I have reached the stage in my latest science fiction epic (in form if not quality!) whereby I need to start solidifying the date system and setting the history of the universe concretely. The date system should be convenient both for characters to discuss and the reader to understand quickly enough.

The date system needs to span at least 10^24 years (it has long been accepted that life will necessarily die out before then, in the book).

I am flirting with the idea of using ordinary earth dates (01/01/1970) up to the year one million. After then, the powers that be decided that writing 01/01/1000000 was not ideal and not very scalable.

Imagine the date: the first of January, four hundred and eighty three trillion two hundred and ninety seven billion five hundred and fifty five million seven hundred and twenty one thousand eight hundred and fifteen (01/01/483297555721815); it doesn't exactly role off the tongue (not to mention it's basically unreadable in any convenient amount of time (with accuracy) for people suffering with certain medical conditions which affect their ability to perceive written numbers).

My universe is has a very explicit timeline and characters will spend time exploring various events of the past and the book will span trillions of years from front to back (there is no time travel in my universe). How can I invent a date system which is human convenient that my characters can use when discussing things, and most importantly, the reader can keep a reasonably simple timeline in their head and understand (ish) when I am talking about.

This problem is quite noticeable when picking two large dates, take the date example above, and this date - 01/01/23365535731776. they look very similar, but there's over 450 trillion years between them.

TL;DR requirements:

  1. Accuracy to the day is important. It is no good saying "at around 10^19".
  2. Not event or loose era based. E.g. LotR uses 'First Age' and SW uses the 'Battle of Yavin' as date markers. I suppose a year (2022) is an 'era marker' in itself, but it is consistent and logical and one can identify its neighbours just by knowing the system.
  3. Span at least 10^24 years (In my book, humanity has retained the idea of earth time, so a year is always an earth rotation around the sun (as recorded in the year 5000).
  4. Be human (reader) convenient and understandable without much mental acrobatics.
  5. The date system must be computer readable also and retain all the functionality that our current date system has.

Ideas I am considering (most likely a combination will be required?):

  1. Some kind of hashing function whereby the universe agrees on some standard set of shorter hashes. (not ideal from a reader perspective as 01/01/1fty32 and 01/01/6hyrrj could be right next to eachother or thousands of years apart).
  2. Scientific notation 1.32322E+25 (still not very convenient).
  3. Breaking the date up into more than three segments. e.g. 01/01/12/45/689/9998 - (this would combat the problem hashing functions have of readers not being able to tell if dates are close or distant from each other, but it is still an awful lot of numbers required).
  4. Associating each number with an alpha mark - 0t10b would be 0 trillion 10 billion. 126t778b etc does not look to pretty though when combined with the rest of the date.

So far my best idea (I think) is to have some sort of consistent era markers, followed a reasonably familiar yet larger system at the end such as aaaaa.365/99999/9999.

  • The aaaaa represents trillions and billions and the next would be aaaab >> aaaaz etc.
  • 365 represents the day of the year (months have been scrapped but the days remain the same length)
  • 99999 represents the year
  • 9999 represents the hundreds of thousands marker

The first day 65 trillion, 77 billion, 10 million, 869 thousand 222 would be: acrfz.001/69222/0108

I think I like this system because it means readers will look at aaaab and aaaac and know they are far apart, but not as distant as they are from ttyxx. It also has specificity and removes the month to reduce visual noise. I would also footnote every date using a written English version of the date to help the reader further understand if necessary.

Anyone with any thoughts on the above system? I like the ordering of 'era marker'.'smallest > biggest' but my main issue is it is still quite 'numbery' - can anyone think of a way to make it simpler but still capable of addressing each and every day?

Finally, history and timescales, and discussion thereon are key story and world building elements crucial to the book so I am not able to just 'pick 50ish dates' that the reader needs to know about and forget the rest.

Sorry for the long post - any help, ideas or pointers to people who have already had ideas is of great help! I should say that my target audience would be those who are more mathematically / scientifically inclined and so I don't need to hold their hand too much but I don't want to exclude everyone who isn't Rain Man.

p.s. I have searched the internet for long term date systems but I have come up very short.

Thank you!

Edit 1

I have noticed in the answers and comments provided below, which have been just as helpful and insightful as I had hoped (and even more plentiful - this question even spent some time in the Hot Network which I was surprised at!), that people are often asking the question 'why?'. Their comments are valid, as humans often don't concern themselves with what's happening a decade from now, let alone thousands or millions of years apart. So I hope to give some (exceptionally brief) clarity as to what I am trying to achieve in case it helps answers have some more context on which to base their answers. This context does not invalidate already given answers in any way and I have found intrigue in all of them; thank you for that.

Added context

As alluded to in the question, history/events/time etc are crucial to my story. My story is centred around events spanning trillions of years. My characters will notice odd things and then questions will be asked about why, who could do this, divine design? evil corporation? Why is earth time such a good fit for all this? The broader picture will spell out a tale of immense planning of cosmic proportion, wherein the very movements of the universe itself are in play. Dates will need to be specific at the very least down to the year of events and often more so, as such events are mathematically linked - detail is important. Indeed, my 'marathon of the middle' will build up to a climax which will see my characters connect dots on great scale which will reveal this so called immense plan in such a way that readers will (hopefully) click; ahhh! It will be seen that certain events and the very universe itself are linked and could not happen in any other way or order. They will use this knowledge to identify future dates of interest and thus we are thrown into the whirlpool of 'the plan' itself and are able to start having a real effect, pressing toward a satisfying and triumphant ending.

tl;dr The maths is important to me and my story. For me, and hopefully my readers, the very beauty of the tale is in the 'grand design' of events and the 'not randomness' of it all. - I hope this answers some people's questions.

Edit 2

Thoughts on the 'frame challenge' answer

I should like to thank the user for giving such a complete and well thought out answer; It contains extremely useful referenced information and makes sound, logical sense.

I have thought long and hard about whether my framing is off for this question, since the frame challenge answer is currently the highest voted; however, I consider that the frame challenge fails in this instance (even though the answer is quite excellent in many ways), for these reasons:

  1. My question does not require that the date system has universal adoption, nor that it be the only date system that exists. I agree with the answers premise that many date systems would be used depending on the frame of reference. Indeed, in regular conversation and the day to day living of my characters, they will adopt phrasing similar to what the answer suggest. However, my characters will also require (as I think is reasonable, stated below) the ability to discuss a larger time frame, with accuracy.

  2. To say that is is "not practical" to have such a system (again I think erroneously implies that it would have universal application, which it does not need to have), is to suggest that across all of creation, there would be no people who ever want to mathematically compare a number of distant dates. I am never entering into my calculator '2022 + "circa the First Age"'; I would use what the answer suggests, and project my current date system back onto past events, even if they didn't use the system back then. Nobody alive BC ever used that to refer to themselves, but we can easily enough today label those time appropriately with our current system.

I consider that there, all but necessarily, must be times when people of the future wish to compare distant dates. Consider geologists discussing stellar events, or historians identifying previous 'dynasties', or calculating how long we have to burn our ion cannons to get out of reach of our slowly approaching, but specifically calculated doom. Remember that science-fiction would also allow for some extension of this reasonableness (for fun of course!). Consider also that the characters could even invent a system up to meet their requirements (which does not invalidate the question, as the question is concerned with both character and reader comprehension).

The answer provides useful commentary on a large number of date system relevant points and I am grateful for the time that was taken to write it. However, I do not consider that it correctly answers the question, and that questions which are looking more into using various 'base' points, or universal constants etc, are closer to the mark set by the question.

As a side note, the concept of Frame Challenge is discussed here, and I don't think the answer meets the criteria discussed there.

Please comment etc and let me know your thoughts on the above guys. It is the top voted answer, so I am expecting some push back!

p.s. I hope this does not come across as dismissive; I really have thought long about this and really am grateful for the ideas in the answer; they have given me a greater understanding of dates than I had before. Thank you!

Edit 2.5

The answer referred to above has been edited by its author to provide a suggestion which does meet the question criteria; as such, I am pleased to say I now consider it to be a valid answer (notwithstanding the frame challenge fail).

  • $\begingroup$ Comments are not for extended discussion; this conversation has been moved to chat. $\endgroup$
    – L.Dutch
    Commented Jul 26, 2022 at 13:26
  • 1
    $\begingroup$ Please eventually write your dates the IEC way, the same way we write numbers, biggest first, so today is 2022-07-26. Then you would obviously be able to prefix with numbered millenia, and possibly named ones, keeping consistent left to right, bigger to smaller ordering. $\endgroup$
    – Neil_UK
    Commented Jul 26, 2022 at 15:19
  • $\begingroup$ Are people with issues processing large numbers in writing an important part of your story? $\endgroup$ Commented Jul 26, 2022 at 22:17
  • $\begingroup$ Will it be a Saturday? $\endgroup$ Commented Jul 27, 2022 at 0:07

22 Answers 22


Frame challenge: no such dating system is practicable

What is the use of a dating system? Obviously, the use of a dating system is to tell when an event took place, or to tell when an event will take place. But also, perhaps a little bit less obviously, is to plan an event ahead, or to locate a past even in a familiar time slot.

Of course, nobody will want to plan ahead events over millions and billions and trillions of years; such planning would be utterly useless. Over a time scale of a billion years, the solar system is chaotic, so that one cannot even tell where Earth will be, even less what season will be in any given city; and no city on Earth has endured for more than ten thousand years or so, anyway. Over a time scale of ten billion years, the entire solar system will have changed beyond recognition. Over a time scale of a hundred billion years, the civilization will have changed home star systems several times.

Moreover, there is no historical example, none, of dating system in continuous use for more than 1,500 years. We date, for example, the accession of Trajan to the position of emperor of Rome to the year 98 CE: but at that time this system was not in use.

In practice, different system for different time scales

Let us consider the example of the dating systems we use in practice.

  • For dates with the last 2500 years or so we use the common year-month-day system:

    • With complications for dates more than 104 years ago -- some countries used old style dating, others used new style dating, and the delta between them varies. The famous example is the Great October Socialist Revolution (for which the cinematical Red October submarine was named), which took place in November as far as the world outside Russia was concerned.

    • With even more complication for dates more than 250 years ago or so -- some countries changed the numbering of the year at dates other than 1 January. For example, until 1752, England changed year numbers on 25 March, such that 24 March 1648 was followed by 25 March 1649.

    • With severe complications for dates more than 2,000 years ago. Quite often, for dates between 2,000 and 2,500 years ago we can mechanically translate a date into our calendar system, but quite often we have only a vague idea what actual day it was. For example, the traditional day of birth of C. Julius Caesar is 12 July 100 BCE; but due to the peculiar way the Romans counted years at that time, all we can say is that it was in the summer of the year 100 BCE, at some point between late June and early August. Yes, the Romans counted that day as 12 July (or actually ante diem IV idûs Quinctiles in their native reckoning); but they started counting days from 1 March, and 1 March was whenever the Great Priest said it was. And don't get me started on Greek dates.

      The Latin means "the fourth day before the ides of Quinctily". Only later were the months of Quinctily and Sextily renamed July and August.

  • Beyond 2,500 years ago the system year-month-day is not used, because it would be useless. Year-month or even year-season is fine and serves all practical purposes.

  • Beyond 3,500 years ago or so we switch to a system counting centuries, or even early-mid-late millennia. That serves all practical purposes well enough. For example, the 4th Dynasty of Uruk is dated to the 22nd century BCE; no more precision is available. (And even that "22nd century" is to be taken with a grain of salt; it could well be "mid-23rd to early 22nd" or "mid 22nd to early 21st".)

  • Beyond 5,000 years ago we count millennia. And since we count millennia, we count them BP, "before present", where "present" is defined as 1950. Yes, for archaeologists even the present is in the past.

  • Beyond 12,000 years ago we stop using year counts and we switch to the geologic time scale. The non-avian dinosaurs got extincted at the bounday between the Mesozoic and the Cenozoic, or, on a finer scale, at the boundary between the Cretaceous and the Paleogene, which is more or less about 66 million years ago. Note that the dating on the geologic time scale is precise (the extinction being the definition of the boundary between the Cretaceous and the Paleogene) whereas the dating by counting years is imprecise, with an uncertainty of tens of thousands of years. (The currently accepted dating by years is 66.043 ± 0.011 million years ago.)

What does this mean for you?

It means that it doesn't make sense to use one single dating system over such a long time span. At any given point in that overlong time span, people would be using

  • A dating system to the day for events within a few thousand years of their present.

  • A dating system to the year for events within say 5,000 to 10,000 years of their present.

  • A dating system to the century for events within maybe 50 to 200 centuries of their present.

  • A dating system to the millennium for events within maybe ten to fifty millennia of their present.

  • Beyond that, they would be counting hundreds of thousand of years, or millions of years.

  • Beyond a few billion years, they would be dating events based on the home solar system of the time: when Homeworld was located in the Manticore system...

As an added bonus, the system would naturally change from time to time, and quite often. Two of the longest lived dating systems in our history are the Greek system of dating by Olympiads, and our present system of counting years since an arbitrary epoch: one endured for about 1,200 years, the other is only about 1,500 years old. (For fun, the year we count as "1 CE" was "the year of the consulship of C. Caesar and L. Aemilius Paullus" in the Roman reckoning, or "1st year or the 195th Olympiad" in the Greek reckoning.) We have no historical examples of dating systems which endured longer. (Discounting the universal habit of dating by the regnal years of the local potentate. This "system" was of course in use everywhere for a very long time, but it has the drawback that it resets whenever the potentate changes.)

On the other hand...

The question can be understood as asking how to write natural numbers up to $10^{24} \times 365.25 \approx 3.65 \times 10^{26}$ in a more-or-less readable way.

(I am discounting the idea of "years" and "months". Our current years and months make no sense whatsoever over a such a long timespan. Wherever Homeworld will be two billion years from now it will most certainly not be on Earth-orbiting-the-Sun-in-One-Year. And there are five hundred thousand billion intervals of two billion years in $10^{24}$ years. Homeworld will have to change locations and orbital periods five thousand billion times in the timespan requested by the question.)

The point is that $\log_2 (3.65 \times 10^{26})$ is about 88.2. To represent a natural number up to $3.65 \times 10^{26}$ one has to represent 89 bits. There is no escape.

  • Using ten digits, 0, 1, 2, ..., 9, (3.3 bits / digit) one needs 27 digits.

  • Using sixteen digits, 0, 1, 2, ..., 9, A, B, C, D, E, F (4 bits / digit) one needs 23 digits.

  • Using 36 digits, 0, 1, 2, ..., 9, A, B, C, ..., X, Y, Z (5.2 bits / digit) one needs 18 digits.

If I were to make a suggestion, I would suggest the following scheme:

  • For dates up to 999,999,999 days before or after the epoch, use the decimal representation.

  • For dates between 1 billion days and 1,679,616 billion days before or after the epoch, use up to four base-36 digits (for the billions) and nine decimal digits (for days in the billion).

    For example, VO2—987,654,321 would be the 987,654,321st day in the 41,042nd billion days.

  • For dates more than 1,679,616 billion days before or after the epoch use a group of up to four base-36 digits followed by a group of exactly four base-36 digits, and then nine decimal digits.

    For example, NF'07WO—987,654,321 would be the 987,654,321st day in the 1,415,926,536th billion days.

  • This goes to 2.8 trillions of billions of days, or $7.7 \times 10^{18}$ years. For even more remote days, add a new group of up to four base-36 digits. (Three such groups are enough.)

Double precision floating point numbers have only 53 bits of precion. One would need quadruple precision arithmetic for those 89 bits...

  • 3
    $\begingroup$ "...when Homeworld was located in the Manticore system"; I really like this. It's (in theory at least) completely calculable and not whimsically based on events that may or may not happen (like wars and natural disasters etc). That allows me to limit the amount of specific sequential letters/numbers required to speak about a particular date. Thanks for your comprehensive thoughts. $\endgroup$
    – user97385
    Commented Jul 23, 2022 at 22:32
  • 3
    $\begingroup$ Arguably, the vagaries of past Earth dating might be due to limitations in measuring and documentation technologies, which might be different in our SF future. "Past performance does not guarantee future results." $\endgroup$ Commented Jul 24, 2022 at 5:26
  • 1
    $\begingroup$ Not to mention that, for civilizations who migrate between star systems, their year and day lengths will almost certainly change. And months will be interesting given that the number of moons and/or their orbital periods, sizes, etc., would change between home worlds. $\endgroup$ Commented Jul 24, 2022 at 20:59
  • 2
    $\begingroup$ @DennisWilliamson and the other issue is that nothing is really "simultaneous" between locations in different star systems, it all depends on the observer's frame of reference. (Technically this applies for any two points, but at non-relativistic speeds and short distances it can be ignored.) $\endgroup$ Commented Jul 25, 2022 at 6:12
  • 1
    $\begingroup$ The different time scales of describing things as happening in the 11th century or 200Mya are often used because they refer to time periods where we do not know or do not care about exact dates. It is both impossible and not worthwhile to guess the specific year a T-Rex lived, so an era designation is sufficient. But the OP requires a system where exact dates are both important and measurable. I don't find the fact that our current calendar is "only" 1500 years old as suggestive that calendars don't last - that's ~1/3 of the time calendars have been around and spans all of modern history. $\endgroup$ Commented Jul 25, 2022 at 13:27

You will name the period after the wavelength of the cosmic background radiation.

cmb graph

The cosmic background radiation is left over from the Big Bang. In our era this radiation has redshifted all the way to microwave frequency.


The CMB is a perfect example of redshift. Originally, CMB photons had much shorter wavelengths with high associated energy, corresponding to a temperature of about 3,000 K (nearly 5,000° F). As the universe expanded, the light was stretched into longer and less energetic wavelengths.

By the time the light reaches us, 14 billion years later, we observe it as low-energy microwaves at a frigid 2.7 K (-450° F). This is why CMB is so cold now.

It is going to keep shifting. There is no limit to how big the wavelength can get.

emr wavelengths


There are frequencies of longer wavelengths than radiowaves. ULF radiation has wavelengths in the kilometer range.

Your dates will be metric units of measurement. One can extrapolate from the steady increase of CMB wavelength to figure out what the wavelength would be at a given time in the future.


However, the Big Bang’s leftover glow will never disappear entirely. No matter how far we extrapolate into the future, even as the density of photons and the energy-per-photon both continue to drop, a large enough, sensitive enough detector tuned to the right wavelength could always identify it.

  • 2
    $\begingroup$ This is a fascinating idea - something ever constant, and predictably changing to refer to a time. How would you deal with 1) skipping numbers: the CMB would be different at 8am as opposed to 3pm on any day (as I am only concerned with specificity down to the day (time will be handled as usual) - meaning some numbers (all those between 12:00am and 11:59 etc would be discarded) - reader confusion maybe?. and 2) requiring large numbers to specify the entire range (trillions of days) - maybe naming the wavelengths after an agreed set of things? "in the '1kg hydrogen atoms in a line' era"? $\endgroup$
    – user97385
    Commented Jul 23, 2022 at 22:43
  • 6
    $\begingroup$ This seems to answer a different question. To specify a date you still need to write down a wavelength of 1.562942648372843911374847540880753868917512663386222353693179318006076672635443338908659593958290563832266131992829026788067520876689250171169620703222104321626954862629631361443814975870122034080588795445474924618569536486444924104432077134494704956584678850987433944221254877066478091588460749988712400765239113 . . . $\endgroup$
    – Daron
    Commented Jul 25, 2022 at 16:46
  • 6
    $\begingroup$ . . . (a million years later) . . . 4857261 nanometers. If I understand the asker is trying to find a better way to write it down. $\endgroup$
    – Daron
    Commented Jul 25, 2022 at 16:47

It can't be done (even in theory).

The bigger problem is not how to display a single scalar (i.e. one dimensional) quantity in an easily readable manner that has a precision at the level of days.

It is more profound.

You can't have a well defined universal dating system for trillions of years that applies everywhere consistently.

This is because, as a consequence of general and special relativity, the rate at which time passes for different observers is different. The kind of system you are imagining only works in a world with Newtonian physics and a flat Euclidian topology. Those are often useful approximations on Earth, but not for these scales of space and time.

Time passes more slower in strong gravitational fields than in weak ones. Time passes more slowly for observers traveling at high speeds (approaching the speed of the light in the limit) than for observers traveling at lower speeds.

For observers who mostly travel at speeds far less than the speed of light and are mostly all traveling at similar speeds, and are in comparatively weak gravitational fields that are all very similar to each other, for periods of millions of years or so (e.g. pre-space travel observers on Earth), the discrepancies are tolerable. But, over trillions of years with observers all over the universe, you simply cannot have, because it is not that well-defined, an absolute dating system with a precision on the level of days.

When you are talking precisions of hundreds of trillions of days, even slight relativistic effects add up in circumstances where scientists today routinely ignore them.

You can only get that level of precision for that time period for a very well defined observer and to be useful, you really need to have very careful definition of the observer whose time measurement is being used. Frequently, someone trying to use the system won't have access to enough information to determine the correct date for a particular observer even if that observer is well defined and there is a process for singling out a date from the perspective of that observer.

Simultaneity, in general, is not well defined over universal distances in long spans of time, for many observers.

According to Einstein's special theory of relativity, it is impossible to say in an absolute sense that two distinct events occur at the same time if those events are separated in space. If one reference frame assigns precisely the same time to two events that are at different points in space, a reference frame that is moving relative to the first will generally assign different times to the two events (the only exception being when motion is exactly perpendicular to the line connecting the locations of both events).

Yet simultaneity is a core axiom of any one dimensional dating system.

What Is Possible?

If you are less ambitious, and only want this system for say, one planet, and you defined a particular location on this planet that was the reference point, you could do well enough. But there is too much variation throughout an entire universe for any system to work over the entire universe (and you need fast movers to have any meaningful interstellar travel which the premise seems to imply the need for).

There are lots of good choices here. For example, the duration of a full rotations of the target location on the target planet around the center of the galaxy (about 200 million years) at the time the system was established could provide one good intermediate unit between planetary years and trillions of years. In one trillion years there would be 5,000 such rotations, which would be a manageable intermediate unit.

Likewise, between a galaxy rotation and a year, a span of say, 10,000 years per unit (India's number system likes division of 10,000 more than divisions of 1,000 for example), you get 20,000 cycles per galaxy rotation. So 5000-20000-10000-365. Four divisions of 10,000 each would be a bit neater: 365 days-10,000 years-10,000 ten millennia eons-10,000 (galactic) cycles. A ten thousand unit item can be done in four decimal digits since 0000 is a possible value. So AAA-XXXX-YYYY-ZZZZ. Also, some periods of time matter more than others, so you could probably have abbreviations that simplify matters rather than spelling out the whole things all the time. There might, for example, be ZZZZ possible galactic cycles, but in a historical account, you'd normally have lots of dates in just one of them, so you might abbreviate with a world like IMP for the imperial dominated cycle, or BIG for the Big Bang dominated cycle, or CE for current cycle (that could also have a different meaning if used, for example, in the YYYY field).

This wouldn't be perfectly defined for other locations even in the same galaxy, but you could have a mapping system from the base location on the base planet to other planets that approximations the average travel time of a space traveler using interstellar transportation at the time the system was adopted to use a standard conversion based upon the spacial coordinates of the location of interest relative to the base location as a conversion factor allowing determinations to be made knowing only one's location in space relative to the center of the galaxy and the location of the base location. This would be arbitrary but workable. But that doesn't generally to the billions of galaxies in the universe since their spatial structure relative to each other isn't so stable. And, that would break down when your base galaxy colliders with another galaxy (as, for example, the Milky Way will with the The Andromeda galaxy in about 4.5 billion years).

  • $\begingroup$ Yes, time dilation occurs, but time dilation is extremely calculable. Someone travelling through a system faster or slower than another, does not invalidate the counting of that system. When Cooper & co land on an ocean planet near a black hole, it doesn't invalidate the Gregorian calendar; they just perceive it differently. "someone trying to use the system won't have access to enough information ... for a particular observer"; I hear you but over many years computation (quantum etc) will have solved this data collection problem (not to mention it has the reasonable protection of fiction). $\endgroup$
    – user97385
    Commented Jul 25, 2022 at 17:23
  • 4
    $\begingroup$ @futurelamb The problem is that at the level of generality of all observers in the universe, the time dilation is not unique between two points. It is path dependent. It is not, in general, possible to say that today on Earth is before or after a particular day that someone is experiencing in another part of the universe. It isn't just a calculation problem. And, if you go back far enough, local observers won't know where they are with enough precision to mark dates in a way that can be converted well by any defined system of conversion that is inherently non-robust. $\endgroup$
    – ohwilleke
    Commented Jul 25, 2022 at 17:31
  • $\begingroup$ "not ... possible to say that today on Earth is before or after a particular day that someone is experiencing in another part of the universe"; See here. I can calculate the time difference, and therefore the date relative to a starting point? I can take mars & earth time from 1 AD and calculate the diff within the same calendar. This is 20th century math, what about in a million years? $\endgroup$
    – user97385
    Commented Jul 25, 2022 at 17:52

I'd go with option 3 you listed: breaking up the year into more groups.

So today would be July 23, 2022, 0001. With 0001 being the millennium.

It keeps the familiar date formats, and much like our current date system it would be common for people to leave out the millennium when talking about nearby dates, same as if I say July 25th, you know I probably mean 2 days from now, not 2 days and a year from now.

After that counter approaches 9999 we either add another digit, or break it up again into how many 10000 millennium have passed.

There's no need for anyone to learn an entirely new date system, and no need to rewrite everything that existed before in the new date system - anything missing the millennia number must have been from 0001. Just like we know anything using a 2 digit year is probably from the 1900s.

Since your millenia field is going to get unwieldy using it over many trillions of years, maybe don't use base10 for it. Using base36 (counting 0-9, then a-z) you can fit a lot more numbers into a much smaller field. It's not as easy for humans to do math with, but it's easy enough to put dates in order and see whether the difference is huge or smaller.

A hundred millennia (100,000 years) from now would be July 23, 2022, 2s. A trillion millennia (1,000,000,000,000,000 years) from today would be July 23, 2022, cre66i9s.

  • 3
    $\begingroup$ @futurelamb You don't have to use 0-9 for your number system. You can keep the years field to numbers for backwards compatibility, and use base36 for your millennia counter. A trillion millennia from now would be July 23, 2022, cre66i9s. Not quite as easy for people to do math with, but easy enough to fairly easily see if a date is bigger than another, and which order they go in. $\endgroup$
    – Grant
    Commented Jul 23, 2022 at 23:27
  • 1
    $\begingroup$ "base36 for your millennia counter" - superb thinking. July 23, 2022, cre66i9s is the sort of length I am looking for and I think it would match the target audience. - very interested in this idea, I will think more on it - thank you! $\endgroup$
    – user97385
    Commented Jul 23, 2022 at 23:32
  • 3
    $\begingroup$ A small argument against base36: so numbers & letters really much look alike. 1 is very easily mistaken with i or l (depending on the case), and similarly 0 and o and 5 and s, and to a limited extent 7 and i/l. For this reason, you may want to go with base32 instead: remove the digits 0, 1, 5, and 7 to avoid confusion. Otherwise: great idea with the change of base ;) $\endgroup$ Commented Jul 24, 2022 at 11:24
  • 1
    $\begingroup$ @MatthieuM. That's a great point. But I'd get confused with just a few numbers missing, so maybe opt for just letters in base26? $\endgroup$
    – Grant
    Commented Jul 24, 2022 at 15:18
  • 3
    $\begingroup$ If you want historical precedent, of course, the Babylonians have you covered with base 60 - although you'll want to use less unwieldy characters for your digits, since you're probably not using clay tablets. $\endgroup$
    – Ottie
    Commented Jul 25, 2022 at 11:31


Fortunately, such a system already exists. Internet Protocol version 6 is the current addressing format used by much of the internet for communication with other devices. It's format is as follows:

IPv6 addresses are represented as eight groups of four hexadecimal digits each, separated by colons. The full representation may be shortened; for example, 2001:0db8:0000:0000:0000:8a2e:0370:7334 becomes 2001:db8::8a2e:370:7334.

They provide the amount of numbers you need, and then some:

IPv6 uses 128-bit addresses, theoretically allowing 2^128, or approximately 3.4×10^38 total addresses.

Your first date of 01/01/483297555721815 converted to hex is 1b78e7b212e57, which would be represented as ::::1:b78e:7b21:2e57. I propose that when read aloud the colons could be pronounced "col," and several in a row can be pronounced using Latin/Greek prefixes to avoid ambiguity. So you'd pronounce it "the first of January, tetra-col one col b seventy eight e col seven b twenty one col two e fifty seven." Still not the most fluid, but it gets some rhythm to it and I'm sure you could internalize them with a bit of repetition. As another example, today's date is "the twenty fourth of July, hepta-col seven e six."

As a bonus, this is very computer readable.

  • 3
    $\begingroup$ So much for making my fortune by patching the Y1M bug. 😄 $\endgroup$ Commented Jul 24, 2022 at 9:52
  • $\begingroup$ but why include the colons at all? given that you could just pronounce the hex number $\endgroup$
    – somebody
    Commented Jul 24, 2022 at 19:40
  • 1
    $\begingroup$ The reason it's useful for IPv6 to collapse all-zero segments (therefore shortening the number for reading and writing) is that for technical reasons, it's extremely common to have a lot of all-zero segments. That wouldn't be true in arbitrary dates, and so this technique would have little value there. $\endgroup$ Commented Jul 24, 2022 at 23:49
  • $\begingroup$ A good overall answer, but you make it more confusing when you pronounce "7b21" as "seven-b-twenty-one". The hex sequence "2-1" is the number thirty-three, and twenty-one is represented in hex as "1-5". Best to avoid the ambiguity and pronounce it as "seven-b-two-one". $\endgroup$
    – bta
    Commented Jul 26, 2022 at 20:52
  • Keep in mind who needs what from your notation:
    • Characters in the story usually need the next couple of days, the next couple of dozen days. "Get me that report by the 42th of Sometember." They might need more, but rarely more than one lifetime.
    • Readers of the story need to get a sense of "not in Kansas any more" -- but not coupled with total bewilderment. They have to follow plot-relevant dates.
  • Except for dates (and measurements in the US), mankind has gone decimal. There are some very good points for octal or duodecimal systems, but decimal is it for your readers.
  • Most of the Western world reads from the left to the right.
  • When one is reading a long string of digits without the checksum of spelling, positional mistakes will creep in. When I write a word like internatonialization, you will understand it despite the spelling error. When I type 053549797435, you have no way of knowing that it should have been 053549779435.

So come up with a system that puts the smallest unit left, followed by blocks which meet human reading and numerical patterns. Three digits per block is convenient because there are less than 1000 days in the year, and blocks can be pronounciated 'thousand' and 'million' and so on. 123-456-789-101-112-131-415 is day 123 of year 456 of millenium 789 etc.

  • 7
    $\begingroup$ I'm most likely biased by living in a country where [CC]YY-mm-dd has been the default date format since before I was born, but I'd argue that the greater unit should be to the left. If nothing else, then for consistency with how we specify time and basically any number. $\endgroup$
    – user95279
    Commented Jul 23, 2022 at 21:46
  • 2
    $\begingroup$ @futurelamb I definitely get that point, and I thought about it both before and after posting my comment. And like I said, I am most likely biased, but having the lesser units to the left seems backwards to me, to the point of almost being awkward. So, let's agree to disagree. :) $\endgroup$
    – user95279
    Commented Jul 23, 2022 at 23:19
  • 1
    $\begingroup$ "...let's agree to disagree. :)": I don't disagree with you - I am grateful for your thoughts as I am (clearly) biased the other way. I hope I didn't come across as dismissive. I suppose it becomes an exercise in deciding what my target audience might favour more, since ultimately, it is the goal of my question to reduce mental fatigue on the reader, while remaining specific. $\endgroup$
    – user97385
    Commented Jul 23, 2022 at 23:26
  • 1
    $\begingroup$ yeah i think smaller -> bigger is a great idea, especially since you'd be able to just truncate the year to ignore anything that's current (current century, current 10k years etc) $\endgroup$
    – somebody
    Commented Jul 24, 2022 at 19:42
  • 1
    $\begingroup$ bigger-to-smaller sorts better. That is Y/M/D is better. $\endgroup$
    – Tony Ennis
    Commented Jul 25, 2022 at 19:25

Put , after every few digits to break up big numbers. People already do this to make it easy to read large numbers. Similar segmentation is done with - in phone numbers or social security numbers, and in credit card numbers.

If people in this world really care about 18-digit dates, they will find a way to remember them. Maybe people will make up songs for different dates. Commercials do this will phone numbers.

A lot of the other schemes you propose violate one of your requirements. Hashes are not human readable. Scientific notation compromises accuracy.

  • $\begingroup$ alternatively / after every few digits: 25/07/22/20/etc... which seems to have been suggested in another answer already $\endgroup$
    – somebody
    Commented Jul 24, 2022 at 19:43

You don't have to use base 10, you're already on that track with your a-z thinking. Base 64 is common for storing data - but could also be used for counting. If you imagine using the span 0-1a-zA-Z for a number - you'd shorten the date by more than half. You cold add letters from other alpahbets; greek, cyrillic, brāhmī, arabic and go for base 256 or some other significant number.

Going for 256 your number 23365535731776 would equal (21x256^5)+(64x256^4)+(54x256^3)+(117x256^2)+(16x256)+64 ie your representation of 21 64 54 117 16 64 for example kβRбfβ.

Edit: Mixing uppercase and lowercase and having characters pronounced the same isn't a good idea for vocal communication. (k-beta-R period, year be-f-beta)

  • $\begingroup$ I agree with this. It's a hell of a lot easier to learn 100 or 200 characters well and have short, pithy dates of a few characters long. I reckon go with whatever number system's base is in use squared or cubed so that it's intuitive for dates near ones own lifetime. If base 10 is used for numbers, use base 100 or 200 for dates. Or an exponent of 2, as many suggest. $\endgroup$
    – user86462
    Commented Jul 26, 2022 at 10:25

What about breaking it down to bigger pieces ?

For example, you could add the unit "Sun", which would be the lifespan of the Sun (estimated to be 10 billion years). Another unit would be an "Ice age", which would be 100K years (approx time-span between ice ages).

You can find better units, obviously.

With your example of The first day 65 trillion, 77 billion, 10 million, 869 thousand 222 , this would be:

The 1st day of Sun 6507, Ice age 70108, Year 69222.

And written shortly: 01/69222/70108/6507, or 1 69222y 70108i 6507s.

Still a lot of numbers, but separated into larger "eras".

  • You can still combine them easily to get the full number.
  • You have some broader numbers if needed ("a period of 3 Suns", "around ice age 10.000 of the 3rd Sun").

I feel like these groups of 4 to 5 numbers are easily readable, and you still use an easily readable base 10. Lots of numbers, but separated into broader chunks.


RFC2550 does not have widespread adoption (and is an April Fool's RFC), but proposes an expansion of the existing ISO8601 standard for recording dates to fix the problem that the string 10000-01-01 comes before 9999-12-31 (because 1 < 9).

It proposes adding a letter to the start of the string, so the date would be A10000-01-01, which gives a potentially useful "when in the universe are we" era marker without having to count the number of numbers.

But this doesn't solve the problems of dates that might be exceptionally long, and there's no way of compressing this data. We can give a short name to a millennium and recognise that there are billions of trillions of millennia named the "cromulent knapsack" era, but we established that the one we're talking about starts in H329910713000, and we're probably not going to have a name collision with any of the other so-named eras. There could also be a custom character for that era, similar to how was coined for the start of the Reiwa era in 2019 in Japan; these would need to be automatically generated like Github Identicons (because otherwise we need trillions of terabytes for all the images!). Although they still don't have enough detail for a human to identify the millenium (only about 24 bits of information compared to 70 bits to uniquely identify a millennium), one could imagine autocompleting the precise date if one has used the same date recently.

This is a hard problem. Good luck!


So, I haven't fully tested it out yet but my first thought is to use a logarithm to make the bulk of the year more manageable. Here are some examples where I rounded down to the millenium, took the log100 of that number, and then added the years I removed for the rounding back to that. This is displaying millenium log100 + years in millenium / month / day.

23,365,535,731,776 = 6.684287872 + 776 / 1 / 12

23,365,535,731,777 = 6.684287872 + 777 / 1 / 12

483,297,555,721,815 = 7.342107299 + 815 / 1 / 12

383,293 = 2.791599387 + 293 / 1/ 12

This produces a number that you can use as a reference for the era and compare pretty easily to other eras and can be converted to an exact year with a couple of simple operations. If it's between 1 and 2 then it's between years 100 and 10000. Between 2 and 3 is between 10000 and 1000000. For every whole number that it increases, add two zeros to the final year.

I'm sure someone better at math than I am can flesh this out a bit.


Dealing with such large dates and timespans without mental acrobatics is inherently difficult, because we are ultimately not made to think in the scale of billions of years.

Therefor I'd suggest to break it down into segments that are clearly separated, to make it easier for humans to read and quickly judge if a given date is further in the past than an other date. Since we already have a separation system at thousand that comes natural to most people, I'd go with this and expand upon it by alphabetical qualifiers. Also omitting the month and just use a fraction of the year to specify the day (with a year having 365 fractions) and using a Format that is something like this: "Fraction of Year" / "Year" / "Millenium" / "Million Years" / "Billion Years" / "Trillion Years"

Qualifier Example: K = thousand, M = Million, B = Billion, T = Trillion.

Your example of The first day 65 trillion, 77 billion, 10 million, 869 thousand 222 then would look like this: 001.222.K869.M010.B077.T065

Characters can then ommit anything either on the beginning or the end if it suits them. Like if its clear they are talking about the current millenium (from their perspective) they can use 001.222 and if they for example know something happened somewhere in 10th million but don't know the exact date yet they can just use M010 and everyone knows they are talking about the timespan from 001.001.K001.M010 to 365.999.K999.M010


Abbreviate it.

We already do this. Today is July 25th, '22. We don't necessarily have to specify "2022" because in the overwhelming majority of circumstances we know which 22 you meant. You would use the long form if the date might be ambiguous "The company was founded in 1902."

I have to imagine the people of the future would still do this. July 25th of the year 22,406,978,455,022? "July 25th, '22".

There might even arise a convention of only using the "significant digits". If it was a century before the above date, it would be printed as "July 25th, '4922" with the single quote denoting "all other digits are the same as today".

It would be something really unusual to require listing exact dates, or to have that come up frequently. "The star went supernova 22 billion years ago" is good enough for conversation. (They may know the exact date was 22 billion, 58 million, 926 thousand, 47 years ago but conversationally, "22 billion" is good enough for most purposes.)

Another possibility is that they simply keep resetting the date back to 0, for this reason, based on whatever (new emperor? Dates reset to 0.) "Long notation" might only be used by scientists and their day 0 is "the big bang". Kinda like how we use F or C for temperature but scientists may prefer Kelvin, which most people never use and wouldn't (off the top of their head) know how to convert.

  • $\begingroup$ very similar to my idea. "Big Bang" would certainly be a "named date" $\endgroup$
    – Tony Ennis
    Commented Jul 25, 2022 at 20:15

The date system we use now has been honed to be efficient for most of what we do. So I'd keep it. For example, if I said something happened in '22, we could make assumptions about the century from the context. When that isn't enough, I'd add a century such as 1922. Whatever system is created here ought to be practical enough to allow such shortcuts.

With a span of trillions of years, we can assume that humans (and their descendent species) have traveled to the stars and will have many local date systems. They all would be translated back to Terra Standard for certain official documents. Otherwise no one would use Terra Standard Time (TST).

So local times are handled - they're local, just like the date we're familiar with, with all the shortcuts we'd expect.

Thus the only users of TST would be AIs and computers. Nothing in the span of humanity lasts long enough, or is important enough, to commemorate an event to the day after a few thousand years. Does it really matter if Stonehenge was completed on August 12th, or 13th?

The only time I see it mattering is for law and banking. When ownership changed hands (the former) and accrued interest (the latter).

Shuffling all this around, then, Terra Standard time is:

The number of 10,000 year epochs The number of years in the current epoch The number of days in the current year The fractional day as a decimal

Easy. Since humans won't use the full format, it doesn't have to be human friendly per se, but just has to be easily manipulated.

Today's date in TST would be 0|2022|220 assuming we use the more or less standard of the beginning of the calendar. Or maybe 2022220 since the date would always be 7 digits. Not all cultures might agree with that arbitrary beginning of the calendar, however, so one science fiction writer, I don't recall whom, suggests the start of the calendar could be July 16, 1945 at 5:30am, the date of the first atomic explosion. That TST date would be 1945211 if we keep our current beginning of calendar, or 0000000 otherwise. I suppose Trinity could be 1945211.23 if that fraction of a day is precise enough.

Certain dates would have names. 1945211 could also be called Trinity. Most current events that seem momentous (Kennedy assassination, invasion of Ukraine) don't actually matter much in the scheme of things. Another date that could be worthy of a name could be the imminent Betelgeuse supernova.

There's a little cheat in TST being the number of 10,000 year epics, then the year of the current epoch. Base 10 being what it is, the epoch and year, together, is just the number of years since the beginning of the calendar. For example, a million years from today is 1002022220. That is, 1,002,022 years and 220 days.


Your civilization uses the 64-bit (and maybe even eventually the 128-bit) Unix time for official recordkeeping, and people just modulo down to 3 significant year digits for everyday use.

Assuming your civilization never goes through periods of such destitution that computers go out of manufacture, every computer ever made will keep merrily counting up the seconds since January 1st 1970. With 64 bits to do it, they'll keep counting for nearly 300 billion years. (I'm sure programmers in the year 292,277,026,596 will curse the programmers of 2038 for not jumping straight to 128-bit.) It seems very likely that digital systems will remain integrated with society for as long as society can support digital systems' supply chains, so the Unix time won't just be an academic curiosity, it will be a critical timekeeping system.

At the same time, if human lifespans cap out somewhere in the 100-200 year range, nobody will care about the 1000s or higher year digit in everyday conversation. We already basically don't care about it, when was the last time you heard someone talk about an event prior to the year 1000 outside of a history discussion? If the year is 252525, people will just say it's 525. Maybe when the millennium rolls over people will keep saying the thousands digit for a century because the year "22" looks silly, just like we are doing right now.

But computers will keep dutifully putting full Unix timestamps on everyone's files, and it will remain possible to date things from the distant past down to the second. That's immensely useful, so people will still think of themselves as being in the year eleventy-billion with all those dated files in their past, even if they're not saying the full year in conversation.


Humans are much better at dealing with graphic representations than they are with numeric/symbolic ones. You can ignore computers because it is usually trivial for them to convert representations. Being unable to just speak out the value could be a problem, however. If you are allowed to use graphics, you can use colors and geometries to pack more raw data into the representation, as well as positional relations. Oh, and also humans are much better at percepting things when they fit within their view (which is tiny, like 5 letters or so), so fitting things in a square/circle is better than in line. And humans are better at remembering things when they can associate them with something, like a name(even if its just a single letter like alpha).

The other side of the coin of precise and efficient data transmission (even when it deals with humans) is the raw data itself. We want to minimise the amount of data transmitted, and amount of time needed to convert the data into information. Instead of using absolute encoding, you could use relative: you pick a reference date, perhaps giving it a name (even if temporary, unique to each conversation or even phrase), and then say how far away the one you need is. You can even create a system where representations integrate that. I recommend looking at the chineese language.

Truly, your problem is incredibly close to raw information communication problems, so you could try looking at how mathematicians and programmers deal with it, even if it is intended for computers.

  • $\begingroup$ Colours as markers is such an interesting idea; I'm imagining a system where rgb16[0,0,0] represents the moment of the big bang, and each one million year era after that, the colours increase through the spectrum [0,0,1] -> all the way to [65535, 65535, 65535] being some arbitrary point in the future beyond life. This has the added benefit of being quite a nice system to discuss generally - "during the 'sky blue' era" and also very specifically when required using the exact rbg numbers. The systems presentation will be dictated by the requirements of the user; elegant! $\endgroup$
    – user97385
    Commented Jul 25, 2022 at 14:45
  • 1
    $\begingroup$ @futurelamb Just using one color would be incredibly inefficient, not to mention that rgb16 is beyond modern humans' perception; modern computers use rgb8! Besides that, you would need to somehow spread all the dates over the spectrum. You can imagine it as a cube with axes of RGB, and then trying to lay a string such that far away parts of it are not too close on the spectrum... Best we have is the Morton order I think. Besides that, just having two colors together would increase the available bandwidth by a power of two! Use multiple systems in conjunction, just don't overload a single one. $\endgroup$
    – user369070
    Commented Jul 26, 2022 at 9:22
  • $\begingroup$ graphicdesign.stackexchange.com/q/3682/2104 - there are only so many human distinguishable colors; likely capping out at around 26. If using computers, colors are less efficient than packed binary (while packed binary can be interpreted as consecutive pixel colors, you'd still need a header indicating when the image wraps). $\endgroup$
    – Iiridayn
    Commented Aug 18, 2022 at 18:35

What you really need to ask yourself when facing such time systems is what are people actually going to do with it? It doesn't matter if you design your nice shiny Julian calendar to be regular if people are going to do calculations based on the orbital period of the Earth (which drifts substantially in the Julian calendar). It doesn't matter if your nice accurate Gregorian calendar fixes these issues if your religious celebrations are tied to a lunar calendar instead.

It doesn't matter if you are capable of counting the number of seconds between two events a trillion years apart if nobody actually wants to do that. So my recommendation is to think carefully about what operations people will actually want to do to your dates.

A few things I've explored which may help:

Not event or loose era based. E.g. LotR uses 'First Age' and SW uses the 'Battle of Yavin' as date markers. I suppose a year (2022) is an 'era marker' in itself, but it is consistent and logical and one can identify its neighbours just by knowing the system.

You notice the oddity of "the year 2022", which might be treated as an era marker, but has some properties which make it fundamentally "better" than other markers. And you're right, except the era marker is not "2022." It's hidden. The era marker is January 1, 1AD, from which you can use simple arithmetic to measure time. Our system of "years" is based on an Epoch of an arbitrary date that we simply all agreed to use. Well, most of us agree. The astronomical community doesn't use this system. They like to use "Julian Days", a measure of the number of days since January 1, 4713 BC from the Julian Calendar. Why that date? Because it was arbitrary. The individuals who suggested the date noted that if you take the solar calendar, the lunar calendar, and the ancient Roman calendar for taxes, you end up with a 7980 year cycle. 4713 BC was chosen as a time they lined up and was before any recorded history, such that all historical events had positive day numbers.

Why mention these? Both year numbers and Julian days share a common trait: they have a single epoch date, however arbitrary, and a numeric value representing time since (or before) that epoch. The epoch is arbitrary. And they lead into the line of reasoning you are: you fundamentally need a number large enough to cover all time, and a number precise enough to measure the durations we care about. You cannot get away from that.

Indeed, there is a fundamental give-and-take between accuracy and the data content of the string describing the number. It doesn't matter if you encode it as "acrfz.001/69222/0108" or the year "65,077,010,869,222" You will have to fight this balance. In astronomy Julian Dates are sometimes not precise enough, so it is common to add the necessary digits such as 2415020.31352 (the julian date of the Bessalian epoch)

So what are your users of this system going to actually do with these numbers? What is the important information content of the date? Do you actually care that acrfz.001/69222/0108 is some number of seconds after acrca.002/6920/0100? Or do your users of this system care that acrca.002/6920/0100 is before the ascension of the God Emperor and acrfz.001/69222/0108 is after?

I would argue that you do not want one system, but two. You want a simple single-epoch system based on an arbitrary epoch of your choosing, and you want a "event" based marker. You specifically stated that you don't want them, but you do. Almost certainly you do. The event based dates would be used for the 99.99999% of date operations people would ever want to do. For the remaining 0.000001%, you can fall back on the single-epoch system like we do today with Julian Dates. On all of those times, your human will have a calculator handy to assist them because there is no way for a human to do the arithmetic on dates trillions of years apart, no matter how clever your encoding.

You then fix the "events" at a specific time since the grand-epoch of your trillion year calendar. On the rare case you have to convert, there is an agreed upon conversion between them.

Which is more useful, I ask: acrfz.001/69222/0108 and acrca.002/6920/0100? Or human.2022/03/05 and robot.0001/02/25, where "human" and "robot" epochs are separated by the rise of AI? Only very rarely do you need to convert those to Julian Dates (perhaps to find out how long it was since we darkened the skies).


As others have said, use a thousands separator as we currently do for very large decimal numbers. Among things used currently on Earth: a comma or a period. Alternatively: a dash or slash. The International System of Units dictates use of a thin space, but I don't think that's very eye-catching.

Personally I would recommend the period (as used in many European countries), as it looks elegant and science-fictiony.


Do it like with star names

The timespan is humongous. It's so large that even epochs have stopped making sense: If you started defining epochs, you'd have more of them than anybody could remember, let alone name or associate any meaningful information with them.

So people will concentrate on events of interest.
Those that have something making them stand out.

And those events are rare enough that they will get names.
In fact nobody will care about the digits that make up the exact date they happened on; you have lists where each event has its exact date/time noted down, just like nobody remembers the exact sky coordinates of Betelgeuze, you just look that up.

And it will be just like stars in other respects, too.
Just like there are "blue giant" stars, there will be "XXX events" of some specific significance, and a wealth of similar but slightly different ones, and there will be classes of events that are a total mystery (until some revelation puts them into perspective).
Scientists will find patterns in the events, and make predictions about when/where exactly to find related events, and most will fail, some will find a pattern but it's pure accident, some will find a real pattern.
And everybody will wonder if the currently-known set of patterns really covers all relevant events. There will be theories about "what's out there but we just don't look in the right place/date, and we don't have the resources to systematically scan everything".


If you assign blocks of time a single character, it will be easy to compare two dates against each other. a = 0 - 10^6 b = 10^6+1 - 10^12 c = 10^12+1 - 10^18 d = 10^18+1 - 10^24 and then make smaller segments of each span for the next character f = 0 - 10^2 g = 10^2+1 - 10^4 h = 10^4+1 - 10^6 with the final digits being the "local" date/time that is relevant to a person's experience (i.e. Yesterday, last year, my birth date, etc...)

Something along the lines of how numbers are assembled now: 1234 = 1000 + 200 + 30 + 4 Each number has an associated multiplier with it, we just commonly use 10 because of 10 fingers. You can use anything you want to shorten the number of digits required and if you use characters and digits you get a bigger "name space" that is still easily compared visually.

bg:2020/02/26 is a date in the year range of 10^6+1 - 10^12 but has a more specific section of that group of 10^2+1 - 10^4 which is the second group within that broad segment. You can break the ranges into however many "buckets" you want to keep the number of characters in a date as large or small as you want.

I suspect the most important thing to the characters will be relative comparisons of dates, so making it easy to know that two things were "close" in time or "far apart", in an easy and fast way.


As there are more than 100,000 (10^5) Chinese characters, and there are ways of assigning an 'alphabetical' order to them you would only need four characters to describe up to 10^20 years. So very space efficient.


Prior Answers Have Good Warnings/Ideas

Agree with prior answers that such a scheme would be useful primarily in academic contexts, not in casual social interaction. Additionally, such a scheme indeed needs a well defined observer, due to relativistic effects, if times are to be meaningfully compared beyond the planetary/solar system scale. Cosmic background radiation lacks the specificity (based on instrument abilities) for precise comparisons at the date level. Unix timestamps record only their local time; but with appropriate calibration (and likely conversion to handle relativistic effects) such computer time should be a good approach for tracking events. Agree also that most significant figures should always be first or last. Dumping days in between months and years is poor Science Fiction, and I know of only one country (with less than 5% of the global population) on the planet which writes dates so irrationally.

With respect to the root of a timestamp system

Unix timestamps count in seconds since an epoch - a fixed date and time. I think setting the base time and duration to 1 year of earth's orbit in the year 5000 is not a great choice, but it's about as bad as many others. More importantly - months are meaningless in this context. I'm sorry, they just are. Days are as well, especially given they don't match up with years nicely. You'll want to pick a single root measurement and have everything else be based on it, in whichever base you choose (decimal if base 10).

With respect to the date root of significant events

Let's suppose your hypothetical superpowerful deity has some attachment to ancient (ancient - trillions of years ago!) Earth months, years, and days for some reason. In that case, the characters should probably have to run a hyperpowerful computer search algorithm to discover these patterns and the associations, just to learn that ancient Earth months matter at all. There is no reason they will continue to be used trillions of years later. Their computers will be better though. Subsequent representations may then be warped to fit Earth dates, but I would argue that that should imply something more than months and years should be at play - why not include the lunar cycle, for example? It doesn't match with years (or months; months are meaningless divisions which poorly approximate the lunar cycle). Week numbers, possibly, or days of the week. None of these are represented in simple numeric time notation.

Say this hyperpowerful being has a thing for Sundays (if it's sentimental about Earth, one can suppose it's played God, and at least one popular God has a thing for Sundays). Every event happens on a Sunday on the Earth calendar. Why would this being limit themselves to the year 5000 representation of time? Would they not have a preference for the 1st day of the 7 day cycle in Earth's current (instead of fixed) timeframe, assuming Earth still exists? Would they not rather prefer to extrapolate Earth's orbit and rotation towards the modern era?

Instead of sentimentality, say the hyperpowerful entity is a computer programmed in the year 5000 with fixed programming (and is somehow relevant trillions of years later). In that case human biases towards months, weeks, and year 5000 year durations may be understandable.


Given the above, that someone will have to discover the pattern, and any meaningful measurement system will likely need to be derived by the characters for this specific purpose, you have a lot of flexibility in choosing the representation. My focus is not on measuring, but on the presentation for the sake of the characters and readers.

Research has shown that representation may not help that much in getting users to recognize small differences in high entropy values (your massive dates are one such example - see "Can unicorns help users compare crypto key fingerprints?" as one example). This applies to both your readers and your characters.

Your base26 system (lowercase letters) is a good idea; readers should be able to understand it pretty intuitively (more than a base 36, 62, or 64, etc scheme). However, if significant events happen every increasing power of 10 years, it would be better to choose that (probably with exponential notation, something like 10e24 + 78953274934173947394793-11-23); essentially, the frequency of significant events should shape the date system, since it'll need to be bespoke to the characters anyway. Every power of e years? e^112 + yyyyyyyyyy-mm-dd, etc. Every 100,000 Thursdays? Yep, you count Thursdays with dates such as 764423 * 10^5 + 68234 Thursdays. Etc.

Whatever is the detected root of the frequency should be the base of the measurement; not planetary years as of 5000 AD, unless that genuinely is the true root of the event repetitions (and why not 4998 AD, or 5257 AD?). Months should similarly not matter at all, again, unless the entity itself has chosen them. Days, again. Lunar orbits, similarly. Etc.


You must log in to answer this question.