davedelong/calendar_fallacies

yourcalendricalfallacyis.com

calendar_fallacies's Issues

There are never more than two dates concurrent at any instant

The same number of years will have passed between two dates, regardless of the calendar used.

For the Gregorian, Hebrew, Julian, etc calendars, the possible error here is 1 year. It gets a lot more confusing when bringing the Muslim calendar into account, again because the years have a different length. 200 years in a solar-based calendar, becomes ~206 years in the Islamic calendar.

A minor addition to an existing fallacy: Like with the Hebrew calendar, the Muslim date is different from the Gregorian one. (It's 1438 this year!)

Add a couple of common date formatter year bugs.

• Y vs y and y vs yyy or yyyy.
• https://twitter.com/davedelong/status/740998081542389762

The week may or may not “start” in the middle of the weekend

Related to the “Saturday and Sunday are the weekend” fallacy, in the United States the week starts on Sunday (part of the weekend), but in Europe the week starts on Monday (the first day after the weekend).

Leap years occur every four years contains incorrect example

The fallacy "Leap years occur every four years" contains this text:

"False. The year 2000 was not a leap year."

2000 was a leap year in the Gregorian calendar. 1900 would be a better example.

date format string fallacies

Common date format string fallacies would be a great addition.

Mention 0-indexed 12 hour time

Some countries (Japan, iirc?), when they use 12 hour time at all (which is rare since they're 24 hour time by default), use 0-11 instead of 1-12.

The month or day may be `00` or `XX`

From here:

the Swiss Government may issue identity cards or passports which do include a birth day of '00' and a birth month of '00',

And:

In the ISO 8601-2 (extensions), 2016 draft standard, the use of an 'X' for an unknown date has been adopted:

4.3 Unspecified
The character 'X' may be used as a replacement character, in place of a digit to indicate that the value of that digit is unspecified.

(via)

Mention YYYY vs yyyy/ISO week-year

One of the most common and mystifying bugs!

The duration between two UTC timestamps can be computed

Due to the unpredictable nature of leap seconds, an interval that includes any date in the future cannot be guaranteed to be accurate. At any time, the IERS can declare a leap second, which must occur in the last minute of the month (June and December are preferred). In practice, the IERS gives about six months advance notice before a leap second, but is not required to do so. Because a leap second could be added at the end of any month, only intervals that do not span the last day of a month can be guaranteed to be accurate to the second.

In order to compute durations between timestamps in the past, a table of all leap seconds must be consulted.

An hour is always 60 minutes fallacy

The Hebrew calendar defines an hour as 1/12 of the number of sunlight minutes in a day. (Wikipedia covers this - I totally forgot earlier, but this fallacy is the entire basis of KosherCocoa)

Adar is not a missing month, it's added on leap years

In a range of months, a month will not be missing in the middle of it

False. The Hebrew calendar uses leap months, which occur during the middle of the year.

In fact, on leap years we add a second Adar (Adar B) right after Adar. Purim is celebrated on Adar B in those years, so if you know Purim is on Adar 15th... You're wrong.

“A date without time needn’t have a time zone”

The iCalendar specification (RFC5545) states that a DATE must not have a time zone, this means that all-day events in one time zone can show on the wrong day in another time zone.

I’ve been repeatedly grumpy about this on Twitter as it means that as someone inhabiting a sufficiently different time zone (17 hours different) than my coworkers, “all-day” events they create always appear in my calendar on the wrong day, and vice versa!

Timezone offsets are always rounded to the minute

(or atleast seconds...)
Wrong:
https://en.m.wikipedia.org/wiki/UTC%2B00:20

Inconsistent title punctuation

Some headings don't have a period but others do

Days don't last longer than 25 hours

Everything you know about calendars is applicable to space travel.

Some headings are a question but are missing a question mark

It is normal that the Sept-, Oct-, Nov-, and Dec- months are numbered 9, 10, 11, and 12

Another exception to the 7 day week: Samoa

Samoa went from Thursday 29 December 2011 directly to Saturday 31 December 2011. There was no Friday, so in that location the week was only 6 days.

There is never a time of day larger than 24:00

False. As discussed in the timezone mail list during 2016 December, the autumn daylight time transition in Spain has sometimes occurred by waiting until hour 25:00 and then falling back one hour to begin the next day. See the tz thread archive at https://mm.icann.org/pipermail/tz/2016-December/024673.html and some of the citations http://www.boe.es/datos/pdfs/BOE//1919/266/A00992-00992.pdf and http://astronomia.ign.es/rknowsys-theme/images/webAstro/paginas/documentos/Anuario/lahoraoficialenespana.pdf

Every city has its own timezone

https://www.cbsnews.com/news/town-in-two-time-zones/

"Each province (or US State) has only one time zone."

False. Several US States and provinces of various contries span more than one time zone. For that matter, several states have more than one time zone within them; for example, Arizona contains both DST and non-DST time zones.

"OK, then, each city or town only has one time zone."

False. At least 14 cities and towns in the US straddle more than one time zone, usually because of a state border. The town of College Corner, Indiana/Ohio has a baskeball court which spans two timezones, making scorekeeping tricky. In Xinjiang, China, each neighborhood chooses between one of two time zones.

The relative velocity of my calendar does not matter.

It matters a lot! Special relativity says so. Calendars with velocities relative to UTC (like those in orbit around the earth, or sprinkled around the solar system) must account for time dilation.

Similarly, if local gravity is not earth-standard, time dilation will affect the accuracy of any timekeeping attempts.

Mention the 1 second long weekend in Paris

It hits a bunch of weird edge cases: "there can only be one weekend in any given hour", for example

(see https://twitter.com/Catfish_Man/status/845068555146035200)

Days are always sequential within a month

That is there are no 'gaps' between days in a month.

In September 1752 (US), 10 days were removed (9/3-9/13) to switch from the Julian Calendar to the Gregoian Calendar.

See Also

https://en.wikipedia.org/wiki/Adoption_of_the_Gregorian_calendar
https://www.timeanddate.com/calendar/julian-gregorian-switch.html

There are 4 seasons, and they always refer to a certain set of calendar months

Lots of places will say 'Coming this summer' meaning June/July/August; however Summer for those in the southern hemisphere is December/January/February.

Also, places in the tropics don't usually have 4 distinct season; instead having 2 - wet and dry. "In tropical climates there are often only two seasons, a wet season and a dry season."

1: https://en.wikipedia.org/wiki/Season
2: https://en.wikipedia.org/wiki/Tropical_climate

Today is Friday; Everyone agrees what "This Sunday" "Next Sunday" means

False! Some people use "this Sunday" to refer to the one 2 days from now, and "next Sunday" to refer to the one 9 days from now; others use "next" to always mean the upcoming date. There is no consensus on where the cutoff between "this" and "next" is. Still others are promoting the word "oxt" to mean "not the first one but the one after that."

It's best to avoid "this" and "next" when referring to days.

fallacy: leap year is always calculated based on a mathematical rule

false: the french revolutionary calendar calculated leap years by defining new years as the vernal equinox in Paris and inserting a leap day if necessary to make that possible.

related fallacy: people are in agreement about how to calculate dates in a given calendar.

false there are 4 different methods proposed to calculated dates in the french revolutionary calendar after the calendar was no longer used but there were still contracts and laws which referenced it.

Linkable fallacies

It would be interesting to have each fallacy linkable / with an anchor so you can get a link to one specific fallacy.

Months always have 30 or 31 days in them

Also Ramadan in the Islamic Calendar may be 29 (actually this month) Link to the current month on the calendar

Israel uses two calendars!

Israel uses the Gregorian calendar along side the Hebrew calendar in one of those odd few places that use more than one calendar at the same time. A significant portion of Israeli jews apparently don't know the Hebrew date of their birthday (translate)

Someday we will be able to create a _perfect_ calendar that fixes all these problems.

Nope.

(This is yours, I just thought it'd be great to put up there.)

UNIX Epoch Fallacy

Jan 1, 1970 is always the start of the UNIX time epoch. Only in UTC, it's 4pm Dec 31 1969 in California.

Fallacy: BC years can be easily modeled by using negative numbers

The Gregorian calendar took us from December 31, 1 BC to January 1, 1 AD (there is no year 0). So if you want to know how many years it was from the summer solstice of 5 BC to the summer solstice of 5 AD, you can't just do "5 minus -5 = 10". The difference was really only 9 years.

Submarines in the US fleet did not use 24 hour clocks until very recently

See:

https://www.military.com/daily-news/2015/10/16/submarine-force-now-24hour-work-day.html

Update information on September, October

It would be interesting to add to the question on the numeric placement of the dates to add the the prefixes of September, October, November and December literally mean 7, 8, 9 and 10 respectively. https://en.m.wikipedia.org/wiki/Numeral_prefix

Historical timezones

Tess Rinearson (@tessr) found this incredible chart of historic timezones: https://twitter.com/_tessr/status/900870734238449664

It should absolutely go in the list somewhere. (Probably the final "timezones are always on the quarter hour" fallacy)

Add Japanese year

Japanese year follows the year of the current Emporer's reign:

http://www.japan-guide.com/e/e2272.html

Happy Heisei 29!

Israel doesn't really use Hebrew Calendar

Most people use the regular calendar. Hebrew Calendar is used to set date for holidays.

"Nobody uses the Julian calendar anymore"

Don't forget java.util.Timestamp, which "use the Julian calendar for dates before October 4, 1582, and the Gregorian calendar for dates starting with October 4, 1582." Tons of code in the PostgreSQL JDBC driver to work around this fact. See https://www.ibm.com/support/knowledgecenter/SSEPGG_10.5.0/com.ibm.db2.luw.apdv.java.doc/src/tpc/imjcc_r0053436.html and https://github.com/pgjdbc/pgjdbc/blob/d28deff57684349707d2b2a357048f59b0861bb1/pgjdbc/src/main/java/org/postgresql/jdbc/TimestampUtils.java#L1255 .

The Julian day is not directly related to the Julian Calendar

The Julian calendar is still used by Eastern Orthodox churches, and the Julian day is used in certain kinds of astronomical calculations.

The Julian day number, used in astronomy, is the number of days since Monday, January 1, 4713 BC, proleptic Julian calendar (November 24, 4714 BC, in the proleptic Gregorian calendar). It has some relationship to the Julian calendar, but it doesn't really depend on it.

https://en.wikipedia.org/wiki/Julian_day

"Month lengths follow regular cycles" fallacy fallacy 😅

Great work, thanks Dave!

I think I spotted a mistake:

In this fallacy you wrote:
[...] and the moon isn’t visible, then the previous month just keeps going… and going…

Actually the visibility only determines wether the month will be 29 days long. If the new moon is sighted, then the following day is the first of the next month, so the current month stops at 29.

If it can't be seen, then the month is only one day longer, 30 days, and the day after the 30th is automatically the first day of the next month, without moon sight.

Furthermore, a "previsional" long term calendar is often established in order to be able to deal with future dates. The 29/30 pendulum regularly merges back with the long term calendar. (Don't quote me on this last one, since there are many sources and differs by countries).

A minute is always sixty seconds long

A UTC minute can be 59, 60, or 61 seconds long, due to "leap seconds." Leap seconds are added (or subtracted) at the end of a month to account for changes in the rotation of the earth, such as those caused by the moon, earthquakes, and man-made events. As of 2017-03-24, 27 leap seconds have been added, and no negative leap seconds have been applied.

When a leap second is added, the last minute of the last day of the designated month has 61 seconds, which is displayed as "23:59:60." In the event of a negative leap second, the last second of a day would be "23:59:58."

ICU does not include leap seconds in its time calculations.

DST change on short notice

I have an even shorter "Daylight Saving Time rules do not change on short notice" example. Basically one day notice.

https://en.wikipedia.org/wiki/Daylight_saving_time_in_Morocco

"However, on Saturday, 28 September, the government announced a change in policy, which moved the end of DST from the last Sunday in September (which was to be the next day) to the last Sunday in October (27 October)"

"OK, DST rules change, but never retroactively."

False. Several times in the last 15 years, specific governments have chosen to enact changes to local DST rules which were either applied retroactively, or applied with so little notice that programmers had to add them to systems retroactively. For example, in 2015 Turkey moved the end of DST by two weeks with less than two weeks notice, and it generally took most OSes up to three months to catch up. Further, historical DST databases sometimes contain errors which must be corrected for historical data.

February never has more than 29 days

Currently true, but Feb 30 occurred in Sweden in 1712 (due to a complicated transition from the Julian to the Gregorian calendar)
Wikipedia link

A given hour will always occur in the same part of the day (eg, 23:00 is always night).

Arctic circle seasonal issue aside - this is untrue when discussing naval or air coordination, which often use Zulu time. (Basically GMT without daylight savings.) This time is the same worldwide, which prevents confusion. So in the above example, 23:00 Zulu time could fall at any time of day - morning, afternoon, etc - depending on the longitudinal distance from Greenwich. Frequently used in aviation flight plans, for example.

New year occurs on the first day of the first month

False. In the Hebrew calendar New year is the first day of the seventh month

Daylight Saving Time rules never change

False. In the US, the Energy Policy Act of 2005 changed the starting and ending dates for DST starting in 2007. In Brazil, prior to 2008, the DST start and end dates were set by government fiat and varied every year.

Days start at midnight (or as close to it as possible)

As an addendum, Daylight Saving Time 2017 in Lebanon will begin at midnight on Sunday, March 26. So technically, the day starts at 1:00 AM.

Fallacy suggestion

Weekends are holidays.
False! Hungarian government may declare "working saturdays" for a year to offset for extra "bridge public holidays". A bridge public holiday is when the government announces a Monday-before-a-hoilday-Tuesday to also be a holiday so that people get a 4 day long weekend. Same goes for a Friday-after-a-holiday-Thursday.

Two countries in the same time zone, will always have the same time.

Daylight savings sometimes begins/ends at different dates even for countries in the same time zone. This sounds innocent! It isn't. I used to coordinate cross-border activities between Israel and Lebanon. Some of these activities were controlled demolitions of bombs or mines. Back then, we switched about a whole month earlier than in Lebanon because of Yom Kippur (for ridiculously illogical reasons you might find entertaining if you google them. I think they've changed it since then, but who can keep track - I'm out of the army now). Anyway, for that entire month, people would forget about the time difference ALL THE TIME. Soldiers on patrol would almost get hurt because they were told to avoid a certain stretch of border during one time period, when in fact the mine was being defused/exploded an hour before. Or an hour later. (It was really confusing, okay??) And then, because personnel got rotated every 1-2 years, you'd have fuckups every. single. year.

A date/month/holiday will always fall in the same time of year.

In the Islamic calendar, because the length of the year is different from the solar year, dates shift slightly with each cycle. Holidays gradually migrate through the seasons. Ramadan was in summer last year; in a number of years it'll be during the winter.