to the sun's closest approach to the horizon, or whatever. You'll have a few minutes of variation in day length to account for: 24:03 o'dawn, in the spring, might still be slightly before 0:00 o'dawn the next day; in the fall, it would be slightly after 0:00 the next day.

These timescales vary with latitude in addition to the variation with longitude that noon-referenced time has, so "time zones" would have to become a grid of squares rather than north-south strips. But in my opinion if timekeeping is that important to you, you should just be using GMT (or TAI) for everything anyway. Otherwise, just use local solar time, and eliminate timezones in favor of a continuous variation across the planet.

Wow, I'm starting to sound like some sort of horology hippie.
posted by hattifattener at 2:32 PM on August 31, 2007

Response by poster: hattifattener, I've read through your recommendation a few times now and I'm not quite getting it. You mention three timescales sliding across each other... Do you mean dawn/dusk referenced time, our current timekeeping system and solar time? Or three different instances of your timescales based on their references to dawn, noon and dusk?
posted by Jeff Howard at 2:55 PM on August 31, 2007

Microcontrollers are basically a simple computer-on-a-chip designed to be embedded in the circuits of devices to perform tasks like turning a clock into the slightly smarter clock that you describe.

These days, a simple microcontroller retails at about $2, and runs happily on about 0.001 Watts. (And they are much cheaper if you buy in bulk).
posted by -harlequin- at 3:00 PM on August 31, 2007

Best answer: We wouldn't have to change the length of a second for this to work, necessarily. We could just add two "Leap Minutes" (slightly more than a minute, technically, I guess) every Saturday night at 2am, right?
posted by Rock Steady at 3:53 PM on August 31, 2007

So, are you saying that it's impossible to design a sundial to do that? Or just that current, traditional sundials don't work that way?

I'm pretty sure it's impossible. Because of the inclination of the Earth, the point on the horizon where the sun rises each day shifts over the course of the year. This means that the shadow of any fixed object at sunrise is going to have to shift over the course of the year. Unless you're going to allow your scales to shift over the course of the year to compensate, the "time" indicated at dawn will shift over the course of the year.
posted by Johnny Assay at 5:18 PM on August 31, 2007

we can build it. we have the technology.
posted by xorry at 7:39 PM on August 31, 2007

Yeah but if you worked from, say, dawn+3 to dusk+1 every day, you'd work way more hours in the summertime. This would suck.

I don't mind working less in the wintertime, but it's gotta be capped at eight hours in the summertime, eh?
posted by bink at 9:38 PM on August 31, 2007

If you live on the equator, sun dials become much more consistent.
posted by smackfu at 9:50 PM on August 31, 2007

In the Roman Empire, 12 hours was, by definition, the interval during which the sun was up. An hour was longer in summer, shorter in winter.
posted by eritain at 1:24 AM on September 1, 2007

I still don't get the logic behind Day Light Savings Time. Do we really have to change the entire system of TIME ITSELF to avoid driving to work in the dark? Why not, I don't know, just get up later?
posted by greytape at 2:28 AM on September 1, 2007

greytape, that's a good point. Daylight Savings Time is something that I've always taken for granted because it's always existed (for me), but I've also always thought that if it never existed before and suddenly someone (imagine GW) went on TV and proposed that we all set our clocks differently twice per year, at the very same time, I think it might sound ludicrous.

Still, if it's a choice between "changing time" and everyone getting up later ... changing time is really much easier. Only one person per household is required to go around and press some buttons or turn some dials. Alternately, each individual would have to remember that their schedules have suddenly been pushed forward/backwards by an hour. ("Oh yeah, band practice was at 8pm, damn." Ad nauseum - school, work, psychiatrist appointment, bedtime, etc ... lots of stuff to remember every day rather than just changing the clocks once.)

I always wondered if we couldn't just change all our clocks by 30 minutes (once), setting time to just in between standard and daylight time, and leave it there all year round. Would that be a viable compromise? I've never done the math. (As it applies to energy usage, etc.)
posted by iguanapolitico at 4:15 PM on September 2, 2007 [1 favorite]

« Older Brain way beyond racked. Need suggestions for a...   |   Sending sensitive information over email? Newer »
This thread is closed to new comments.