Talk:2542: Daylight Calendar

Explain xkcd: It's 'cause you're dumb.
Jump to: navigation, search

When did y'all in the US "fall back" your clocks? It has a look of being (askewedly) inspired by DST reversal, and I know you did one of the switches at a different typical weekend than us (UK BST>GMT was last weekend of October), but I thought it was 'first weekend of month-after-(the-month-that-it-is-our-last-weekend-of)'. You know, I could have just looked this up. 00:11, 16 November 2021 (UTC)

Second question, more easily expressed and less obviously answered, which sun-up/sun-down is this calculated by? Nautical, civil, etc? 00:11, 16 November 2021 (UTC)

I think at the equator, you get one day per day. At the pole you get two days per day in summer, then one six month long day.Template:Unsigned

Ah, I just mentioned that, in my edit. Though it depends upon how close to the pole as to how long you wouldn't get one full day for (and how the shifting boundaries align, possibly). I haven't worked out if those "further north" people are necessarily Arctic, or merely Canadian/northern-States even. I know that in the UK we're north enough to technically never get beyond civil twilight in the 'summer' months (the Sun isn't low enough below the horizon, as it passes below the northern rim, to be proper 'night') but we're still short of the actual Arctic Circle and true days-without-night/light, accordingly. I'm still not sure what edge-case is imagined. Perhaps intentionally left vague? 00:59, 16 November 2021 (UTC)

Is this supposed to be about whether it's cloudy? 00:17, 16 November 2021 (UTC)

I think not. If cloud cover were taken into account, the date would fail to ever sync up from place to place, and in heavily-overcast areas (*cough* Cleveland *cough) you'd only have a few days a year. Whereas, if cloud cover is irrelevant and a "day" is simply defined as the sun being above median horizon level (whether visible or not) for a total of 12 hours, then the date in each time zone will sync up once a year. Presumably you would arrange things so that this happens on New Year's Day. (If you don't arrange things and just switch to the calendar at some point in time, the date would sync up on the anniversary of the switch, assuming everyone switches at once.) Additionally, timezones would work roughly as they do now, if you travel due east or due west: driving west from Eastern time to Central, for example, would take you to a place where it's one hour earlier on the same date. Things would only get really weird if you go north or south at all, and even that would be sufficiently regular and formulaic that school children could be taught how to calculate it. --Tsadok

If I knew where to start (too many assumptions needed), I'd be tempted to make an "xkcd Calendar" that works like the xkcd Clock, but there are so many possible configurations (e.g. when is the 'epoch' of synchronisation? When do you count daylight from/to? Do you assume 6AM day-starts and work up from there?) before you then have to plug in your lat/lon to get your highly personalised datetime result that may well differ significatly even from someone a few miles away, when the time-boundaries involved have misaligned just enough and haven't shifted back together again (perhaps!) by the next epoch-point... 01:34, 16 November 2021 (UTC)

If you use the same epoch for everyone, the date and time would sync up within each timezone once a year, on the anniversary of the epoch. If everyone just naively starts at their then-current date/time on the switch date, then the switch date is the epoch. The ideal epoch and switch date for minimizing confusion would be midnight January 1st; this also has precedent (e.g., the Unix epoch is midnight January 1st 1970). This leaves the question of whether to start each day's time at 00:00:00, or start it at whatever you have to start it at to make noon happen at 12:00:00. The latter would mean starting the clock at negative times in the summer, positive in winter, and any given latitude's start-of-day times would average out pretty close to 00:06 over the course of a year. (The average wouldn't be _exactly_ 00:06 with infinite precision over a single year, because you're only averaging a finite number of days. But the average would asymptotically approach 00:06 over large numbers of years (unless the DST change being not-at-midnight or leap days being not-at-New-Year throws it off in a systematic way; but I am guessing this proposed calendar would replace and obviate DST; leap seconds *are* added at midnight, so they're ok; that leaves leap days as a potential monkey wrench if they aren't moved to align with the epoch), and it would be "close enough" for garden variety everyday purposes even after just one year.) --Tsadok
Ok, maybe go with this, start it at Unix timestamp 1609459200 (Midnight Jan 1st 2020, GMT), and you'd need to look up sunrise/sunset (0 degrees with respect to horizon) in the location (possibly round to the nearest degree of latitude for the closest conurbation, except when in the ocean) for the year. There may be a formula for that, or there may just be an API you can hit. Things wouldn't quite sync up, because the orbit isn't a unit number of sidereal days long. I'm wondering how bad the drift gets in 400 years. Erin Anne (talk) 17:14, 17 November 2021 (UTC)
Hi, I just wrote some python code that hopefully works to check an API for sunrise and sunset and then calculate from the start of the year: 15:15, 8 December 2021 (UTC)
I like it, maybe add an explicit summary: "It is currently <datetime>, Day Number <day#>. Today will end at <day-end>, <x> seconds from now" Erin Anne (talk) 15:36, 8 December 2021 (UTC)

I know it is released close to daylight saving change. But has it actually anything to do with that? It is not mentioned at all, and only the darkness of November has any relation to the change. I'm not sure I would include it in the DST category... Randall has often made it clear that he dislikes DST but this new calendar is no guarantee they would not also include DST anyway. Hopefully we will stop with the DST in Europe from next year, so that we will not change back to summer time next spring! --Kynde (talk) 07:53, 16 November 2021 (UTC)

I thought it would have been closer to (US) changeover to be directly related. But perhaps Randall took a week to 'run some numbers' after being inspired, to get some (implausibly) plausible scenario to depict.
As for stopping putting the clocks forward... You could do that, but I'd suggest waiting a few years to see if people accept that before you also stop turning them back again in autumn, just in case... ;) 10:53, 16 November 2021 (UTC)
The conversation is clearly a parody of the kinds of conversations that occur regularly after the change to/from DST. I was at the eye doctor yesterday, and when she asked me about night driving I mentioned that this didn't really become an issue until the DST change, because now it's dark in the early evening. So these conversations do occur more than a week after the change. Barmar (talk) 17:18, 16 November 2021 (UTC)

Has anyone noticed the different style of Megan in this comic? I think that it may not be her… --Obscure xkcd reference (talk) 16:50, 17 November 2021 (UTC)

Three-Day Days

What I want to know is, what latitude do you have to be at, to get three-day days in November? --Tsadok

Anywhere north of the tropic, probably, if you count partial days (if a day starts before sunset, goes over anything even a smidgen less than 12 hours of daylight the next day, then completes itself soon after sunrise the day after).
But a bit of looking around suggests that today anywhere around 70°N has less than 4 hours of daylight so would easily need three full daylights in a row (or all-but) and possible unconsidered fractions of the neighbouring days depending on how the modified day-boundries land, which you may wish to ignore if you're strict). Wainwright, AK, would qualify, amongst other US settlements (most with native-names) and a number of Canadian ones, assuming you mean North from the US. There's Denmark (i.e. Greenland), Norway/Sweden/Finland and of course Russia with places too. Murmansk is slightly too far south today, I think, but I haven't checked for later dates in November and I'm sure it'll get included before too much longer.
Looking for a current six-hour daylight (i.e. any daylight not included in the period the day trips over upon is exactly made up for at the beginning of the third day touched before the new-calendar day ends) suggests reaching ~65°N will suffice, which adds a fair few other places 'up north' that would apply to right now. 21:15, 16 November 2021 (UTC)

Of course, before mechanical clocks, all days were 12 hours long. It is just that the length of the hour varied, as it was always one twelfth of the day, or night. Arachrah (talk) 18:04, 17 November 2021 (UTC)