mastodon.gamedev.place is one of the many independent Mastodon servers you can use to participate in the fediverse.
Mastodon server focused on game development and related topics.

Server stats:

5.8K
active users

demofox

my god... sorting the numbers 0-99 in german makes a low discrepancy sequence. wtf.
Thanks for the link @andrewhelmer47
reddit.com/r/dataisbeautiful/c

@andrewhelmer47 The explanation is interesting too.
paraphrased "the least significant digits are said first, like "one and twenty" instead of "twenty one", so it's like Hammersley, but in base 10."

@demofox @andrewhelmer47 It's a completely inhumane way of expressing numbers let me tell you!

@WaitForPresent @demofox

There is a society (Verein) that wants to change that.
They demand! that the German language changes!
And there are at least seven of them. That is the number you neev for a registered society (e.V.)

zwanzigeins.jetzt/startseite

zwanzigeins.jetztZwanzigeins – Startseite

@ospalh

I sort of want to see them battle the fascists from the society for the German language. 😅
@WaitForPresent @demofox

@Fripi

Yeah, stick it to the Verein Deutsche Sprache!

But you have to be careful not to mix them up, especially when translating the name. Those folks are very much not the Gesellschaft für deutsche Sprache.

Or did i miss some developments at the GfdS?

@WaitForPresent @demofox German (and also Dutch) is more consistent than English though. In German, the concept of saying the units before the tens is also applied to the numbers 13-19 in a sense. English does as well, but then switches to the tens-units order starting with 21.

@WaitForPresent @demofox I'm German and almost 40 and I struggle with numbers between 21 and 99. I hate it. German numbers are like dates in month-day-year format.

@demofox in English both ways of saying the numbers are acceptable though eg "four and twenty blackbirds"/"twenty four blackbirds". Just modern English favours the latter.

@demofox @andrewhelmer47 The fun thing is that you swap ones and tens in every group so 123456 is "einhundertdreiundzwanzig tausend, vierhundertsechsundfünfzig", literally "one hundred three and twenty thousand, four hundred six and fifty"

@demofox @andrewhelmer47 i kind of want to see it in larger, too, because for 123 it‘s like „one hundred, three and twenty“

does it end up as like, a bunch of squares?

@halcy @demofox I implemented this in my web UI for Kap, the code is pretty short and you can see it for Swedish here:

Scatter plot for Swedish numbers

Just click on "send" to run it.

It's reasonably easy to adjust to any other language and any other length of words. My example programmatically generates the words to save space, so the code may seem a bit difficult, but if you have the pattern you want to try I can help write it (or, one could just replace the calculation with the list of words)

kapdemo.dhsdevelopments.comKap client

But for the fun of it only in the tenth...

four thousand three hundred one twenty (4321)

@demofox
@andrewhelmer47

@demofox my last shower thought on that is that German numbers have the same flaw as American dates. Month/day/year is neither big endian nor little endian, and so is hundreds/ones/tens.

@demofox Believe me, it's hell when you speak both German and English, and always have to rewire your brain to the corresponding digit order. 😅

Also it's not just for the least significant digits. 21,483: "Einundzwanzigtausendvierhundertdreiundachtzig" (one-and-twenty thousand four hundred three-and-eighty)

@demofox @andrewhelmer47 Now that makes me wonder what would happen if the sorting would be done backwards.
Like english would start off like this:

$ rev top10.txt | sort | rev | tr '\n' ' '
three nine one five ten seven zero two four eight six

(note: I know this isn't how endianess works which is roughly what's happening with German).

@demofox @andrewhelmer47 German has little endien numbers? Just wrong. (As are must current computer CPUs, BTW.)

@demofox @andrewhelmer47

The irritating point is basing on the fact that western cultures are writing in latin way and calculating in the indian way.

In ancient rome the people reads and calculated from the left to the right and used "our" letters and numbers like XIIV.

In ancient india the people reads from the right to the left and used "our" numbers - which they have invented - and a "strange" set of letters.
1/3

@demofox @andrewhelmer47

We are reading from the left to the right and calculating from the right to the left.

Try in the way you learned in primary school:

123
+456
--------
...79

In the graph you can see, that in all languages are sorted like in german in the area of 1 to 15 or 20. At this point the languages flipping from a calculation order to a readable order.
2/3

@demofox @andrewhelmer47

The difference in german is based on german Gründlichkeit. German is not flipping somewhere, it's flipping exactly after the second digit.
3/3

@demofox It's only for two digit numbers, e.g. 4321 is "four thousand three hundred one and twenty". Similar to what English has for the 10s (see e.g. fifteen), but expanded for all two digit numbers.

@demofox Huh... I thought surely it was a joke, but it seems not. A quick search shows that that to say 53, instead of saying "50, 3", you would say "3, 50". I guess that makes sense!

@slembcke @demofox 53 would be „drei und fünfzig“ (correctly written without the spaces so dreiundfünfzig) or translated in english „three and fifty“. In german we have an „und“ between the numbers

@hubot @demofox Ah, ok. I noticed "und". Makes sense that it just means "and". :)

(I only know English and enough Norwegian to be confused by my wife's relatives then confuse them back in return. >_<)

@slembcke @hubot @demofox

Then it will make you really "happy" to know that we can do the same in Norwegian, and many (like me do) so for 53 you can say "femtitre" or "treogfemti" mostly depending on dialect.

@sotolf @hubot @demofox interesting… I just learned a bit (mostly from books and a few movies). My wife studied abroad in Tromsø, and had relatives in Trondheim she talks with sometimes. I’ll have to see if she’s heard that. :)

@hubot @slembcke @demofox Slovenian's very similar – tri_in_petdeset (three_and_fiveten).

@hubot @slembcke @demofox it’s the same in Arabic. 33 would be 3 and 30 with ‘and’ between each number (‘and’ is pronounced ‘wah’). And even though their script is written right to left, their numbers are left to right so your brain does a little wobble when reading a sentence containing numbers!!

@demofox My hash function is now just translating the input to German

@mcc @demofox old and busted: blue noise sampling pattern
new and hot: german numbers sampling pattern

@halcy @mcc it's like wind talkers but for monte carlo

@demofox @andrewhelmer47 oh neat, now I need to see this for Danish (where the numbers are arguably even more messed up than in French 😉)

@demofox @floe They are not messed up at all. The etymologies for the words for 30, 40, 50, 60, 70, 80, 90 are rooted in a base-20 system and so a little strange compared to the other Germanic languages, but these etymologies are uninteresting and actually largely unknown to native speakers. They're just words, as arbitrary signs as any other. And they're combined just like in German, reversed compared to English but base-10.

(Native Dane, MA linguistics here)

@niklasnisbeth @demofox Well, of course, but for me as a non-native speaker, the numbers were (and still are) the most difficult part of the language to wrap my head around.

E.g. whenever I hear someone say a number >= 50, I can basically feel my brain tripping over and completely loosing track of the rest of the sentence. 🤷

I've also always wondered: why is 100 not "fems", if 90 is "halvfems"?

(PS. No need to state you're a linguist, that was entirely evident from your post itself 😉)

@demofox @floe Same reason it's not tenty in English, I think? Although we could keep up that ruse a bit longer, I don't think twentyty would ever catch on.

But generally language is like a strict parent. If you ask it why it just stares back and says "because I say so."

@niklasnisbeth @demofox @floe as a German linguistics drop-out who speaks danish¹:
I enjoyed your conversation SO MUCH! 🤩

¹except for the numbers of course.

@demofox phanpy warned me when I boosted that there wasn't alt text. May I suggest some?
Four graphs titled "The Numbers 0-99 Sorted Alphabetically In..." for English, French, Spanish and German. For English, French*, and Spanish, the graphs have prominent clusters, but for German the plane is covered pretty evenly. A note at the bottom: "*specifically French as spoken in France (Belgian and Swiss French use a decimal system for 60-99)"

@demofox i wouldn't know how to properly describe this kind of graph, so i'm just asking you: could you add an alt text to your post?

@demofox pretty much you organize your number output by saying "one and twenty", "two and twenty", and son on, as opposed to the way numbers are organized in most other languages ( if not all others!)

@demofox Makes sense. About the only German numbers most people my age would have heard are "ein" and "neun und neunzig" (as in: Luftballoons), which hints at the pattern numbers between 21 and 99 would follow...

@demofox @andrewhelmer47 Zeigt das nicht dem Verein Deutsche Sprache. Die finden da nur einen weiteren Beleg für die Überlegenheit der deutschen Sprache.

@demofox @andrewhelmer47
Belgian French is a mix of French and Swiss: 60,70 and 90 as the Swiss do but 80 as the French do

@demofox @andrewhelmer47 TIL there's at least *one* somewhat cool property of the German language's weird "middle-endian" number format!

@demofox
Sorting alphabetically is silly, anyhow. As I already provided in 2020, you have to sort those number names by length!

twitter.com/datenhalde/status/

Time for a scatter plot, presumably...

@andrewhelmer47

@demofox
It's slightly confusing, that the dots are not on a straight line for German. As all numbers with one as a second digit (1, 11, 21, 31,...) start with the same letter, I would have expected them to be on a horizontal line.
Same for all other numbers ending 1 to 9.
@andrewhelmer47

@Peranka its not "grouped by first letter", but the words sorted alphabetically.

@Peranka Does the dictionary write all words starting with "a" in one line, or do they put them in an order?

@HeNeArXn
Ok, I get your point.
But still, if I assume, that the y-axes is equidistant, I would expect the words starting with "acht" being less far spread (lowest row).
But maybe, my assumption is just wrong.

@Peranka @HeNeArXn I really stumbled over that y-axis, too. It really seems they took the 100 words, ordered them, and reserved a complete line for that point.

I'm now interested in that graph with the y-Axis showing the position of a word on the permutation spectrum between 'aaaaaaaaaaa' and 'zzzzzzzzzzzz' (basically taking the words as a Base26 number, filled with a's at the end to normalize them to same length).

@demofox @andrewhelmer47 It probably helps that in German the written names of larger numbers mostly start with the name for the last digit, so if the single digits are spread out, this should replicate.

@demofox @andrewhelmer47 Specifically important to understand the graph is that the vertical axis doesn't show the absolute position in the alphabet but the word's ranking in the sorting order of all words, sorted alphabetically.