A writing system that nobody has ever considered adopting

[fivemack]

Comments

[akirlu.livejournal.com]

Man, that image buggers up my FL page like howdy.

[bugshaw.livejournal.com]

They are lovely squiggles.

[fivemack.livejournal.com]

Thanks.

They seem to be just a little too complicated to write by hand without looking at the paper; at least, whenever I try I realise that I've drawn them so as to self-intersect nastily. I suppose the first three Rs I drew without looking at the paper probably also self-intersected, of course, though I'd hope my fine motor control is better at 31 than it was at 3.

[del-c.livejournal.com]

Just for once, Opera behaved nicely for that one, even before it went behind the cut.

I count 412 squiggles. What's the significance of that? I thought at first it was going to be a power of two.

Dyslexic people would utterly hate you if this got adopted. Have you read any works on readable text? It turns out descenders, ascenders etc. are more than just nice patterns, they actually have readability advantages (letters that are near-mirrors of each other are not nice to said dyslexics, however).

One of the new Vista fonts has numerals that ascend and descend, instead of sticking resolutely to the same height. Once you get used to the weirdness, they actually do make it easier to read and memorise large numbers. apparently it's only in recent times that numbers had to be even in height, and this was a fashion statement.

[simont]

Just for once, Opera behaved nicely for that one, even before it went behind the cut.

Safari did okay as well: it managed to put line breaks between glyphs as necessary. Firefox insists on putting them all on one very long line, unfortunately. (Though my friends page is unfazed by that, muhahaha. My LJ style permits people to render their own posts and comments illegibly wide, but protects every post and comment from strange HTML perpetrated in any other.)

I count 412 squiggles. What's the significance of that? I thought at first it was going to be a power of two.

(After quickly hacking up a Python program to check) 412 is exactly the number of distinct squiggles which obey the following rules:

• 5×5 grid of dots
• single unbroken line
• line starts at the top left dot and begins by moving one space right
• line then visits every dot in the grid exactly once
• line travels orthogonally between adjacent dots at every step.

Looks as if fivemack has generated all the possibilities and translated each one into a nice-looking glyph.

[lnr]

Um, *my* (linux) firefox (2) did not insist on putting them all on one line. And experimenting with it and (windows) firefox (3) they quite happily re-wrap them as I resize the window.

[simont]

I should have said Iceweasel, I now realise. Though I have no idea whether that's what makes the difference...

[simont]

Ah, now I see. It's because my LJ style doesn't explicitly open a <p> tag between the innermost <td> and the string of images, whereas other styles do. Adding an explicit <p> turns on the wrapping behaviour even in Firefox 2, whereas Safari was happy to wrap regardless. I've now fixed my style; thanks for giving me the hint that it was something I could fix at my end!

[ceb.livejournal.com]

Can I use them?

[fivemack.livejournal.com]

Yes, of course!

[megamole.livejournal.com]

majQa!

[simont]

If it'd been me, I think, I'd have done 1 setlinecap as well as 1 setlinejoin. (Or equivalent if you didn't actually use PostScript, of course.)

[fivemack.livejournal.com]

I did it in GD::Simple, which defaults to the state it's in; I suppose I could add small circles at the start and end points.

[dd-b.livejournal.com]

So we don't actually *know* if maybe, if I stare at it too hard, Cthulhu might appear? Guess I'd better be careful; I certainly feel *something* weird happening behind my eyes when I look at it.

[simont]

Rather surprisingly, I note that the sequence 1,4,26,412,11072,... does not appear to be in OEIS. I think that's the first time I've ever looked up a sequence in it that turned out not to be there! Perhaps you should send it in. (After first computing another five terms, I expect, by a cunning dodge I haven't thought of and/or the application of more raw computation than I can conveniently bring to bear or both.)

[fivemack.livejournal.com]

I'm wondering if I'm missing some other obvious symmetry which would bring the numbers down a bit and make the series different; I too only got to n=6, but was running in Python for minutes and could probably manage to run in C++ on eight cores for a week if anyone were interested in the answer.

Initially I was counting designs of this sort on a 3x3x3 cube, which by choice of axis always begin either 0125 or 014, but there are fewer symmetries of the square.

412 is quite a nice sort of number - comparable to the syllable-count of English, I think, which could lead to some spectacularly cryptic inscriptions slightly too hard to decipher by hand to be fun. The thing I liked about these squiggles is that any two of them do look reasonably different, in the way that two similar Chinese characters look different; they're encoding eight and a half bits while not looking (to me - I have no frustrated nine-year-old Chinese kid to ask) too complicated for humans to work with.

Someone suggested carving them on stone, which I suspect would work rather nicely with the right CAD package and a round-tipped router bit; but I don't need more reasons to buy a CNC router, I just need somewhere to put it. I assume the monumental mason along the road could tell me where to get an A4 slate slab.

There are 45 squiggles which end in the top left-hand corner and so could be written cursively, but I imagine writing cursively would look incredibly confusing; and these will look sillier punctuated than even Chinese in one of those fonts where 002C 'COMMA' is the same size as 9F4E 'TO PRESENT RESPECTFULLY USING BOTH HANDS' or 齉 'SNUFFLE'.

I suppose shorthand is the way to go to look for an extra nineteen characters when writing English; I've naively merged N-gram frequency tables up to N=4, but the recommended polygrams overlap a lot: THE, AND, ING, HIS are clearly useful, DTH is more common than I'd have thought ... at this point I should probably ask hilarityallen about what sort of characters were used by medieval scribes.

The most common five-letter words (as opposed to five-letter sequences) are OTHER WHICH ABOUT WOULD WRITE THESE THING COULD SOUND WATER; single-character nouns would seem too specialist, maybe four conjunctions and twelve prepositions would be a way to go.

[del-c.livejournal.com]

If 4x4 is 26 patterns and 5x5 is 412, what is 4x5? When a) the line the line begins by moving in the 4 direction, b) the line begins by moving in the 5 direction?

What would it look like if presented as rectangles 4 wide by 5 high, with the lines starting in the lower right corner with a visible descender, except in those cases where the line happens to end at the top pointing up, where there is a visible ascender instead?

How many patterns in a 4x4 grid if the line need not start in a corner?

[fivemack.livejournal.com]

5x4 (5 wide, 4 across), beginning by moving in the 5 direction gets 73 patterns; beginning by moving in the 4 direction gets 87. I left the software to do the drawings on the other computer, will try to get them rendered before I flee the country.

Descenders and ascenders sound like an idea; there's also serif potential but I wonder if that might not just look messy.

4x4 starting in arbitrary places gets me 552, but I suspect a lot of those look identical (since the line has no direction to it), and a lot more are rotations or reflections of one another. Basically you can start at 0,0 (52, but they come in pairs depending whether they go to [1,0] or [0,1] first); or at 0,1 (25); or at 1,1 (36)

[simont]

412 is quite a nice sort of number - comparable to the syllable-count of English, I think

And the number of such glyphs on a 4×4 grid is exactly 26, which suggests a much simpler encoding of English text. Perhaps while the adults are puzzling out your 5×5 syllabic script, the children can be working out a simple substitution cipher based on the 4×4 version :-)

[fivemack.livejournal.com]

I _think_ term 7 is 755223.

[davidcook.livejournal.com]

... possibly because Chinese already exists :-)

I'll take del_c's word for it that there are 412 - which is very close to the number of distinct sounds (not counting tones) in (Mandarin) Chinese. I'm sure there would be some clever way to map all of the non-intersecting paths through a 5x5 grid (I guess that's what they are, at least) to the Chinese syllables ...

[davidcook.livejournal.com]

... of course, if I'd refreshed the page before commenting, I would have seen simont's various comments and done rather less stating-of-the-obvious. Oh well ...