▲ 140 points
back
51 comments
I am a fan of using Cistercian for date stamps, with a glyph for the year, month+day, hour+minute. Like this: <https://me.micahrl.com/blog/cistercian-dates/>
I also made a little decimal to Cistercian translator once. The numerals aren't in Unicode, so the font I use has to make use of private code points, which was kind of fun to get working properly. <https://cistercian.micahrl.com>
How did the Unicode people possibly miss Cistercian numerals?
Well there is a background document for future consideration: https://www.unicode.org/L2/L2020/20290-cistercian-digits.pdf
However, it says:
> It is not clear that there is any current need for Unicode support. Western Michigan University’s Medieval Institute (Cistercian and Monastic Studies) has a digitization project of Cistercian mss, but as of end 2020 they have not expressed any desire for computer encoding of the numerals. This document is therefore background information so that work is not duplicated if a need arises later.
Unrelated rant: The Unicode people have missed a lot more than that. There are an enormous amount of, for instance, musical accidentals that are missing, including ones that have been in use for at least a hundred years, such as the half flats or half sharps in e.g. Arabic or Turkish music, which are used by probably hundreds of millions of people in the Middle East. (They do have a symbol called "half sharp" and "half flat", but it isn't the same symbol.)
But at least they have the multiocular O, I guess.
If there's something missing from Unicode, you can propose it and there's a good chance they will add it to Unicode. Source: I got 7 characters added to Unicode.
We also have a poop emoji and no Tengwar
Unicode afaik focuses on taking already existing & supported character encodings, and incorporates those. Given there are 10,000 distinct glyphs, without some group of people already pushing for a specific encoding because it's what they use, Unicode probably doesn't have that strong a motivation to create their own (only to find out that it doesn't meet the needs of real users).
At least Unicode today doesn't need 10,000 codepoints to encode 10,000 distinct glyphs. Cistercian might make sense of ZWJ sequences of existing numerals plus a "cistercian mode selector". Authoring font ligatures to support that might not be fun, though, but also the cistercian glyphs are pretty simple concatenative structures almost perfectly designed for simple ligatures.
You're right, that does seem a better way to encode these—but that kind of encoding design I believe Unicode is wary of doing themselves. They seem to prefer sticking with a community-picked encoding instead.
Luckily Unicode makes exceptions for charsets/languages that are overlooked, also they take encouragement (via joining the Unicode Consortium) if you have a special symbol that you need to add for a private interest.
How do you tell the difference between 10,000 and 100,000 and 1,000,000?
If you like that, check out my favorite Cistercian numeral library: https://github.com/hsablonniere/cistercian-numerals
Here's a demo of it in action on the author's blog: https://www.hsablonniere.com/a-clock-based-on-cistercian-num...
I ended up using that library to create a simple Cistercian clock screensaver. My hope was that I'd learn to read Cistercian numbers fluently though I'm still working on that.
OMG, you made my day. Glad it helped you ;-)
Reminds me a lot of the D'ni numbers from Riven.
D'ni numerals overlay two base-5 values in the same space, one horizontal & the other vertical, to represent 0–24. There's even a one-character numeral for 25, though it rarely comes up in-game. Different, of course, but I wouldn't be surprised if Cistercian was the inspiration.
Another number system, which I learned about recently on HN, are the Kaktovik numerals, which are even closer to D'ni. Interestingly, they were developed just a couple of years before Riven was released.
https://en.wikipedia.org/wiki/Kaktovik_numerals
It's not a base-5 system exactly. It's base-20, where the rotated-and-superimposed digit is only one of the first three numerals. I guess you could call it a mixed-radix system, where the bases alternate like 4 5 ... 4 5 4 5 and where each group of 4 5 combines into a single symbol.
It’s a fun task to generate these from the five base shapes.
I like these.
I feel like there's a missed symmetry here though and almost symmetrical things are like a wobbly tooth in my brain.
The right line appears at 6, but if you added that at 5 you could repeat the pattern you did for 1-4 for the horizontal lines, which it almost does. 7 and 8 are 6+1 and 6+2.
I'd make it binary-like, like another commenter said.
Change 3 to 2+1, 6 to 4+2.
8 takes the current 6, and 7 becomes 6+1, 9 becomes 8+1 (current 7). That also gives you hex digits to work with for free, each symbol could store up to FFFF.
It wouldn't change a lot, just removes the downstroke from 3. Of course, 4 bits is an arbitrary choice, one could keep the current 5 bit/segments as well, but not much is needed to normalize it to 4 bits, which is interesting in itself.
There is also a Numberphile video going into their history [0]. About monks with a vow of silence.
Related:
Show HN: A CLI tool to generate Cistercian numerals - https://news.ycombinator.com/item?id=27853033 - July 2021 (5 comments)
Cistercian Numerals - https://news.ycombinator.com/item?id=26523363 - March 2021 (66 comments)
There’s something very satisfying about the idea that the numerals themselves can be generated algorithmically.
For example, here’s a nifty Cistercian numeral generator: https://shelby.cool/#/sketch/cistercian-numerals
One SVG to display 1 to 9,999 by using just a bit of JavaScript to toggle CSS classes to show/hide segments and update the SVG's accessible name.
The problems are that this is 1) hard to use for arithmetic, and 2) brittle.
"Brittle" because you're peacefully sitting there at 9999 coins or whatever, and you earn another one, and suddenly you can't represent it. Arabic (Hindu) numerals just slap another digit to the left, but Cistercian numerals break.
(You could fix that by using an Arabic notation - two Cistercian numbers, with the left one meaning "multiply this by 10,000". At that point, Cistercian numbers become a compressed Arabic notation. Still harder to use for arithmetic, but that's true of many compressed formats.)
> The problems are that this is 1) hard to use for arithmetic, and 2) brittle.
I wonder if these constraints are inate to the problem (inescapable trade-offs) or circumstantial to the particular choices of the Cistercian monks...
Any fixed-number-of-digits (or bits, or whatever) system is brittle against overflow.
Any compressed format is hard(er) to use for arithmetic. Pretty much any grade-school kid can memorize the addition table, and can work it out on their fingers if they forget. But that table only has 100 entries. The addition tables for Cistercian numbers has 100,000,000 entries; nobody will memorize that.
But maybe I'm thinking wrong. Cistercian numbers really only have 0-9, in four possible orientations. So maybe you can just add them, going from units up through thousands, writing down each quadrant as you go. (Multiplying still looks like a mess, though...)
It's like a 16 bit integer, you can only store so much…
This looks a lot like how Chinese characters are composed of individual "radicals", arranged like tetris blocks inside the same square box. Some of them affect the meaning of the letter, while others only signify the sound it's supposed to make. The resulting writing system is not easy for foreigners to learn, but it makes quite an efficient use of space. This might have been more valuable at a time when writable surfaces were expensive.
For your embedded cistercian projects: https://www.tindie.com/products/savageelectronics/rgb-cister...
(I'm not the maker, I just noticed these last year on Tindie and picked up a couple.)
Not my work, but I found this 3d printed digital clock using cistercian numeral system online. It looks like a cool project: https://www.instructables.com/Cistercian-Digital-Clock/
Until someone makes a Cistercian display automatic wristwatch I'm not interested.
here is some research that was done on adding cistercian to unicode https://www.unicode.org/L2/L2020/20290-cistercian-digits.pdf
you can still email the author. he would love for someone to continue the research
Oh how I wish the glyphs started at the top right and moved anti clockwise one quadrant for each power of two, to be consistent with right handed coordinate systems
Glyphs have to be written in an orientation; bidirection (or higher symmetric) layouts would impose either writing the words left to right & right to left, which would cause smearing when writing with ink with a dominant hand, or mirroring the glyphs themselves, which would make reading & writing words much more difficult.
Yup, and besides the counter-clockwise for consistency, I would like them even more if:
* Each stroke were 1 bit, so the numerals would be just a more "compact" representation for binary numbers.
* Even better if they can be represented with a seven-segment display (or other similar displays).
(Though I can't blame them, since these two things didn't even exist back then.)
This method could be used to compress numbers? You get a conversion from 4 to 1 digit.
As an interview question I might ask how to one might extend this system
Out of interest, what would you consider to be a good reply?
The 'obvious' answer is to have a second glyph (and third, etc.), but that's rather a boring answer.
Perhaps making a cross-shape so you have four-arms to work with (and can thus get to 10.000.000).
Or making a capital H shape would let you have the next four digits, which could then be extended out indefinitely; |-|-| etc.
Well, another obvious extension would be to prolong the vertical line, since it's already ordered by rows (two of them).
Or, indeed, you could make an array out of them, making a super-glyph (kind of like a block matrix). Four blocks of four digits would get you to 10^4^4 = 10^16 so 100,00,00,00,00,00,00,00 (or 10000,0000,0000,0000, or 10,000,000,000,000,000 depending on your grouping), which is 10 Quadrillions in short- scale. Obviously grouping digits by blocks of 2 or 4 makes mor sense, since that's a row or a glyph.
Interesting discussion though, which is probably the goal here, rather than a good or bad reply: it allows you to see how the candidate thinks or approaches a problem.
It’s intentionally open-ended to engage in discussions about functional and quality attributes. Does the candidate ask questions or just jump into an answer?
I love it because it’s very open-ended and demonstrates problem solving. It’s a great conversation piece.
You could use multiple dimensions. Or colours. Or fractals. Or any one of a hundred things.
But the real answer to any "how would you...?" questions is to reply with "Who are the users and what are their needs?"
Can you generate the higher valued numerals using OpenType ligatures?
That probably came from a culture where writing materials were really expensive. Did they use vellum?
Plus I can imagine that it also helped as a form of encryption...
It would be interesting to have a "Top-50" page of topics ranked by how many times they have been posted on HN.
IMHO, Having something reposted every now and then is good. This gives the opportunity for people to see it if they missed it the first time.
Judging for myself, this is the first time I’ve seen Cistercian numbers and although I’m a frequent HN user, I’ve never seen them before.
If a topic is interesting it doesn’t hurt to repost it :-)