really well put together video! these are some very compelling arguments for binary. you're right that I dismissed it for superficial reasons without second thought in my videos; the immense advantages it has for arithmetic and information density shouldn't be overlooked, and there definitely is a case to be made that these may matter more than the things I focused on. in the few contexts where the notion of an "objectively best base" actually makes mathematical sense as a thing to care about, binary is a clear winner. I don't think I'm completely convinced that binary is necessarily the absolute best choice for a human-scale base (the "coincidental" advantages seximal has for working with small primes are just too good) but I am convinced that it can work as a human-scale base to begin with, which I hadn't even properly considered before. it definitely deserves a seat at the infinite table with the other SHCN bases.
@@wilh3lmmusic thats basically what the video said, you can combine 2 or 3 digits of binary to make it shorter like base 8, but with all the advantages of base 2
@@leggyjorington3960 Not exactly. The grouping system in the video is distinct from hex or octal in a few ways. First of all, if you just want to use hex, then the notation suggested is very inefficient. I'm not a big fan of how we usually write hex (0,1,2,3,4,5,6,7,8,9,a,b,c,d,e,f) but it's much more efficient than iiii iiil iili iill... etc. The same is basically true for octal. Second of all, the advantage of his system is that it isn't hex or octal, it's both (and any other 2^x base). You can very easily re-analyse a number as any of those bases to get its benefits. Sevens aren't great in hex, but they are easy in octal, so just group by three bits. Octal sucks at fives, so just group bits by fours and treat it like hex. Wilh3lm is, imo, correct that the benefits of the more flexible system are outweighed by the practical issues with actually writing these numbers. I spent several months using almost exactly the notation suggested in the video and it simply isn't practical.
I frequently see children using this system to express the number 4 to me when a school bus drives past. It's truly amazing to see something adopted with such enthusiasm at a young age and gives me a lot of hope for future generations.
Another fun fact, since microprocessors out number humans, and most of them do arithmatic significantly more often than the average person does, technically binary is the most often used base currently.
Hi, not sure if anyone’s mentioned this yet but your use of pitches following the harmonic series to accentuate numbers you’re talking about is absolutely incredible and I didn’t want it to go unnoticed.
I think I found my favorite genre of videos: the hour long math rabbit hole. Videos such as this one, "HACKENBUSH: a window to a new world of math" by Owen Maitzen, or " The Continuity of Splines" by Freya Holmer. Really loved the use of music and sound in this one.
I love both of the videos you mentioned so much, if you have any other similar ones to share I'd love to know! I can also go through my favourites and find similar ones if you want me to tomorrow ^^
@@codenamelambda The recent 3b1b explanations of light slowing down when travelling through matter are definitely a must-watch, if you haven't seen them already. Oh, and also the "Why can't you multiply vectors?" by Freya Holmer
This point is addressed in the video. Using just three fingers on each hand still lets you go further than seximal, and binary is the only base where you can use any amount of fingers for finger counting
The amount of times I got caught by the “‘well actually’ -You“ moments scared me. Every time I felt like I had a valid argument to make, there was a direct response to it.
I really don't like the way that binary is proposed to be written in this video. The bottom connection thing is actually really nice, but the whole "short ticks for 0, long ticks for 1, and downward short ticks for the radix point" thing seems like it'd be really prone to accidental slip-ups and unnecessary ambiguity, especially without some sort of guide on the paper.
Agreed. A good way to distinguish them could be to write 0 like the lowercase Greek gamma and 1 like the cursive lowercase L. It would be fast, less prone to errors and easy to read.
@@polymloth I just sat down to try and find a system based on this comment, and this was what I came up with lol, connecting adjacent characters like cursive. This is a pleasing analog of chiral topology, something like over- and under-crossings in a knot diagram. You can use word breaks to indicate digit groups, and a slash to indicate fractions as usual (I don't actually think having a positional notation for fractions gets you much idk).
the thing is, if natural languages have no issue making this distinction (take the word יוון in Hebrew), this shouldn't be an issue with numbers. we can also always take inspiration from Hebrew crossword solvers and turn the short tick to a short cross.
Great video. Although the "speaking system" is quite flawed. So i made a new one. Firstly, the biggest flaw is recursiveness. It's new and neat, but when you want to say a number to other person, it's better to convey number's magnitude right away. For example in decimal you would say "world population in 1975 was 4 billion and dot dot dot". However in binary it would be "three four three hex two BYTE two four one hex four three SHORT dot dot dot", and only when you say SHORT the person can sense the magnitude. As a follow up, what if the person is writing the number down? For example when he hears "four int two..." how many zeroes should he put before writing down "two"? If the number is "four int two short" then 14 zeroes. If it's "four int two byte" then 22 zeroes. If it's "four int two byte short" then 6 zeroes. Secondly, phrases are a bit bigger. A small number in decimal (255 - "two hundred fifty five") would be "three four three hex three four three": 22 symbols versus 37 symbols (or 6 syllables versus 7 syllables). We sometimes omit hundreds, so it's minus 2 syllables. Thirdly, "speaking/writing system" interferes with the idea of grouping bits into groups of 2, 3 or 4 bits. System works with groups of 2 and 4, but does not with groups of 3. As it's said in video there is a learning curve, where person first learns arithmetic on group2 then group3 then (maybe) group4. But what if he considered group4 arithmetic too complex and stopped at group3 ? After he's done calculations on group3 he has no choice rather than regroup the whole thing and only then say the number out loud. For the first problem I'd kinda go traditional method (millions, billions, trillions etc.) For the second problem I'd compress numbers up to hexadecimal digits For the third problem I don't know. Either group3 people will have to regroup, or make another speaking system for group3 representations (which is quite bad). The digits Imho it would be worse to use digit name, that we already use, so I gave new names, trying to reflect "binariness". Also these numbers should be fast and easy to pronounce and phonetically distinct from each other , because they will be used a lot in speech. I'm not a conlanger, but I tried my worst wan du ti ro rówan ródu róti ko kówan kódu kóti kro krówan kródu króti hes Here ó is a stressed o. These words represent names for following hexadecimal numbers: 1 2 3 4 5 6 7 8 9 A B C D E F 10 I used "hes" instead of "hex" for sixteen, because I think it's faster to pronounce it this way. For 0 we could use "zero" I think "r" prefix for 4 and "k" prefix for 8 work quite good with binary. With this system we can count up to 255 (in a similar way we count up to 999 in decimal without involving power names, such as thousands, millions, billions). "hes" is used as a connecting word between quartets (similar to "hundred" in decimal). If the right half of number is zero, "hes" is omitted. ........ - zero .....|.| - rowan |||||||| - kroti-hes-kroti ...|.... - hes ..|..... - du-hes |..|.||. - kowan-hes-rodu Now for power names. Let's call collection of 8 bits as bunches (not bytes cuz we will use this keyword). Bunch in decimal would be 3 digits. Every bunch in decimal is followed by a power name (thousand, million, billion...) If we do the same in binary with suggested names (byte, short, int, long, overlong, byteplex), these "power names" (technically power phrases) will be like that: 256 ^ 1 = byte 256 ^ 2 = short 256 ^ 3 = short byte 256 ^ 4 = int 256 ^ 5 = int byte 256 ^ 6 = int short 256 ^ 7 = int short byte 256 ^ 8 = long (We suppose that most significant words come first) Here I would suggest other naming system (which in general will have more syllables). However each power name will be represented with one word, pronouncing these names I think will be easier, since "int" "short", "long" are not pronounced well together. Also every name will end on "-yte", indicating, that it is indeed a power name: 256 ^ 1 = byte -> byte 256 ^ 2 = short -> plyte 256 ^ 4 = int -> fryte 256 ^ 8 = long -> ksyte 256 ^ 16 = overlong -> znyte (I did not come up with an alternative for byteplex) Now instead of "overlong long short byte" we would get "znyte ksyte plyte byte". But we need to combine these words. The rules are: 1) Last word gains prefix "o-" 2) Other words turn into prefix form The prefix forms are: plyte -> pil fryte -> fer ksyte -> kas znyte -> zun So by these rules "znyte ksyte plyte byte" will convert into "zun-kas-pil-o-byte", or "zunkaspilobyte". Here are first 14 power names: byte plyte pilobyte fryte ferobyte feroplyte ferpilobyte ksyte kasobyte kasoplyte kaspilobyte kasofryte kasferobyte kasferoplyte And now one example with all of this: distance to the Sun in nanometers: |... ...|||.. ...|.||. .||..||. ||..||.. .|.|.|.. .|.||... ..|..||. .|.|.||. "ko ksyte wan-hes-kro ferpilobyte wan-hes-rodu feroplyte rodu-hes-rodu ferobyte kro-hes-kro fryte rowan-hes-ro pilobyte rowan-hes-ko plyte du-hes-rodu byte rowan-hes-rodu"
I'm pretty sure if we had developed binary & hexadecimal counting, we would not be translating numbers from base 10, or from metric, especially since considering the metric system was developed in a world where base 10 was well-estsblished. What if we had established a measuring system based on 100,000 [in hex) of the diameter of the earth through the poles? The "hex stick," if I may, would be about 16 cm in length. A little small, but still usable. As much as we like to make up words, I'm sure we'd still have million and billion, or an equivalent in hex language & a little different in magnitude. Mil & bil are kind of arbitrary labels, but if hex had an equivalent, you could again feel the magnitude. Let's just pretend for a bit that hun, thou, mil, & bil applied to the number in the 3rd, 5th, 7th, and 9th digit respectively. Distance to the Sun would be about 37 milhex, if I may. A population of 4 billion would be EE milhex, which is just under 1 bilhex. Magnitude problem solved. 🙂
@@livingpicture Well yeah, we solved magnitude problem in our ways. But thou, mil, bil are names for 3*n hex digits (or nibbles) which is kinda arbitrary. I tried to follow the idea in the video: to establish names only for 2^n digits. The traditional approach (3*n digits) is probably easier to remember, but idk, I didn't learn my system :)
I kinda prefer this (but what do you guys think about this?): 0 = zero 1 = one 10 = 2 = two 100 = 4 = four 1000 = 8 = eight | 1111 = 15 = fifteen (there should be alternative names for 10[dec] to 15[dec]) 1,0000 = 16 = *hex* 10,0000 = 32 = two hex | 1111,0001 = 241 = fifteen hex one 1,0000,0000 = 256 = *byte* 1111,0111,0101 = 3,957 = fifteen byte, seven hex five (fun fact: this is 3 syllables shorter than its decimal name) 1,0000,0000,0000 = 4096 = hexabyte (or hexbyte?) 1111,1110,0100,1001 = 65,097 = fifteen hexbyte, fourteen byte, four hex nine 1,0000,0000,0000,0000 = 65,536 = *"short"* (there's probably a better alternative name for this we can borrow from Computer Science) This gets absurd... 10,0000,0000,0000,0000 = "two short" 1,0000,0000,0000,0000,0000 = "hex short" 1,0000,0000,0000,0000,0000,0000 = "byte short" 1111,0000,0000,0000,0000,0000,0000,0000 = "fifteen hexbyte short" Maybe we should think of an alternative for "hexbyte"?
James Grime convinced me of 12, almost 12 years ago Misali of 6, almost 6 years ago You just convinced me of 2 Who said there was no progress in history?
in 2 years some will convince you of base 1 and then 1 year later someone else will tell you base 0 is better, on which there will be a imediate response claiming base -1 is optimal, while then you realized that you actually had memories of someone claiming 1 year ago that base -2 was better and so on
honestly, this video has now converted me into a true binary supporter also i did not expect all of the arithmetic stuff with binary to be SO simple and easy to do
The bit about the square root algorithm made me literally get up out of my chair, scream "what??" at my phone multiple times, and roam around my apartment for several minutes rethinking life
In my humble opinion this video goes into the youtube's mathematical hall of fame. A deep and new point of view shedding light on a topic that everyone can relate to yet few thought consciously before. I have no other words than to thank you for your work.
Programmers and hardware engineers have been doing this forever, using 8 or 16 as the "compressed" written format, but dropping back to bits for actual arithmetic, especially when making machines to do so. Anyone in those groups worth their salt can convert from 0-F their binary quartets and back intuitively, and when learning to do that, often use fingers in ways that resemble your grouped bits. The main problem with binary as a written or spoken system is that its hard to read at a distance, and the octal/hexidecimal are "error correcting" in that a smudge on a piece of paper or a dent in a sign makes simple tally marks unreadable, but leaves letters and hindu-arabic numerals mostly in tact.
I'm incredibly surprised this video didn't mention that the GREATEST advantage binary has, is that it's a system we could ACTUALLY switch to without nearly as much hassle as any other system. I'll be honest, I went into this video thinking "huh, interesting. Id like to see a new point of view", got to the twist reveal that it's about binary and went "ok, this is either a joke or I'm in for an interesting if unconvincing response", but now you've really convinced me. Holy cow, I had no idea what I was in for. The counting is fun too since I don't have to remember so much, and somehow these numbers are easier to understand with my dislexia too. They could also be made easier to understand for dyslexic people with a few simple tweeks so that's comforting. And on top of all that, binary numerals would be so fun to make fonts for, as you can pretty much make the symbols whatever you want as long as one is "less" in some way than the other. Hollow/full circle, down/up arrow, Mario/Luigi, literally infinate options lol. Anyways awesome video, please make more! I'd absolutely love more specific video lessons on how to use binary with your numeral and naming systems!!
I think language is the part where I'm least impressed here. There are studies suggesting that languages that express numbers in fewer syllables lead to people doing faster mental arithmetic. A lot of languages have specific words not only for the singular digits 1-10, but also for some numbers into the teens. English has "Twelve", for example, instead of "Twoteen". French has a single digit words for eleven, twelve, thirteen, fifteen, and sixteen. And I think there's also some linguistic benefit to having words for numbers like twelve rather than "ten two" than just a historical linguistic artifact--small integers are just the numbers we will use the most, being able to express them quickly and unambiguously is inherently valuable, even if it makes the language harder to learn than Toki Pona. It's also worth noting that languages tend to have specific words for 20, 30, 40, 50, etc. In English: twenty, not "two ten"--now there's no syllable advantage here (although in some languages the word for 20 is one syllable) but there may still be a linguistic advantage to having a separate word--even if hearing is an issue, you'll never mistake "twenty" for "ten" or "two". Rather than express number words in base 4, I think it would probably be advisable to have language work in...probably hexadecimal honestly? Either that or Octal, but I would lean hexidecimal for two reasons. First because it's 2^4, and 4 is a power of 2, so it'd be easier to deconstruct if you were already thinking in binary. Second having number names for numbers higher than 10 is fairly reasonable, cause a lot of languages already have that (1-12 in English and German, 1-16 in French and Spanish). But honestly, this is all hypothetical, really. People are never going to switch number bases unless governments force it by changing money to a different number base. As long as people have 10s, 20s, and 50s in their wallet, they're going to think in decimal.
I always felt, instinctively, that binary "should" be the best base, but it just seemed too cumbersome to use in practice. Thanks for your excellent work.
You present your arguments well, however: I am nonbinary. So I'll have to stick with Seximal, or maybe Balanced Ternary for the small number and negative advantages
Perhaps you would like base negative two, or negabinary? It uses alternating negative and positive powers of two so negatives can be written just like positives without the need of a minus sign.
Sadly, any way of speaking base-2 numbers will be non-portable. Some ancient cultures will say that int is 2^16, while modern ones say 2^32. And then there's the oft-forgotten cultures where byte (or char, as they'd say) is e.g. 2^31. Instead the names for the double powers of two should follow a simple and memorable system, e.g. 2^16 could be "int underscore least sixteen underscore tee". Since that's a mouthful you should also introduce "intmax underscore tee" as shorthand for "the biggest power of two I feel like thinking of right now".
just use the x86 names: byte, word, dword, qword / quad, and dqword / double quad. i can't remember what the next two were but i could've sworn it was ddqword and qdqword. super convenient and not confusing at all
@@rubixtheslime after qword is either oword (octa) or xmmword (128 bit simd register) after oword is xword (hexadeca) or ymmword (256 bit simd register)
I feel like I've stumbled on a piece of forbidden knowledge. The funniest thing was trying to come up with some counter-arguments while noticing I had already used a binary representation of octal (tri-octal to be more precise) to convert an alphabet, and already figured on my own how easy it was to apply a vigenaire cypher mentally on it thanks to how trivial it is to do arithmetics on it
re: chapter six. i'm not sure your naming scheme really does a great job here. it definitely feels like "ninety one" is less information to parse than "four one hex, two four, three", and in general, standard names tend to feel less cumbersome. perhaps there's a less mathematically elegant, but more linguistically practical, way of handling binary number naming.
"four one hex, two four, three" is more analogous to "nine ten one" like some languages do. For me it's more natural with -s on the things that should be plural, so four one hexes two fours three. (like "nine tens one") I think something that could help solve the cumbersomeness is basically doing base 16. So make all numbers less than hex compound words, so you can treat each as a single unit. Like ninety one would be four-one hexes two-fours-three. Compared to base 10 English, the digits are more syllables, but that happens in other languages. Like in greek the numbers up to 10 are all 2 or 3 syllables. With this system the numbers up to 16 are all 1, 2 or 3 syllables. The fact that it's base 16 means it's actually less cumbersome than base 10 for large numbers. It also keeps all the elegance, since it's just a conceptual reframing to help interpretability. There's still a slight problem that, for example, "four-one hexes two-fours-three bytes hex one" could be read as "four-one hexes" + "two-fours-three bytes hex one", you'd need to say "(four-one hexes two-fours-three) bytes hex one" somehow. I think this is just a problem that comes with the b^(2^n) system, you might have to switch to b^(kn), maybe 2^(4n) so powers of hex. Tho you could have the power names be like hex, byte, hex-byte, short, hex-short, byte-short, hex-byte-short, int and so on, so it'd be "four-one hex-bytes, two-fours-three bytes, hex, one".
@@d.l.7416 in written form, using the 2^(2^n) system, your latter example isn't technically ambiguous. but when spoken it's unclear what order of magnitude is being referred to at any point until the number ends, because at any point while telling you the number, someone could just say "short" and suddenly the number you were thinking of is 16 orders of (base 2) magnitude too small, etc.
my proposal for naming numbers is to simply read off the bits, using short syllables that can be flexibly strung together and said quickly. If 1 is pronounced like "wun" or "nun", and 0 like "oh", "wo", or "no", then a sequence like 1101 1001 becomes "wununowun wunowowun"; and you could optionally insert the magnitude words like hex between gaps. This mirrors the written form and retains its advantage of being able to group digits into whatever sized chunks are most convinient.
Hearing "nine ten one" in English is a little hard to parse, but in my native Hokkien where numbers work exactly like this, it's perfectly natural (九十一 káu-tsa̍p it). I think this just goes to show that it's all a matter of getting used to it. The "four one hex two four three" system is perfectly fine.
@@jfb- I think that would become very confusing very quickly. People already mishear fifteen (wunununun) and fifty (wununo nowuno) in decimal numbers, and you're expecting them to parse wununowun from wunonunun in fast speech (not to mention that both of those numbers are a mouthful). At the very least I think you should pick sounds with greater contrast. Maybe something like ko and mi, where the sounds of each differ in as many features as possible. You'd probably still have people getting mimikomi and mikomimi mixed up if you rattled off a couple of mikokoko-bit numbers, but it would be an improvement.
16:00 I thought I'd heard that somewhere; it's a "trick" used to store floating point numbers in the IEEE standard; the leading 1 in the mantissa is implied
In the end, it's not about "which system is generally better", but "which system is better after I came to a conclusion which were my criteria and priorities".
I think the proposed method on how to say the binary numbers has a couple of major drawbacks: 1) The recursive, non-linear conversion between words and symbols makes it hard to dictate, and non-trivial to write a dictated number down. 2) The symbolically easy doubling becomes unintuitive, e.g. 3 4 1 doubled becomes H 2 4 2 3) It can bury the most significant part towards the end, for example 3 4 2 H 3 4 2 B. You have to listen to all of the spoken numbers to make sure you're even in the right order of magnitude.
i think 1 and 3 come from the fact that it's a 2^2^n system instead of a 2^kn system. you could instead make it a 2^4n system (powers of hex), but name the powers using the 2^2^n system. So hex, byte, hex-byte, short, hex-short and so on. like you'd say for example, three-fours-one hex-bytes three bytes hex two-fours, which is basically just how standard english base 10 words.
Take a cue from the world of programming, where our spoken numbers are base 16 but our math is base 2. If we used the numerals from this video it would make the conversion trivial.
43:29 "Traditionally, to notate a recurring fraction, the entire recurring segment is marked, but that doesn't make a lot of sense: It's simpler to just mark where it starts." I think your opinion comes from the age of typing and digital text representations. In hand-writing, marking the recurring segment makes more sense for the following reasons: (1) Often, you will get this repeating decimal from just long division until you enter a cycle, and adding a line of the repeating part is something you can do after you've written the number, which doesn't require you to erase anything or rewrite the number, and also separates the repeating segment from any digits of the next repetition you may have written before realizing it was repeating. This doesn't apply if you're rewriting the number somewhere else, but if you're copying a hand-calculated quotient with a long repeating decimal, you'll want to use some mark like the over-bar to mark the repeating segment of the decimal anyway, so any other notation like the "r" you're using will just be another notation to learn IN ADDITION to something like the over-bar. Thus, I think this is the most important reason. (2) In hand writing, writing a line over several numbers is not significantly harder or more time consuming than writing an r before them. It's actually a slightly simpler shape than an "r", though the length and precision required mean I'll just say it's about as hard overall. This is different from typing or digital text, where putting bars over numbers requires special characters or text formatting, both things that are much more of a pain to deal with than just typing "r", both just in typing it, and in the sneaky problem that things like this sometimes won't render properly on other people's computers or on printers. (With typewriters, it usually requires some kind of tricks involving typing over the same text I suspect.) That being said, one downside of the over-bars specific to hand-writing is that, due to the imprecision of handwriting, it can often be unclear to readers exactly which numbers the over-bar is over. In typing, this would only really be a problem if your method of creating over-bars was bad, and using preceding "r" would get rid of this issue in handwriting (as would using parentheses or circling, although those aren't quite as convenient to add on to an already written sequence of digits as an over-bar is).
In my school they taught us to write repeating digits usung brackets: 1/7 = 0.(142857) This eliminates the precision problem as brackets are easier to write and mark exactly where the repetition begins and ends
Finaly managed to crack the text at 8:20 on my own. The substitution is 啊 = A, 痹 = B, 雌 = C, 低 = D, 婀 = E, 付 = F, 佮 = G, 喝 = H, 乙 = I, 咳 = K, 刕 = L, 冪 = M, 妳 = N, 我 = O ,仳 = P, 儿 = R, 絲 = S, 偍 = T, 無 = U, 予 = V, 劸 = W, 牙 = Y. And the actual text is that one comment from jan Misali's Ido video: " YOU HAVE GOT TO BE ABOUT THE MOST SUPERFICIAL COMMENTATOR ON CON-LANGUES SINCE THE IDIOTIC B. GILSON. DID I MISS THE ONE WHERE YOU SAID WHICH CONLANG YOURE FLUENT IN AND READ AT LEAST THREE TIMES A WEEK AND CAN READ NEW BOOKS IN EVERY WEEK OF EVEN ONE YEAR OR LISTEN TO RADIO SHOWS IN EVERY WEEK? NEW RADIO SHOWS? "
I went into this with a strong preference for hex. I got pleasantly surprised by the suggestion of binary, and when you started grouping the bits it's essentially a multi base system around binary. Hex is really just groups of 4 bits, which is why it's good. I think the naming would sound more natural if it's done for groups of bits instead of what's suggested here
I love the notation system you created and the grouping shorthand, it’s very elegant and also makes the fundamental patterns you discuss visually obvious at a glance, without having to translate into numbers, even for someone totally new to the notation without having built base specific intuition. I could easily learn to do a lot of that math with your binary notation without converting into or out of decimal, which makes an excellent argument for it’s naturalness and simplicity
Don't you love it when you have that one really strong opinion that no one else cares about, but then you stumble across an hour long video essay about it at 3 am
i would love to see some examples of this binary notation used in some common settings. some examples >in a car for speed and distance. >prices and measurements in a grocery store. >numbers in a video game. >a deck of playing cards. that batch of 4 trick seems like it would be very readable. i would also like to see something like "babies first numbers video", that would really show how easy binary actually is. a webpage for "try math in a new base" could be a nifty demonstrator for it's usability. i think the symbols themselves need a direction notation, the underline may be enough but I'd like it more obvious.
I made a userscript that you can find on greasy fork called "convert to binary" that attempts to convert numbers on webpages to their binary representation. The only pitfall of it is that it doesn't accurately convert non-integer numbers (like it'll turn 10.32 into lılı.lııııı)
@@ARockRaider It doesn't look any better. The issue here is both conversion and size. It's pretty silly to suggest that using rather indistinguishable characters is somehow better here. There is *some* ability to parse a misread sign simply becuase of possible interpretation, but that's because we're operating within a presumed 'field' of possible speeds. It's also more important that I used a number like 22, which I can defend, and you instead focused on what the digits look like. Which tells me you either didn't take my critique seriously or you are too predisposed to the idea of it working without considering it's real-world applications.
@@jibbjabb43 I had assumed that you picked a number to look the most outlandish, i was pointing out that you wouldn't use actual 1s and 0s and that you would be using a notation that makes the numbers every bit as clear. that you picked 22 makes my assumption of an intentionally outlandish number very clear, on top of that the exact speed would be adapted to the number system 20 for example would be i .i.. and as a group with proper underlines would be very different from 40 or i. i... neither look anything like 30 or i iii. but i expect that you wouldn't be using multiples of 5 for speed-limits, rather you would pick a new batch that always stays neat and round.
Amazing. Genuinely, i was amazed multiple different times. I'm a computer science major myself, I had already discovered the bit about multiplication, but the section on factoring? I had never thought to do that, and it blew my mind. Division? So much better than base 10. Your notation is beautiful, I'm definitely adopting it for when I do binary work, and I'm geniunely strongly considering switching to binary in my everyday life.
14:49 I can't speak for jan Misali, but I think he probably just meant that if you write binary with Hindu-Arabic numerals, the numbers get unwieldy in size, and base 4 in Hindu-Arabic numerals is a good way to compress it. 15:36 This fact isn't what makes binary the most efficient, it just makes it the most efficient by a long shot. Even if we didn't consider the fact that the leading digit can't be 0, by replacing b-1 with b in the formula, binary would still be more efficient than base 3. side note: I don't know how we're measuring which is better rigorously, even though it's clear intuitively. Because the graph for base 3 does sometimes dip below the graph for base 2, like for example between 8 and 9. Is it the one which is the lowest for the most numbers, which has the lowest average, or what? 16:27 I wouldn't be so hard on seximal. jan Misali admits that in seximal, numbers are written longer, and the justification isn't radix economy, it's that the square base is small enough to be used for compression, it's only too large to do basic arithmetic. In fact, I think it doesn't make all that much sense to use radix economy as an argument at all, our brains don't simply look at the representation of a number and take a specific amount of time to process each digit based on the log of what they expect the digit to be. Although it would be fun to see a study comparing each base in its own writing system and the speed of writing or reading numbers.
holy shit, *holy shit*, i think this might be my favourite video. so so so many cool points brought up. so many new ways of thinking that were a delight to be introduced to. it has like, all my favourite things. this video is a masterpiece
I really loved the video, but I think, as many commenters have pointed out, that the legibiliy of your line system could be improved. Some might even compare it to a certain Minecraft optimised number notation system. My suggestion would be to change out the short line for a small circle. This not only de-clutters the lines by separating them more cleary, but also prestents a great opportunity for ligatrues utilising the existing latin alphabet and also reducucing stroke number. or for l. and and for .l maybe for l.l, and maybe for .l. Keep in mind the arcs would be proportionally smaller in written notation I'll flesh this out further and come back to this comment later, I'm expecting some grouping and priority conflicts, but there might be a nice way to deal with them.
"just use only three fingers on each hand" leads to communication issues. The great thing about addition only decimal finger counting is no matter how you count you always get the same result. It is unambiguous and inclusive to everyone who has any fingers. But yes it also means that you can only communicate very few numbers at one time. Basically decimal counting is TCP (sacrificing speed and efficiency for making sure it is as unambiguous as possible within the system) while binary counting is UDP (so prioritizing speed and data density while risking the wrong message being sent)
18:14 What happens, is that you'd stop being able to count it easily by sight at a distance - counting "how many fingers are extended" doesn't care about positioning, binary counting does. And the issue with this, is that when using hand signals to convey numbers, you're probably doing that _because your voice can't reach the other_. Most situations where that happens are scenarios in which stopping for a couple seconds to reorient & count the exact numbers for yourself is an awkward use of time. As an example: if I am on a construction zone with operating machines, and I need the handyman to replanish the third team with beams, screaming "BRING BEAMS TO THREE!" will probably be only partially heard, and after hearing the expected "WHAT?", being able to just raise three fingers while hollering "BEAMS TO THREE!!!" adds a layer of important redundancy, as the handyman will be instantly able to note that one word is 3, and the other is the object to bring, allowing them to process it better across the other sounds. And yes, this is also true with binary, but with that system you'd be forced to use and read for the specific fingers which hold specific meanings, which can easily mean that the handyman needs to ask clarification _again_ because either they didn't catch if the fingers raised were for 2 and 4, or 1 and 2, or they caught that, but not what to bring. It's already hard enough with all the redundancy we use, and needing to clarify and reclarify everything is stressful and awkward for everyone involved. In this sense, using binary for finger counting is strictly worse. Seximal, or well, the other counting methods named came about because of that need to balance redundancy and information density. Sometimes, being able to alternate between asking for 2 of something and 60 of something is important, and between the thumb being quite ubiquitous and only needing to pay attention to "which hand are they raising", reading fingers at a distance using only 4 parts, is plausible. Using 10 parts, making order important for every finger, is simply too awkward to use. However, I will concede some other thing that's reasonable: if instead of an specific number you are asking for a general amount, then having each finger imply a different order of magnitude is useful, albeit there's already an order-independent system for that in which the amount of fingers raised is how many 0 are after a 1. So yeah, binary is really damned good for counting... In paper and computers. It isn't good for hand counting tho, but honestly it doesn't need to be - hand symbols can simply not change alongside written ones.
Some notes as I go: - I see you don't want to be able to write the numbers 0 and 1, if the leading digit doesn't matter for information. - You seem to be neglecting the human tendency to ossify details. A binary system wouldn't stay a binary system for long. Within a few generations, eople would start writing shorthand octal or hex with a couple strokes rathet than four or five, and those would become the basic unit. That grouping trick is actually a negative when you look at language evolution. - "What's the Most Commonly Used Prime Which is Incompatible with This Particular Base?" got a laugh out of me, good work on that one! - Your repeating symbol and your grouping marks would be super easily confused in hasty handwriting. That's not an inherent binary problem though, just one with your notation. - 2⁴ should be called nybble or nyb. It's cuter and sounds better in practice. - ... I both love and hate you for pointing out to me that stack is a number name. It is though, saying something like "three stacks twelve" is perfectly normal in minecraft contexts. - I think seximal is still more aesthetically pleasing, to be honest. And that matters because the only context switching away from decimal will ever happen is art. ... though balanced ternary might win there, it's just the right amount of weird to be really fun.
For your first point: I do agree that they handled this poorly. But I think all is not lost. Instead, you can think of what they are talking about as a consequence of the fact that simply knowing the bit length of a number tells you the value of its most significant bit (which has to be 1). The “length” of the number does not give you so much information in other bases, this is definitely true. As for your second point, so long as people don’t forget where these ossified forms come from and they are visually similar to unossified forms and can be easily decomposed into them, then I don’t think this is actually an issue.
Since the measure of information per digit is considering all infinitely many numbers and the number of single digit numbers is finite, the amount of extra information carried by that first digit tends towards zero in the limit of considering all numbers since a finite portion of an infinite space is always 0% of the total space. I think this means that when considering information on the long scale this simplification is fine, since the context of the discussion was already the long term efficiency of a base and not its efficiency writing small numbers.
This feels like an induction into a religion. Now I see how things should be, and my duty is to convert the non-biners to the true light of one and zero.
The thing that fascinates me the most is that base 2 is every base 2ⁿ at the same time, so it is capable of using properties from all of them. Amazing video.
Great video! I wasn't even half way through and you had me convinced. After watching the procedure for the square root, I managed to find a way to get any integer root! It works in every base, but its way simpler in binary. Basically instead of separating the original number in pairs, you separate it in base-of-the-root groups, and then instead of checking if its larger than or equal to the number above with a 01, you check if its larger than or equal to the following sum: sum from k=1 to q of (q choose k)*(2p)^(q-k) with q being the base of the root and p being the number above.
I think I might be the first person to discover this method, the method for square roots obviously already existed but I couldn't find anything about cubic or any other roots using this digit-by-digit method.
Grouping "bits" by up to for 4 is also pretty practical in a human sense, since around 4 is the natural range of human "subitizing" i.e telling how many of a thing there are at just a glance
Actually, I have a more general counterargument: Redundancy is good, actually. All of these arguments demonstrate that binary is the most efficient base in a similar way to how Ithkuil is the most efficient language. Sure, that might be true on a technical level, but if you miss more or less any information, it's harder to recover. For example, something weird happens with the HDMI on my TV, so the edges of the screen get cut off when I try playing a Switch game on it. But because all of the Hindu-Arabic numerals are fairly distinct, I was still able to follow the timer in the Star raids in pokémon, despite the top half of the number being cut off. Meanwhile, if the top half of your binary numbers got cut off, that'd be it. There would be no way to distinguish numbers anymore. Or similarly, consider seven-segment displays. They're actually *horribly* inefficient, since you can display 128 distinct figures with them, and we only use 10. But because we use so few, depending on which segment goes out, you can still distinguish most numbers. And even in a case like the middle segment going out and making 8 and 0 indistinguishable, context clues help. If you see 1-"thing that looks like it's supposed to be a 9" change to 10, you can infer that it's probably counting down and is probably supposed to be 18. But if it changes to "thing that's probably a 2"-0, you can infer that it's counting up and that it *is* supposed to be a 0. Or similarly, finger counting. Functionally, "binary" finger counting is actually duotrigesimal. No one's going to actually think of each finger as its own digit, so you're effectively expecting people to learn 32 different hand shapes. So if you're just using hand counting to show someone a number, like a kid going "I'm this many", it's going to be less efficient compared to just holding up a number of fingers and subitizing. Or similarly, if you're actually counting up or down, with both seximal and jisanbeop, you only ever need to change either 1 finger at a time or all 5 fingers. (And with all 5 fingers, it's also always either resetting the hand or changing 4 fingers to a thumb) Meanwhile, with binary finger counting, you might have to change any number or combination of fingers. And that's actually so well known of a problem that Frank Gray came up with the idea of reordering binary numbers so you only ever have to change 1 bit at once all the way back in 1947. (With the idea itself going back to at least the 1870s)
I actually agree with most of your points, but have a few nuanced points to make. 1) This one's simple, actually using binary doesn't mean blocking out half the digit would leave it unreadable. Binary is actually pretty good about that, you can *always* tell if you can see part of the digit. 2) I use decimal for readability purposes, but I find binary a better system for doing simple counting tasks. Remember that you could use different systems for different purposes, it's done in some cultures in other areas, such as speaking one language and writing another. (Looking at you, Switzerland...) 3) While your argument about finger counting has some merit, I actually think that's only a problem for children. Since we're actually using a *unary* system for that task, binary finger counting is strictly better but also more complex. (Seeing each finger as a digit with two states, the *number of digits* doesn't have anything to do with the counting system - that's just a physical limitation.) It's too complicated to teach to children without a whole course around counting systems, which doesn't make sense at that stage of life. For adults in the mathematics or comp sci sphere though, it's easy enough to adopt and therefore we probably should do so on an individual basis. Not sure what the video itself seeks to argue, as I'm not about to watch someone try to convince me of the merits of a counting system I already use, so don't take any of this as an argument toward the creator's points. I'm simply arguing for what I believe to be best in my own experience.
@@impishlyit9780 1. Eh, not necessarily. One of the video's weaker arguments has to do with number length. It basically makes a *font* based argument, where if you change binary numbers to tall (1) and short (0) lines, the horizontal width of numerals becomes comparable to other bases. And that really *does* run into that issue. Or more broadly, I forgot to include this bit for contrast, but imagine a binary display. If any one segment goes out, you cease to be able to distinguish anything in that place. At a minimum, you'd need to do something like pairs of dots, where only one can be on, for you to be able to lose a segment and still be able to distinguish numbers. Meanwhile, even though only the top left and bottom right segments lack minimal pairs, you can still generally make things out on a 7-segment display, even if one of the segments goes out. You'd need 2-3 to go out before it starts really impacting your ability to read things. 2. No counterargument here. 3. It's also more than just kids. For example, how often do you *actually* need to count on your hands? It's typically smaller things, like how you might silently count down from 5. And while I'll grant that people probably count down specifically from 5, in part because we have 5 fingers on each hand, unary feels a lot easier for that. In a way, it's sort of like sorting algorithms. Merge sort is more effective for really big lists, but it also requires a lot of overhead. So for shorter lists, the less asymptotically efficient insertion sort winds up running faster overall. The main time I can think of that it's potentially useful to display bigger numbers would be something like days, but there are also systems like the Medieval Arabic hand numbering system that do it more easily than finger binary. It's actually even more efficient than finger binary. It encodes one digit on your thumb and index finger and one digit on your other three fingers, so it can go up to 9999 on two hands, as opposed to 1024. But overall, my main criticism of the video really is that it successfully argues that binary is the least redundant numbering system possible, but not that redundancy is something we need to be avoiding
You should into your TV's settings and turn on either "overscan" or "just scan", or set the aspect ratio to "full" or "just scan". Turning on "game mode" might also fix this.
I really appreciated jan misali's introduction to the advantages of seximal, and have considered myself a convert for many years. You present a strong case here, and it's certainly convenient that your choice happens to be binary. One of the two systems that I have to know anyway. I'd like to see separate (short) videos on some of the topics here. How to write binary numbers. How to speak binary numbers. How to divide binary numbers. Etc.
Personally as programer I chose binary with hexadecimal compression because it is the best of both worlds. Your number to long, just start compressing into hexadecimal, need to do math decompress it. It just allows better writing efficacy and in low level programming you just do base two operations on hex nums anyways.
I am also a programmer, I love the idea of having two different ways of writing numbers, I previously only thought about hexadecimal and even made my own writing system for it wich also uses a sub-base of 4, 2*2 = 4, and 4*4=16, I love the symmetry Binary for math operations and hexadecimal for showing numbers!
The music argument in the early chapter is interesting. Our notation is specified in fractions of powers of two - but they are fractions of four beats per measure. We also work just as often in three (triplets). However, we count in groups of 4 or 6, and often 8 (dance) and 12 (triplets over 4 beat measure). When dealing with irregular numbers like 5 or 7, the rhythm is usually done in syncopated emphasis of groups of three and two. For example, 5 is usuall 3+2 or doubled at 3+3+2+2; 7 is usually 3+2+2. Syncopated division of 8 is common with 3+3+2 (typical of latin and jazz) and 12 with 3+3+2+2+2 in flamenco compas. You further get the complication of one instrument playing a phrase in triplet over the rest of the band playing in duple. The final pattern is that phrases tend to also be groups of four measures, a prime example being the three sections in a twelve bar blues. This is super internalized, and my brain stops and pays attention if a phrase isnt that multiple.
4-bar phrases and 16-bar sections are common enough that when I'm counting measures of rest or something, it usually feels most natural to count in a base 4 system. On each hand I use my 4 non-thumb fingers and with 2 hands I can count up to 16. For example: 1 = ...| 2 = ..|| 3 = .||| 4 = |||| (1-4 had implicit |... right) 5 = ...| left, ||.. right 6 = ..|| left, ||.. right 7 = .||| left, ||.. right 8 = |||| left, ||.. right 9 = ...| left, |||. right etc. On both hands I start from my index finger, but with each one visualized palm-facing down, that results in the left hand counting right-to-left and the right hand counting left-to-right, but the point is that it feels right in the music.
My first reaction was "wait, binary??", especially as i watched the jan Misali's videos before, but then it turned into "oooh, thats how we can do it", and then "it's beautiful as heck"
This really is exceptional content. As a nerd I really appreciate the work that was put into this video. Great work. My only note is the pacing was too fast for me to grasp some of the details that were claimed to be “immediately obvious”.
@@matheuscabral9618 First, there's no need to make assumptions and attack people. Please refrain from it and maintain a civil discussion. Secondly, my point is that they said their video would've been longer, which is unexpected to me seeing how it responds to a 10 minute video; this video is already so much longer than the video it responds to!
My favorite numerical base is centovigesimal, or base-120, because it is really good for divisibility tests and representation of fractions... not only because 120 is very composite, but its neighbors are also composite (119 = 7 • 17 and 121 = 11²).
Having been in computer science, I and all the computers in the world agree with you, because it can also be used to represent pretty much any concept that can be stored or represented digitally ... until it comes time for a human to interpret what's being said, because it's horribly inefficient! This is why we compromised and use 4-bit and 8-bit, which transfer nicely to hexadecimal, and lead to more efficient ways to represent extremely large numbers, which has become necessary in the age of the gigabyte, terabyte, petabyte, and beyond.
Honestly the only reason why I never considered binary in this nerd debate of "best counting system" was the number length. With tally-mark-like digits this already makes so much more sense, and that's just chapter . in the video. we even have ascii characters for binary numbers: |.|. looks basically like the marks. regardless of what you think remember, every base is base 10 :3
What a high quality video! I wonder if I’m really slow or most could keep up without extensive pausing and replaying. Congratulations on such a well made video regardless.
1:08 "Perhaps that too is the case because 2 is the smallest integer greater than 1." Yes, I think so. The note an octave above is the first frequency in the harmonic series of a pitch other than the pitch itself, and so an octave can be thought of as the smallest maximum consonance interval. It is interesting to consider systems that use other integer ratios in similar ways to how we use octaves though.
Non-binary music scales are neat to explore but they never seem to really click for me. Any system with a base larger than 2 has to avoid including a 2-octave between any of its notes, otherwise it kinda overpowers the 'real' base of the scale.
I have to do some maths and check how unary does in all metrics mentioned in the video. Surely, it won't score much on fractions, but it might get some absurdly good results wherever logarithms appear.
I think many of your ideas have basis. But i dont think that it works quite as well in practice. I am a fan of base 6, because of jan Misali's video, but as a programmer and computer scientist, i have to acknowledge the advantage of binary for raw computation strength. There is a good reason we use it in computers, and computation is a place where numbers shine. But unfortunately for us humans, we are not easily wired for base 2. It's hard to say how the real world efficiency of human calculations would turn out if we completely switched to binary, but i think that our brains are slightly better wired to understand larger bases. As someone who has poked around with basic arithmetic in both binary and seximal, i can say that binary is easier algorithmically, but if i had to do a lot of math, i would prefer base 6. Part of this is the practical problem of we need to write down the numbers we are doing math with and i feel like bigger numbers are better for cramps. I think it would be interesting to do legitimate experiments teaching people variou bases, and see who in the long run uses the bases most effectively. Seximal, for me, just seems to be a nice balance between small bases, which are easier for math, and big bases, which are easier for brains. Curious to see if jan Misali responds, but i still think i am on the seximal side for the best base for humans.
I suspect that, were this to be actually used, it might be easier to use hex (or maybe octal) and convert to binary (which is really easy) every time you need to calculate something. Hex is easier to remember numbers in (try remembering 00101101 vs just 2D), to write down (6 strokes and 5 pencil moves for |.||.| vs 4 strokes and 1 move for 2D) and to input on a keyboard (two keys pressed for 2D vs 6 for the binary).
As someone who played with numerals a lot as a teen, and who's done a lot of programming, I find it VERY telling that we programmers who work with it often will ALWAYS display it as hexadecimal. The 'best base' is really about what is the best base for humans, and minimizing how it pushes against our mental limits. I really liked your new binary-based symbols for hexadecimal numbers. I wonder if a society based on binary math and those symbols would work better and learn math easier. I think the binary naming in this video got really awkward and long to speak. Instead, use the groupings and name those. I think a society would need to use octal or hexadecimal for communication, while doing arithmetic in binary.
As a programmer myself, during my education (specifically when learning assembly), I always found working with the binary numbers far more intuitive / less cumbersome than working with the hex numbers. Like, at all points, I would think about it in the binary expansion, but then every time I had to actually write the number in the program I would have to group the binary by 4s and convert to hex, its just a mess. This is not to say hex is bad per se, but that it is necessarily just a layer of translation/abstraction over the actual numbers, which are binary. "x5555" is obscure, needs translation. "101 101 101 101" is the real number, the thing you actually need to work with. Anyways..
Well let’s see. The majority of programming is done staring at a text rendered with a monospace font. Why of course it would be quite a nightmate to juggle literals 32+ characters wide. There’s no real compressibility unless the programming language allows macros or something to use a custom number literal format (or the IDE used allows to work with proportional fonts in comfortably). I suspect there can be more mundane reasons other than this one.
@@05degrees Yeah, suppose we start with a system like the proposed. I think it wouldn't take much for people to start comming up with symbols and meanings that more readly represents the common quantities and groupings.
Felt compelled to decipher the text at 8:20. It reads: "You have got to be about the most superficial commentator on con-langues since the idiotic B. Gilson. Did I miss the one where you said which conlang you're fluent in and read at least three times a week and can read new books in every week of even one year or listen to radio shows in every week? New radio shows?" Was a fun challenge, thanks!
Very interesting. These are some convincing points for doing arithmetic in binary. I do think the proposed naming scheme is the biggest weakness however. As the process to convert between the spoken and written forms is relatively involved, it loses a lot of the written form's strengths. You can't very naturally think of the spoken form of a number in terms of whatever sized groups of bits are most convenient (e.g. size 3 groups are tricky); and the extremely simple operations of doubling and halving can completely change how a number is said. For finger counting it's very nice to just use however many finger i'm comfortable freely manipulating when counting something for myself, but it becomes an issue with communication. If someone shows me some various combination of fingers raised on each hand, how do I know whether each hand represents 3, 4, or 5 bits? Edit: I have an alternate proposal for naming numbers. You dismissed the idea of simply reading off the digits; however, that could actually work. If you pick short syllables and ensure the consonants between them can flow smoothly (e.g. no mandatory glottal stops), then you can read off bits quickly. For example, if you pronounce 1 like "wun" or "nun", and 0 as "oh", "no", or "wo" depending on the previous bit, then you can read a number like 1010 0011 like "wunowuno owowunun". (yes this means you'd say "owo". deal with it.) This way, you can group together digits in whatever sized chunks you want, like with the written form; as well as have easy bitshift operations; and between gaps you can *optionally* insert the "magnitude words" like hex,byte,etc. (like in base 10, the magnitude words are sometimes optional - you can say "three fifty" rather than "three hundred and fifty"). And you could start recognising the pronounciation of short bit sequences as words themselves.
I take some umbrage with the “efficient finger counting” argument, because finger counting is primarily a tool for teaching children. Binary finger counting (and even seximal finger counting) is more complex than simply counting the number of fingers raised. Bases are an arbitrary construct and humans don’t think in terms of them, so a child will never intuitively understand binary finger counting the way they understand “base 10” finger counting (though i would argue it’s more accurately described as base 1 finger counting)
Many cultures have counted in dozenal or hexadecimal on one hand. Many Asian cultures to this day count to ten on one hand and find the western way to count on fingers jarring. Your argument is eurocentric.
I’d argue the mentioned systems are still ultimately base-1 counting systems. They use 12 or 16 naturally ordered positions of the hand. Counting up to 12 with the dozenal thumb-thing is still just moving your thumb through a “number line” of hand positions. It still retains the simple linearity of western counting that’s lost with binary counting.
well actually maybe base infinity would be more accurate but my point is that neither 10 finger counting nor 12/16 one-hand counting have to actually deal with the things that make bases complicated (i.e. multiplying by powers of the base according to the position of the digit)
@@considerthehumbleworm During the Middle Ages doing arithmetics with your fingers was common enough that almost every book on mathematics had a chapter on it. For example in _'Liber Abaci'_ by _Leonardo Fibonacci_ finger counting is the first chapter. Systems to count up to 100.000 were around for quite some time. Conflating finger counting to what _we_ today in the west teach to toddlers is a mistake. Drawing conclusions from oneself about others rarely works out.
Finger counting is what we teach to toddlers tho. I think it’s a bad idea to offer binary finger counting as a replacement in those contexts. It’s useful to be able to count higher than only 10 in other contexts but it only muddies the water for learning basic arithmetic. Honestly, teaching toddlers systems that go up to 8 or 16 is probably preferable to 10 to ease the transition from “number line counting” to using actual positional binary (even binary finger counting, eh?). That transition’s a bit of a given with 10 finger counting and base 10, so it’s worth considering how that works with binary. Also we don’t have to hate each other because we disagree about numbers lol
Unary is quite interesting Chap 0) Numbers can be written as a series of Dots that follow some Line. This is obviously hard to read and long if the line stays straight, but if we have the Path the Dots follow not be Straight then we can Group them. We could just Grid the Dots with some set Side Length to group them. I also kinda like the idea of Coiling the path then Dots are all Coiled and can group parts by when the Spiral passes some spot, let's say the rightmost spot. Since as we move outward, if the Distance between points on the line remains the same then the Value Difference of each Layer of the Coil Increases as we move Outward, which is kinda cool. We could also just change the distances between them as we move Outward such that the Layers all represent the same Amount. Chap 1) Log base 1 of x doesn't work so this breaks here. But, if each digit can only be in one state then all digits contain no information (in the same sense that the first Binary digit contains no information) hence Unary is super efficient? Chap 2) Counting the Joints and Fingertips I can count to 38 on my hands, using a finger from the other hand to just point at the joint. I like to list it from my Right Thumb Tip to right Thumb Joints moving closer to the hand, then next finger etc. but it really doesn't matter. The only confusion I can think of is when the Index Fingertip is Counted, it might be Ambiguous as to Which Finger is being Counted and which is Pointing, and ig you could have one be on top or something. Other than that it's pretty easy to read and avoids the Recognisabilty issue. If you can't use some fingers, that's fine if ur joints are still there cuz u just point at them, and if ur missing fingers then u just count based on how many joints you have. Chap 3) Unary Arithmetic is Mindlessly simple. Addition: Just stack them on End (This might be more annoying with the spiral method to immediately visually see, but pretty easy with the Grid method). Multiplication: Just stack it onto itself the number of times ur multiplying it by. Division: Count thru the Dividend and reset every time you count to the Divisor and then ur left with the Quotient with some Remainder. Square Root: Just re-stack them into a square. U can do this procedurally by starting at the corner and expanding from there until u run out so u don't have to know the sqrt beforehand. Chap 4) No divisibility tests work as far as I can tell. Chap 5) I don't think you can do fractional expansions in Unary. Chap 6) BABABABABABABABABABABABABABABABABABABABABABABABA.... where each BA is a single number. Chap 7) Unary is obviously the Most Fundamental. Objects just Exist, there isn't really the Non-Existence of things to be Counted. Look to Monism for a Philosophical Exploration of One This is why we should all revert back to Unary.
The problem with that is that numbers start to get uncomfortably lengthy at an exponential rate. The number 41 in decimal would be 11111111111111111111111111111111111111111 and good luck trying to compress that
let's just go to base infinity, every number has its own unique symbol or something like that i guess or maybe instead of base 2 let's go with base 1 so every number you just have to count out like tally marks
59:18 I feel like using kibi, mebi, gibi, etc. for powers of 10n would be good for familiarity's sake, since it's common to approximate 2^10n to 10^3n, and the words already exist.
(RE: Section ||.) I find it quite difficult to actually understand what is going on with these "three four hex four three two one"- stuff. As others already have pointed out in the comments that these names for numbers are just too long and complicated to parse intuitively, I came across the fantastic hint that the human brain is capable and moreover the natural language does evolve to number names that exceed the number base (base-ten). E.g. eleven and twelve in English, which are not "oneteen" or "twoteen". For this, it has been proposed to just use a hexadecimal naming convention for this way to count. I sat down and get more thoughts on that till I came up with something comparable to that, what natural languages like English or German had done. A second thing I implemented is a distinction between numbers in higher or lowers orders/register, i.e., we use another name for some partial numbers if they are going big. Or for the multiplier numbers. (we'll come to that later, you'll see what I mean) The first number that has been significantly changed through my system is the number after twelve, which in decimal is call'd 13 (thirteen), but in binary shall be "five-eight" (Note that in binary the 1101 consists of 1000 "eight" and 101 "five"). I now present to you the further titling of the numbers in binary: 0: zero 1: one 10: two 11: three 100: four 101: five 110: six 111: seven 1000: eight 1001: nine 1010: ten 1011: eleven 1100: twelve 1101: fiveeight "five-eight" 1110: sixeight 1111: seveneight 1 0000: hex 1 0001: hex one... 1 1111: hex seveneight 10 ----: bihex... (pseudo-latine-graeco prefixes for distinction) 11 ---- : trihex... 100 ----: tetr(a)hex... 101 ----: penthex... 110 ----: seshex (probably the tongue twister one lol) 111 ----: septhex 1000 ----: octhex 1001 ----: nonhex 1010 ----: dechex 1011 ----: elhex (shortened eleven) 1100 ----: twelv'hex 1101 ----: penteighthex (from fiveeight) 1110 ----: seseighthex 1111 ----: septeighthex (from seveneight) 1 ---- ----: a byte.... and so on As you please, you can replace the interjacent hexes with nybbles or (or do both) use plural 's'. Final test: The number 1111 1111 1111 1111 looks like this: septeightnybbles-seveneight bytes septeighthex-seveneight Now you would argue, this is LOOONG. But the same problem applies to the decimal base with numbers like 999,999, which is: Nine hundred ninety-nine thousand Nine hundred ninety-nine. What do you think?
I feel like this system could be improved by using unique labels in spoken language for every permutation of 4 bits, effectively treating it like hexadecimal when talking. You already concede that spoken binary would create some very long number labels and offer quartery as a compromise, so you might aswell embrace the pseudo-hexadecimal nature of grouping 4 bits together by adding a horizontal line at the bottom by covering these groupings with a single label each.
Exactly! And since lower numbers don't have decimal based names (compare "eleven" to "twenty-one") we can start by borrowing those. In spanish we have unique names up to 15 (it's "quince" as opposed to "diecicinco") so we already have the firt 4 digits covered. From that point, I'd say the next 4 digits should have a prefix (so, I.II. would be something like hexasix), but we would need to find something that rolls of the tongue (I love stack for 64, but it sounds clunky as a prefix)
Numberphile convinced me a decade ago that dozenal was the best base. Then jan Misali comes in and shows how seximal was better. After being fascinated about the topic and thinking about it on and off for years, I thought I knew almost all there was to know about number bases, and was convinced for so long that factors of the base are the biggest determining factor for what makes a base good. I never thought anyone else could show me otherwise. This video showed me just how wrong I was, and that there was a surprising amount of unexplored depth to the topic of number bases. It's videos like these that I love the most, and I wish for this video to explode in popularity. Maybe binary is the best base after all.
really well put together video! these are some very compelling arguments for binary. you're right that I dismissed it for superficial reasons without second thought in my videos; the immense advantages it has for arithmetic and information density shouldn't be overlooked, and there definitely is a case to be made that these may matter more than the things I focused on. in the few contexts where the notion of an "objectively best base" actually makes mathematical sense as a thing to care about, binary is a clear winner.
I don't think I'm completely convinced that binary is necessarily the absolute best choice for a human-scale base (the "coincidental" advantages seximal has for working with small primes are just too good) but I am convinced that it can work as a human-scale base to begin with, which I hadn't even properly considered before. it definitely deserves a seat at the infinite table with the other SHCN bases.
please do a responde video! not only. for exposing more of your points but also for giving this guy more attention here on youtube
I think the answer then would be some more usable power of two such as 8 or 16
+
@@wilh3lmmusic thats basically what the video said, you can combine 2 or 3 digits of binary to make it shorter like base 8, but with all the advantages of base 2
@@leggyjorington3960 Not exactly. The grouping system in the video is distinct from hex or octal in a few ways.
First of all, if you just want to use hex, then the notation suggested is very inefficient. I'm not a big fan of how we usually write hex (0,1,2,3,4,5,6,7,8,9,a,b,c,d,e,f) but it's much more efficient than iiii iiil iili iill... etc. The same is basically true for octal.
Second of all, the advantage of his system is that it isn't hex or octal, it's both (and any other 2^x base). You can very easily re-analyse a number as any of those bases to get its benefits. Sevens aren't great in hex, but they are easy in octal, so just group by three bits. Octal sucks at fives, so just group bits by fours and treat it like hex.
Wilh3lm is, imo, correct that the benefits of the more flexible system are outweighed by the practical issues with actually writing these numbers. I spent several months using almost exactly the notation suggested in the video and it simply isn't practical.
drama in the counting community
lol
When are the diss dropping?
Things are heating up in the number fandom
6 finally cancelling 7 cuz it 8 9
I frequently see children using this system to express the number 4 to me when a school bus drives past. It's truly amazing to see something adopted with such enthusiasm at a young age and gives me a lot of hope for future generations.
Wouldn't they also be expressing the numbers 128 and 132?
@@mehulpandya4761 The most avid mathematicians on the back seat usually flaunt their counting skills in this way.
New euphemism just dropped!
Another fun fact, since microprocessors out number humans, and most of them do arithmatic significantly more often than the average person does,
technically binary is the most often used base currently.
My phone did arithmetic to like ur comment
I mean the term “use” implies some sort of conscious input so no
I'm a bit excited
a
Because Vötgil has so many vowels
I am waiting for a formal debate
vötgil :3
Because this episode is a first in a few ways
59:03 Excuse me, but four bits is a *nibble* (half a byte!). I will not drop the cute name. I love it.
fr agree, how wouldnt you wanna say 8DEC as "a nibble"
"gimme a nibble burgers"
"i would like a nibble kilo of ice cream"
"give me nibble apples"
@@im-radio "Can I just have a nibble of your sandwiches?"
"NO!"
If you wanted to keep it monosyllabic you could shorten it to "nib" :P
@@im-radio36332 right?
Hi, not sure if anyone’s mentioned this yet but your use of pitches following the harmonic series to accentuate numbers you’re talking about is absolutely incredible and I didn’t want it to go unnoticed.
I just appreciate how unnecessarily hostile the video is. It's honestly hilarious.
Yes, and it makes the thesis so much more compelling
One cute fraud.
yooo wikipedia bisexual lighting skeleton
having beef with base four for over an hour
I think I found my favorite genre of videos: the hour long math rabbit hole.
Videos such as this one, "HACKENBUSH: a window to a new world of math" by Owen Maitzen, or " The Continuity of Splines" by Freya Holmer.
Really loved the use of music and sound in this one.
I love both of the videos you mentioned so much, if you have any other similar ones to share I'd love to know! I can also go through my favourites and find similar ones if you want me to tomorrow ^^
@@codenamelambda The recent 3b1b explanations of light slowing down when travelling through matter are definitely a must-watch, if you haven't seen them already. Oh, and also the "Why can't you multiply vectors?" by Freya Holmer
we're honored to be compared to videos like these! glad you enjoyed
"Seximal may win a sprint, but binary wins the marathon."
So that's why you made a 1-hour video to counter jan Misali's 18 min one.
i'm not sure you understand how a response works
youre missing the point but its still funny
Guys it's a joke, don't take it as him being serious
kekw
@@katie-ampersand
short vid = sprint
Some fingers are significantly harder to extend individually than others. This applies both physiologically and culturally.
Ah yes, the four (l..) finger ;)
This point is addressed in the video. Using just three fingers on each hand still lets you go further than seximal, and binary is the only base where you can use any amount of fingers for finger counting
@@OMGYavanithree fingers per hand still means you can flip someone off, and arbitrarily skipping the middle finger would introduce ambiguity.
finger counting 19+113 💀💀
i can do it just fine, so... skill issue
The amount of times I got caught by the “‘well actually’ -You“ moments scared me. Every time I felt like I had a valid argument to make, there was a direct response to it.
+1 for playing overtones behind fractions
i loved that
I really don't like the way that binary is proposed to be written in this video. The bottom connection thing is actually really nice, but the whole "short ticks for 0, long ticks for 1, and downward short ticks for the radix point" thing seems like it'd be really prone to accidental slip-ups and unnecessary ambiguity, especially without some sort of guide on the paper.
Agreed. A good way to distinguish them could be to write 0 like the lowercase Greek gamma and 1 like the cursive lowercase L. It would be fast, less prone to errors and easy to read.
@@polymloth I just sat down to try and find a system based on this comment, and this was what I came up with lol, connecting adjacent characters like cursive. This is a pleasing analog of chiral topology, something like over- and under-crossings in a knot diagram.
You can use word breaks to indicate digit groups, and a slash to indicate fractions as usual (I don't actually think having a positional notation for fractions gets you much idk).
I also think it would pay to write numbers little-endian.
@@duncanw9901 Exactly what I had in mind!
the thing is, if natural languages have no issue making this distinction (take the word יוון in Hebrew), this shouldn't be an issue with numbers.
we can also always take inspiration from Hebrew crossword solvers and turn the short tick to a short cross.
Great video. Although the "speaking system" is quite flawed. So i made a new one.
Firstly, the biggest flaw is recursiveness. It's new and neat, but when you want to say a number to other person, it's better to convey number's magnitude right away. For example in decimal you would say "world population in 1975 was 4 billion and dot dot dot". However in binary it would be "three four three hex two BYTE two four one hex four three SHORT dot dot dot", and only when you say SHORT the person can sense the magnitude.
As a follow up, what if the person is writing the number down? For example when he hears "four int two..." how many zeroes should he put before writing down "two"? If the number is "four int two short" then 14 zeroes. If it's "four int two byte" then 22 zeroes. If it's "four int two byte short" then 6 zeroes.
Secondly, phrases are a bit bigger. A small number in decimal (255 - "two hundred fifty five") would be "three four three hex three four three": 22 symbols versus 37 symbols (or 6 syllables versus 7 syllables). We sometimes omit hundreds, so it's minus 2 syllables.
Thirdly, "speaking/writing system" interferes with the idea of grouping bits into groups of 2, 3 or 4 bits. System works with groups of 2 and 4, but does not with groups of 3. As it's said in video there is a learning curve, where person first learns arithmetic on group2 then group3 then (maybe) group4. But what if he considered group4 arithmetic too complex and stopped at group3 ? After he's done calculations on group3 he has no choice rather than regroup the whole thing and only then say the number out loud.
For the first problem I'd kinda go traditional method (millions, billions, trillions etc.)
For the second problem I'd compress numbers up to hexadecimal digits
For the third problem I don't know. Either group3 people will have to regroup, or make another speaking system for group3 representations (which is quite bad).
The digits
Imho it would be worse to use digit name, that we already use, so I gave new names, trying to reflect "binariness". Also these numbers should be fast and easy to pronounce and phonetically distinct from each other , because they will be used a lot in speech. I'm not a conlanger, but I tried my worst
wan du ti ro
rówan ródu róti ko
kówan kódu kóti kro
krówan kródu króti hes
Here ó is a stressed o. These words represent names for following hexadecimal numbers:
1 2 3 4
5 6 7 8
9 A B C
D E F 10
I used "hes" instead of "hex" for sixteen, because I think it's faster to pronounce it this way. For 0 we could use "zero"
I think "r" prefix for 4 and "k" prefix for 8 work quite good with binary.
With this system we can count up to 255 (in a similar way we count up to 999 in decimal without involving power names, such as thousands, millions, billions).
"hes" is used as a connecting word between quartets (similar to "hundred" in decimal). If the right half of number is zero, "hes" is omitted.
........ - zero
.....|.| - rowan
|||||||| - kroti-hes-kroti
...|.... - hes
..|..... - du-hes
|..|.||. - kowan-hes-rodu
Now for power names.
Let's call collection of 8 bits as bunches (not bytes cuz we will use this keyword). Bunch in decimal would be 3 digits.
Every bunch in decimal is followed by a power name (thousand, million, billion...)
If we do the same in binary with suggested names (byte, short, int, long, overlong, byteplex), these "power names" (technically power phrases) will be like that:
256 ^ 1 = byte
256 ^ 2 = short
256 ^ 3 = short byte
256 ^ 4 = int
256 ^ 5 = int byte
256 ^ 6 = int short
256 ^ 7 = int short byte
256 ^ 8 = long
(We suppose that most significant words come first)
Here I would suggest other naming system (which in general will have more syllables). However each power name will be represented with one word, pronouncing these names I think will be easier, since "int" "short", "long" are not pronounced well together. Also every name will end on "-yte", indicating, that it is indeed a power name:
256 ^ 1 = byte -> byte
256 ^ 2 = short -> plyte
256 ^ 4 = int -> fryte
256 ^ 8 = long -> ksyte
256 ^ 16 = overlong -> znyte
(I did not come up with an alternative for byteplex)
Now instead of "overlong long short byte" we would get "znyte ksyte plyte byte". But we need to combine these words. The rules are:
1) Last word gains prefix "o-"
2) Other words turn into prefix form
The prefix forms are:
plyte -> pil
fryte -> fer
ksyte -> kas
znyte -> zun
So by these rules "znyte ksyte plyte byte" will convert into "zun-kas-pil-o-byte", or "zunkaspilobyte". Here are first 14 power names:
byte
plyte
pilobyte
fryte
ferobyte
feroplyte
ferpilobyte
ksyte
kasobyte
kasoplyte
kaspilobyte
kasofryte
kasferobyte
kasferoplyte
And now one example with all of this: distance to the Sun in nanometers:
|... ...|||.. ...|.||. .||..||. ||..||.. .|.|.|.. .|.||... ..|..||. .|.|.||.
"ko ksyte wan-hes-kro ferpilobyte wan-hes-rodu feroplyte rodu-hes-rodu ferobyte kro-hes-kro fryte rowan-hes-ro pilobyte rowan-hes-ko plyte du-hes-rodu byte rowan-hes-rodu"
I'm pretty sure if we had developed binary & hexadecimal counting, we would not be translating numbers from base 10, or from metric, especially since considering the metric system was developed in a world where base 10 was well-estsblished. What if we had established a measuring system based on 100,000 [in hex) of the diameter of the earth through the poles? The "hex stick," if I may, would be about 16 cm in length. A little small, but still usable. As much as we like to make up words, I'm sure we'd still have million and billion, or an equivalent in hex language & a little different in magnitude. Mil & bil are kind of arbitrary labels, but if hex had an equivalent, you could again feel the magnitude. Let's just pretend for a bit that hun, thou, mil, & bil applied to the number in the 3rd, 5th, 7th, and 9th digit respectively.
Distance to the Sun would be about 37 milhex, if I may. A population of 4 billion would be EE milhex, which is just under 1 bilhex. Magnitude problem solved. 🙂
@@livingpicture Well yeah, we solved magnitude problem in our ways. But thou, mil, bil are names for 3*n hex digits (or nibbles) which is kinda arbitrary. I tried to follow the idea in the video: to establish names only for 2^n digits. The traditional approach (3*n digits) is probably easier to remember, but idk, I didn't learn my system :)
someone please like this comment later today so I remember to come back to this amazing comment
I kinda prefer this (but what do you guys think about this?):
0 = zero
1 = one
10 = 2 = two
100 = 4 = four
1000 = 8 = eight | 1111 = 15 = fifteen (there should be alternative names for 10[dec] to 15[dec])
1,0000 = 16 = *hex*
10,0000 = 32 = two hex | 1111,0001 = 241 = fifteen hex one
1,0000,0000 = 256 = *byte*
1111,0111,0101 = 3,957 = fifteen byte, seven hex five (fun fact: this is 3 syllables shorter than its decimal name)
1,0000,0000,0000 = 4096 = hexabyte (or hexbyte?)
1111,1110,0100,1001 = 65,097 = fifteen hexbyte, fourteen byte, four hex nine
1,0000,0000,0000,0000 = 65,536 = *"short"* (there's probably a better alternative name for this we can borrow from Computer Science)
This gets absurd...
10,0000,0000,0000,0000 = "two short"
1,0000,0000,0000,0000,0000 = "hex short"
1,0000,0000,0000,0000,0000,0000 = "byte short"
1111,0000,0000,0000,0000,0000,0000,0000 = "fifteen hexbyte short"
Maybe we should think of an alternative for "hexbyte"?
@@Gelatinocyte2we already have:
ten
eleven
twelve
draze
eptwin
fem
James Grime convinced me of 12, almost 12 years ago
Misali of 6, almost 6 years ago
You just convinced me of 2
Who said there was no progress in history?
in 2 years some will convince you of base 1 and then 1 year later someone else will tell you base 0 is better, on which there will be a imediate response claiming base -1 is optimal, while then you realized that you actually had memories of someone claiming 1 year ago that base -2 was better and so on
2 months ago
@@LucasFerreira-gx9yh he may end it up studying Gaussian integers. Hahahahaha.
the pattern shows that in 6 years youll be learning base .5
@@smoceany9478 12/(t/6)!?
honestly, this video has now converted me into a true binary supporter
also i did not expect all of the arithmetic stuff with binary to be SO simple and easy to do
no way its the crystal garden golden person
i frogelined and peacelined
oh hey it's the |..|eg guy
I did not expected that usage of binary in human scale is not only possible, but pretty convenient too and it has a lot of advantages
The bit about the square root algorithm made me literally get up out of my chair, scream "what??" at my phone multiple times, and roam around my apartment for several minutes rethinking life
I shared your astonishment when I realized there's an algorithm for that. It just shows how the compactness of binary makes it so versatile.
In my humble opinion this video goes into the youtube's mathematical hall of fame.
A deep and new point of view shedding light on a topic that everyone can relate to yet few thought consciously before.
I have no other words than to thank you for your work.
npc ahh comment
@@mikechad27npc reply
Programmers and hardware engineers have been doing this forever, using 8 or 16 as the "compressed" written format, but dropping back to bits for actual arithmetic, especially when making machines to do so. Anyone in those groups worth their salt can convert from 0-F their binary quartets and back intuitively, and when learning to do that, often use fingers in ways that resemble your grouped bits. The main problem with binary as a written or spoken system is that its hard to read at a distance, and the octal/hexidecimal are "error correcting" in that a smudge on a piece of paper or a dent in a sign makes simple tally marks unreadable, but leaves letters and hindu-arabic numerals mostly in tact.
She truly made a channel and a dedicated trailer for the counting video essay. Dedication like that deserves my full attention and like!
Check the credits! The writer/editor has a feminine name and isn't the narrator
@@timperkin9 thanks! I fixed it in the original comment :) I never check the description or anything so I missed that one :(
The fact that the little sound effects match with the number on screen and the harmonic series is a very nice touch
I'm incredibly surprised this video didn't mention that the GREATEST advantage binary has, is that it's a system we could ACTUALLY switch to without nearly as much hassle as any other system.
I'll be honest, I went into this video thinking "huh, interesting. Id like to see a new point of view", got to the twist reveal that it's about binary and went "ok, this is either a joke or I'm in for an interesting if unconvincing response", but now you've really convinced me. Holy cow, I had no idea what I was in for. The counting is fun too since I don't have to remember so much, and somehow these numbers are easier to understand with my dislexia too. They could also be made easier to understand for dyslexic people with a few simple tweeks so that's comforting. And on top of all that, binary numerals would be so fun to make fonts for, as you can pretty much make the symbols whatever you want as long as one is "less" in some way than the other. Hollow/full circle, down/up arrow, Mario/Luigi, literally infinate options lol.
Anyways awesome video, please make more! I'd absolutely love more specific video lessons on how to use binary with your numeral and naming systems!!
STOP DOING BINARY
"Hello, I would like 🙆🙅🙆 apples please!"
*They have played us for absolute fools*
I think language is the part where I'm least impressed here.
There are studies suggesting that languages that express numbers in fewer syllables lead to people doing faster mental arithmetic.
A lot of languages have specific words not only for the singular digits 1-10, but also for some numbers into the teens. English has "Twelve", for example, instead of "Twoteen". French has a single digit words for eleven, twelve, thirteen, fifteen, and sixteen. And I think there's also some linguistic benefit to having words for numbers like twelve rather than "ten two" than just a historical linguistic artifact--small integers are just the numbers we will use the most, being able to express them quickly and unambiguously is inherently valuable, even if it makes the language harder to learn than Toki Pona. It's also worth noting that languages tend to have specific words for 20, 30, 40, 50, etc. In English: twenty, not "two ten"--now there's no syllable advantage here (although in some languages the word for 20 is one syllable) but there may still be a linguistic advantage to having a separate word--even if hearing is an issue, you'll never mistake "twenty" for "ten" or "two".
Rather than express number words in base 4, I think it would probably be advisable to have language work in...probably hexadecimal honestly? Either that or Octal, but I would lean hexidecimal for two reasons. First because it's 2^4, and 4 is a power of 2, so it'd be easier to deconstruct if you were already thinking in binary. Second having number names for numbers higher than 10 is fairly reasonable, cause a lot of languages already have that (1-12 in English and German, 1-16 in French and Spanish).
But honestly, this is all hypothetical, really. People are never going to switch number bases unless governments force it by changing money to a different number base. As long as people have 10s, 20s, and 50s in their wallet, they're going to think in decimal.
you have got to make a video for the better way to say numbers
I have to say. I was extremely skeptical at first, but the elegant way of writing binary numbers you came up with really sold me
Try it for a month and you will change your mind. It's a nice idea, but I can tell you from experience that it doesn't work in practice.
@@Salsmachevthis video's been out for like 3 days
it's just too bad they're completely unreadable at a glance
@@mrosskne I'm curious if you're saying this having used that notation for a while or not
@@starstufs I literally just said "at a glance"
I always felt, instinctively, that binary "should" be the best base, but it just seemed too cumbersome to use in practice. Thanks for your excellent work.
Drama in the black background white text youtuber community
You present your arguments well, however: I am nonbinary.
So I'll have to stick with Seximal, or maybe Balanced Ternary for the small number and negative advantages
Perhaps you would like base negative two, or negabinary? It uses alternating negative and positive powers of two so negatives can be written just like positives without the need of a minus sign.
This comment is pure perfection.
@@lemoneer7474 Wouldn't nonabinary be base 18?
I'm non-decimal. It's a shame no-one supports me! /jk
@@lemoneer7474base √-1 -1 (called i-1) works well too
Sadly, any way of speaking base-2 numbers will be non-portable. Some ancient cultures will say that int is 2^16, while modern ones say 2^32. And then there's the oft-forgotten cultures where byte (or char, as they'd say) is e.g. 2^31. Instead the names for the double powers of two should follow a simple and memorable system, e.g. 2^16 could be "int underscore least sixteen underscore tee". Since that's a mouthful you should also introduce "intmax underscore tee" as shorthand for "the biggest power of two I feel like thinking of right now".
Devastatingly nerdy comment, love it
just use the x86 names: byte, word, dword, qword / quad, and dqword / double quad. i can't remember what the next two were but i could've sworn it was ddqword and qdqword. super convenient and not confusing at all
@@rubixtheslime after qword is either oword (octa) or xmmword (128 bit simd register)
after oword is xword (hexadeca) or ymmword (256 bit simd register)
Based and Embeddedpilled
More like "the biggest power of two I feel like thinking of right now, except I must never change my mind or everything I've said before will break".
I got converted to binary after seeing the square root algorithm, knowing how complex it is in base 10
gd guy
its funny cuz the main narrator kepe also made yBot, small world huh
@@g_vost WHAT
I feel like I've stumbled on a piece of forbidden knowledge.
The funniest thing was trying to come up with some counter-arguments while noticing I had already used a binary representation of octal (tri-octal to be more precise) to convert an alphabet, and already figured on my own how easy it was to apply a vigenaire cypher mentally on it thanks to how trivial it is to do arithmetics on it
re: chapter six. i'm not sure your naming scheme really does a great job here. it definitely feels like "ninety one" is less information to parse than "four one hex, two four, three", and in general, standard names tend to feel less cumbersome. perhaps there's a less mathematically elegant, but more linguistically practical, way of handling binary number naming.
"four one hex, two four, three" is more analogous to "nine ten one" like some languages do.
For me it's more natural with -s on the things that should be plural, so four one hexes two fours three. (like "nine tens one")
I think something that could help solve the cumbersomeness is basically doing base 16. So make all numbers less than hex compound words, so you can treat each as a single unit.
Like ninety one would be four-one hexes two-fours-three.
Compared to base 10 English, the digits are more syllables, but that happens in other languages. Like in greek the numbers up to 10 are all 2 or 3 syllables. With this system the numbers up to 16 are all 1, 2 or 3 syllables.
The fact that it's base 16 means it's actually less cumbersome than base 10 for large numbers.
It also keeps all the elegance, since it's just a conceptual reframing to help interpretability.
There's still a slight problem that, for example, "four-one hexes two-fours-three bytes hex one"
could be read as "four-one hexes" + "two-fours-three bytes hex one", you'd need to say "(four-one hexes two-fours-three) bytes hex one" somehow. I think this is just a problem that comes with the b^(2^n) system, you might have to switch to b^(kn), maybe 2^(4n) so powers of hex. Tho you could have the power names be like hex, byte, hex-byte, short, hex-short, byte-short, hex-byte-short, int and so on, so it'd be "four-one hex-bytes, two-fours-three bytes, hex, one".
@@d.l.7416 in written form, using the 2^(2^n) system, your latter example isn't technically ambiguous. but when spoken it's unclear what order of magnitude is being referred to at any point until the number ends, because at any point while telling you the number, someone could just say "short" and suddenly the number you were thinking of is 16 orders of (base 2) magnitude too small, etc.
my proposal for naming numbers is to simply read off the bits, using short syllables that can be flexibly strung together and said quickly. If 1 is pronounced like "wun" or "nun", and 0 like "oh", "wo", or "no", then a sequence like 1101 1001 becomes "wununowun wunowowun"; and you could optionally insert the magnitude words like hex between gaps. This mirrors the written form and retains its advantage of being able to group digits into whatever sized chunks are most convinient.
Hearing "nine ten one" in English is a little hard to parse, but in my native Hokkien where numbers work exactly like this, it's perfectly natural (九十一 káu-tsa̍p it). I think this just goes to show that it's all a matter of getting used to it. The "four one hex two four three" system is perfectly fine.
@@jfb- I think that would become very confusing very quickly. People already mishear fifteen (wunununun) and fifty (wununo nowuno) in decimal numbers, and you're expecting them to parse wununowun from wunonunun in fast speech (not to mention that both of those numbers are a mouthful). At the very least I think you should pick sounds with greater contrast. Maybe something like ko and mi, where the sounds of each differ in as many features as possible. You'd probably still have people getting mimikomi and mikomimi mixed up if you rattled off a couple of mikokoko-bit numbers, but it would be an improvement.
16:00 I thought I'd heard that somewhere; it's a "trick" used to store floating point numbers in the IEEE standard; the leading 1 in the mantissa is implied
i dont think so. they have to represent 0 too
They represent 0 by setting the exponent to its lowest possible value, at which point that leading 1 is treated as a leading 0 instead.
jan Misali said this in his floating point number video iirc
In the end, it's not about "which system is generally better", but "which system is better after I came to a conclusion which were my criteria and priorities".
Nah. It's more about "which system am I already using".
However, I think binary wins when we start talking about how to TYPE these numbers out.
As a programmer seeing hex byte short int used as power names is both horrifying and amazing
I think the proposed method on how to say the binary numbers has a couple of major drawbacks:
1) The recursive, non-linear conversion between words and symbols makes it hard to dictate, and non-trivial to write a dictated number down.
2) The symbolically easy doubling becomes unintuitive, e.g. 3 4 1 doubled becomes H 2 4 2
3) It can bury the most significant part towards the end, for example 3 4 2 H 3 4 2 B. You have to listen to all of the spoken numbers to make sure you're even in the right order of magnitude.
as far as I can see, these are all issues with existing spoken number systems. this system isn't much different than the one we use for decimal
i think 1 and 3 come from the fact that it's a 2^2^n system instead of a 2^kn system.
you could instead make it a 2^4n system (powers of hex), but name the powers using the 2^2^n system. So hex, byte, hex-byte, short, hex-short and so on.
like you'd say for example, three-fours-one hex-bytes three bytes hex two-fours, which is basically just how standard english base 10 words.
@@thebestwaytocount These issues are much more significant in binary since they show up much earlier on than in something like decimal
Recursive algorithms are simple algorithms. No one that has any experience reasoning about procedures thinks any different.
Take a cue from the world of programming, where our spoken numbers are base 16 but our math is base 2. If we used the numerals from this video it would make the conversion trivial.
There are 10 types of people, those who understand binary, and those who don't
Underrated comment
And those who weren't expecting a ternary joke
There are 10 types of people, those who understand the hexadecimal system, and F the rest.
@@MuzikBikethats still 2 #binaryandseximalareslaynumbersystems
Wrong, there are |. types of people
43:29 "Traditionally, to notate a recurring fraction, the entire recurring segment is marked, but that doesn't make a lot of sense: It's simpler to just mark where it starts."
I think your opinion comes from the age of typing and digital text representations. In hand-writing, marking the recurring segment makes more sense for the following reasons:
(1) Often, you will get this repeating decimal from just long division until you enter a cycle, and adding a line of the repeating part is something you can do after you've written the number, which doesn't require you to erase anything or rewrite the number, and also separates the repeating segment from any digits of the next repetition you may have written before realizing it was repeating. This doesn't apply if you're rewriting the number somewhere else, but if you're copying a hand-calculated quotient with a long repeating decimal, you'll want to use some mark like the over-bar to mark the repeating segment of the decimal anyway, so any other notation like the "r" you're using will just be another notation to learn IN ADDITION to something like the over-bar. Thus, I think this is the most important reason.
(2) In hand writing, writing a line over several numbers is not significantly harder or more time consuming than writing an r before them. It's actually a slightly simpler shape than an "r", though the length and precision required mean I'll just say it's about as hard overall.
This is different from typing or digital text, where putting bars over numbers requires special characters or text formatting, both things that are much more of a pain to deal with than just typing "r", both just in typing it, and in the sneaky problem that things like this sometimes won't render properly on other people's computers or on printers. (With typewriters, it usually requires some kind of tricks involving typing over the same text I suspect.)
That being said, one downside of the over-bars specific to hand-writing is that, due to the imprecision of handwriting, it can often be unclear to readers exactly which numbers the over-bar is over. In typing, this would only really be a problem if your method of creating over-bars was bad, and using preceding "r" would get rid of this issue in handwriting (as would using parentheses or circling, although those aren't quite as convenient to add on to an already written sequence of digits as an over-bar is).
In my school they taught us to write repeating digits usung brackets:
1/7 = 0.(142857)
This eliminates the precision problem as brackets are easier to write and mark exactly where the repetition begins and ends
i fully support the use of "stack" as an official name for 2^6
10^110
and i use nibble for 16
@@peridotisthegoat96or hex
Back for a rewatch. This is levels of autism I strive for
fr lol
yess
on my third rewatch rn
I audibly laughed at 5 in the morning when you proposed “a stack” for 8^2, but it sounds only fair when you think about it
MINECRAF???!?!??!!
wow I didn't get it on the spot that's just super funny x'D
Finaly managed to crack the text at 8:20 on my own.
The substitution is 啊 = A, 痹 = B, 雌 = C, 低 = D, 婀 = E, 付 = F, 佮 = G, 喝 = H, 乙 = I, 咳 = K, 刕 = L, 冪 = M, 妳 = N, 我 = O ,仳 = P, 儿 = R, 絲 = S, 偍 = T, 無 = U, 予 = V, 劸 = W, 牙 = Y.
And the actual text is that one comment from jan Misali's Ido video:
" YOU HAVE GOT TO BE ABOUT THE MOST SUPERFICIAL
COMMENTATOR ON CON-LANGUES SINCE THE IDIOTIC
B. GILSON.
DID I MISS THE ONE WHERE YOU SAID WHICH CONLANG
YOURE FLUENT IN AND READ AT LEAST THREE TIMES A
WEEK AND CAN READ NEW BOOKS IN EVERY WEEK OF
EVEN ONE YEAR OR LISTEN TO RADIO SHOWS IN EVERY
WEEK? NEW RADIO SHOWS? "
this is gold. thank you.
interestingly the characters pronounced in chinese are close to the sound of their corresponding letters
amazing dedication
I went into this with a strong preference for hex. I got pleasantly surprised by the suggestion of binary, and when you started grouping the bits it's essentially a multi base system around binary. Hex is really just groups of 4 bits, which is why it's good. I think the naming would sound more natural if it's done for groups of bits instead of what's suggested here
16:14 this is exactly what floating point does. It doesnt store the leading one to save a bit, which allows the number to have a bit more precision
I love the notation system you created and the grouping shorthand, it’s very elegant and also makes the fundamental patterns you discuss visually obvious at a glance, without having to translate into numbers, even for someone totally new to the notation without having built base specific intuition. I could easily learn to do a lot of that math with your binary notation without converting into or out of decimal, which makes an excellent argument for it’s naturalness and simplicity
Don't you love it when you have that one really strong opinion that no one else cares about, but then you stumble across an hour long video essay about it at 3 am
It's my favourite feeling tbh
> that one really strong that
I think the noun is missing.
Regardless, yes.
@@zairaner1489 oops
i would love to see some examples of this binary notation used in some common settings.
some examples
>in a car for speed and distance.
>prices and measurements in a grocery store.
>numbers in a video game.
>a deck of playing cards.
that batch of 4 trick seems like it would be very readable.
i would also like to see something like "babies first numbers video", that would really show how easy binary actually is.
a webpage for "try math in a new base" could be a nifty demonstrator for it's usability.
i think the symbols themselves need a direction notation, the underline may be enough but I'd like it more obvious.
I made a userscript that you can find on greasy fork called "convert to binary" that attempts to convert numbers on webpages to their binary representation. The only pitfall of it is that it doesn't accurately convert non-integer numbers (like it'll turn 10.32 into lılı.lııııı)
First speed limit sign you see that says 10110 is basically the end of the emperiment.
@@jibbjabb43 well you wouldn't be using Arabic numerals for the task, you would use binary combs like the ones showed in the video.
@@ARockRaider It doesn't look any better. The issue here is both conversion and size. It's pretty silly to suggest that using rather indistinguishable characters is somehow better here. There is *some* ability to parse a misread sign simply becuase of possible interpretation, but that's because we're operating within a presumed 'field' of possible speeds.
It's also more important that I used a number like 22, which I can defend, and you instead focused on what the digits look like. Which tells me you either didn't take my critique seriously or you are too predisposed to the idea of it working without considering it's real-world applications.
@@jibbjabb43 I had assumed that you picked a number to look the most outlandish, i was pointing out that you wouldn't use actual 1s and 0s and that you would be using a notation that makes the numbers every bit as clear.
that you picked 22 makes my assumption of an intentionally outlandish number very clear, on top of that the exact speed would be adapted to the number system 20 for example would be i .i.. and as a group with proper underlines would be very different from 40 or i. i...
neither look anything like 30 or i iii.
but i expect that you wouldn't be using multiples of 5 for speed-limits, rather you would pick a new batch that always stays neat and round.
Amazing. Genuinely, i was amazed multiple different times. I'm a computer science major myself, I had already discovered the bit about multiplication, but the section on factoring? I had never thought to do that, and it blew my mind. Division? So much better than base 10. Your notation is beautiful, I'm definitely adopting it for when I do binary work, and I'm geniunely strongly considering switching to binary in my everyday life.
1:07:16 That Toki Pona fact at the end... This man doesn't know mercy
I like how we all do this research but will still all ever only use decimal casually
things heating up in the counting fandom
14:49 I can't speak for jan Misali, but I think he probably just meant that if you write binary with Hindu-Arabic numerals, the numbers get unwieldy in size, and base 4 in Hindu-Arabic numerals is a good way to compress it.
15:36 This fact isn't what makes binary the most efficient, it just makes it the most efficient by a long shot. Even if we didn't consider the fact that the leading digit can't be 0, by replacing b-1 with b in the formula, binary would still be more efficient than base 3.
side note: I don't know how we're measuring which is better rigorously, even though it's clear intuitively. Because the graph for base 3 does sometimes dip below the graph for base 2, like for example between 8 and 9. Is it the one which is the lowest for the most numbers, which has the lowest average, or what?
16:27 I wouldn't be so hard on seximal. jan Misali admits that in seximal, numbers are written longer, and the justification isn't radix economy, it's that the square base is small enough to be used for compression, it's only too large to do basic arithmetic.
In fact, I think it doesn't make all that much sense to use radix economy as an argument at all, our brains don't simply look at the representation of a number and take a specific amount of time to process each digit based on the log of what they expect the digit to be. Although it would be fun to see a study comparing each base in its own writing system and the speed of writing or reading numbers.
TLDR: binary > every other base because binary < every other base
unary.
holy shit, *holy shit*, i think this might be my favourite video. so so so many cool points brought up. so many new ways of thinking that were a delight to be introduced to. it has like, all my favourite things. this video is a masterpiece
IKR?
"Alright buddy, you've pissed me off."
*counts to 20 in binary using my hands*
counts to I .I.. using my hands*
I really loved the video, but I think, as many commenters have pointed out, that the legibiliy of your line system could be improved. Some might even compare it to a certain Minecraft optimised number notation system.
My suggestion would be to change out the short line for a small circle. This not only de-clutters the lines by separating them more cleary, but also prestents a great opportunity for ligatrues utilising the existing latin alphabet and also reducucing stroke number. or for l. and and for .l maybe for l.l, and maybe for .l. Keep in mind the arcs would be proportionally smaller in written notation
I'll flesh this out further and come back to this comment later, I'm expecting some grouping and priority conflicts, but there might be a nice way to deal with them.
"just use only three fingers on each hand" leads to communication issues. The great thing about addition only decimal finger counting is no matter how you count you always get the same result. It is unambiguous and inclusive to everyone who has any fingers. But yes it also means that you can only communicate very few numbers at one time. Basically decimal counting is TCP (sacrificing speed and efficiency for making sure it is as unambiguous as possible within the system) while binary counting is UDP (so prioritizing speed and data density while risking the wrong message being sent)
18:14
What happens, is that you'd stop being able to count it easily by sight at a distance - counting "how many fingers are extended" doesn't care about positioning, binary counting does.
And the issue with this, is that when using hand signals to convey numbers, you're probably doing that _because your voice can't reach the other_. Most situations where that happens are scenarios in which stopping for a couple seconds to reorient & count the exact numbers for yourself is an awkward use of time.
As an example: if I am on a construction zone with operating machines, and I need the handyman to replanish the third team with beams, screaming "BRING BEAMS TO THREE!" will probably be only partially heard, and after hearing the expected "WHAT?", being able to just raise three fingers while hollering "BEAMS TO THREE!!!" adds a layer of important redundancy, as the handyman will be instantly able to note that one word is 3, and the other is the object to bring, allowing them to process it better across the other sounds.
And yes, this is also true with binary, but with that system you'd be forced to use and read for the specific fingers which hold specific meanings, which can easily mean that the handyman needs to ask clarification _again_ because either they didn't catch if the fingers raised were for 2 and 4, or 1 and 2, or they caught that, but not what to bring.
It's already hard enough with all the redundancy we use, and needing to clarify and reclarify everything is stressful and awkward for everyone involved.
In this sense, using binary for finger counting is strictly worse.
Seximal, or well, the other counting methods named came about because of that need to balance redundancy and information density.
Sometimes, being able to alternate between asking for 2 of something and 60 of something is important, and between the thumb being quite ubiquitous and only needing to pay attention to "which hand are they raising", reading fingers at a distance using only 4 parts, is plausible.
Using 10 parts, making order important for every finger, is simply too awkward to use.
However, I will concede some other thing that's reasonable: if instead of an specific number you are asking for a general amount, then having each finger imply a different order of magnitude is useful, albeit there's already an order-independent system for that in which the amount of fingers raised is how many 0 are after a 1.
So yeah, binary is really damned good for counting... In paper and computers. It isn't good for hand counting tho, but honestly it doesn't need to be - hand symbols can simply not change alongside written ones.
Some notes as I go:
- I see you don't want to be able to write the numbers 0 and 1, if the leading digit doesn't matter for information.
- You seem to be neglecting the human tendency to ossify details. A binary system wouldn't stay a binary system for long. Within a few generations, eople would start writing shorthand octal or hex with a couple strokes rathet than four or five, and those would become the basic unit. That grouping trick is actually a negative when you look at language evolution.
- "What's the Most Commonly Used Prime Which is Incompatible with This Particular Base?" got a laugh out of me, good work on that one!
- Your repeating symbol and your grouping marks would be super easily confused in hasty handwriting. That's not an inherent binary problem though, just one with your notation.
- 2⁴ should be called nybble or nyb. It's cuter and sounds better in practice.
- ... I both love and hate you for pointing out to me that stack is a number name. It is though, saying something like "three stacks twelve" is perfectly normal in minecraft contexts.
- I think seximal is still more aesthetically pleasing, to be honest. And that matters because the only context switching away from decimal will ever happen is art. ... though balanced ternary might win there, it's just the right amount of weird to be really fun.
i love this comment so much. Agreed with everthing you said 😂
For your first point: I do agree that they handled this poorly. But I think all is not lost. Instead, you can think of what they are talking about as a consequence of the fact that simply knowing the bit length of a number tells you the value of its most significant bit (which has to be 1). The “length” of the number does not give you so much information in other bases, this is definitely true.
As for your second point, so long as people don’t forget where these ossified forms come from and they are visually similar to unossified forms and can be easily decomposed into them, then I don’t think this is actually an issue.
Since the measure of information per digit is considering all infinitely many numbers and the number of single digit numbers is finite, the amount of extra information carried by that first digit tends towards zero in the limit of considering all numbers since a finite portion of an infinite space is always 0% of the total space.
I think this means that when considering information on the long scale this simplification is fine, since the context of the discussion was already the long term efficiency of a base and not its efficiency writing small numbers.
@@keldwikchaldain9545 1 and 0 are kind of important numbers though.
i love the binary multiplication table because it's also the logical AND gate
This feels like an induction into a religion. Now I see how things should be, and my duty is to convert the non-biners to the true light of one and zero.
The thing that fascinates me the most is that base 2 is every base 2ⁿ at the same time, so it is capable of using properties from all of them.
Amazing video.
Great video! I wasn't even half way through and you had me convinced. After watching the procedure for the square root, I managed to find a way to get any integer root! It works in every base, but its way simpler in binary.
Basically instead of separating the original number in pairs, you separate it in base-of-the-root groups, and then instead of checking if its larger than or equal to the number above with a 01, you check if its larger than or equal to the following sum:
sum from k=1 to q of (q choose k)*(2p)^(q-k)
with q being the base of the root and p being the number above.
I think I might be the first person to discover this method, the method for square roots obviously already existed but I couldn't find anything about cubic or any other roots using this digit-by-digit method.
no matter what you think, base 10 is always the best (as long as you read it in your base of choice, of course)
got us in the first half
This video is exactly 200 minutes in seximal. Cheeky.
Are minutes a seximal unit?
@@Arbarano(kinda sexagesimal)
Grouping "bits" by up to for 4 is also pretty practical in a human sense, since around 4 is the natural range of human "subitizing" i.e telling how many of a thing there are at just a glance
Actually, I have a more general counterargument:
Redundancy is good, actually. All of these arguments demonstrate that binary is the most efficient base in a similar way to how Ithkuil is the most efficient language. Sure, that might be true on a technical level, but if you miss more or less any information, it's harder to recover. For example, something weird happens with the HDMI on my TV, so the edges of the screen get cut off when I try playing a Switch game on it. But because all of the Hindu-Arabic numerals are fairly distinct, I was still able to follow the timer in the Star raids in pokémon, despite the top half of the number being cut off. Meanwhile, if the top half of your binary numbers got cut off, that'd be it. There would be no way to distinguish numbers anymore. Or similarly, consider seven-segment displays. They're actually *horribly* inefficient, since you can display 128 distinct figures with them, and we only use 10. But because we use so few, depending on which segment goes out, you can still distinguish most numbers. And even in a case like the middle segment going out and making 8 and 0 indistinguishable, context clues help. If you see 1-"thing that looks like it's supposed to be a 9" change to 10, you can infer that it's probably counting down and is probably supposed to be 18. But if it changes to "thing that's probably a 2"-0, you can infer that it's counting up and that it *is* supposed to be a 0.
Or similarly, finger counting. Functionally, "binary" finger counting is actually duotrigesimal. No one's going to actually think of each finger as its own digit, so you're effectively expecting people to learn 32 different hand shapes. So if you're just using hand counting to show someone a number, like a kid going "I'm this many", it's going to be less efficient compared to just holding up a number of fingers and subitizing. Or similarly, if you're actually counting up or down, with both seximal and jisanbeop, you only ever need to change either 1 finger at a time or all 5 fingers. (And with all 5 fingers, it's also always either resetting the hand or changing 4 fingers to a thumb) Meanwhile, with binary finger counting, you might have to change any number or combination of fingers. And that's actually so well known of a problem that Frank Gray came up with the idea of reordering binary numbers so you only ever have to change 1 bit at once all the way back in 1947. (With the idea itself going back to at least the 1870s)
I actually agree with most of your points, but have a few nuanced points to make.
1) This one's simple, actually using binary doesn't mean blocking out half the digit would leave it unreadable. Binary is actually pretty good about that, you can *always* tell if you can see part of the digit.
2) I use decimal for readability purposes, but I find binary a better system for doing simple counting tasks. Remember that you could use different systems for different purposes, it's done in some cultures in other areas, such as speaking one language and writing another. (Looking at you, Switzerland...)
3) While your argument about finger counting has some merit, I actually think that's only a problem for children. Since we're actually using a *unary* system for that task, binary finger counting is strictly better but also more complex. (Seeing each finger as a digit with two states, the *number of digits* doesn't have anything to do with the counting system - that's just a physical limitation.) It's too complicated to teach to children without a whole course around counting systems, which doesn't make sense at that stage of life. For adults in the mathematics or comp sci sphere though, it's easy enough to adopt and therefore we probably should do so on an individual basis.
Not sure what the video itself seeks to argue, as I'm not about to watch someone try to convince me of the merits of a counting system I already use, so don't take any of this as an argument toward the creator's points. I'm simply arguing for what I believe to be best in my own experience.
@@impishlyit9780
1. Eh, not necessarily. One of the video's weaker arguments has to do with number length. It basically makes a *font* based argument, where if you change binary numbers to tall (1) and short (0) lines, the horizontal width of numerals becomes comparable to other bases. And that really *does* run into that issue. Or more broadly, I forgot to include this bit for contrast, but imagine a binary display. If any one segment goes out, you cease to be able to distinguish anything in that place. At a minimum, you'd need to do something like pairs of dots, where only one can be on, for you to be able to lose a segment and still be able to distinguish numbers. Meanwhile, even though only the top left and bottom right segments lack minimal pairs, you can still generally make things out on a 7-segment display, even if one of the segments goes out. You'd need 2-3 to go out before it starts really impacting your ability to read things.
2. No counterargument here.
3. It's also more than just kids. For example, how often do you *actually* need to count on your hands? It's typically smaller things, like how you might silently count down from 5. And while I'll grant that people probably count down specifically from 5, in part because we have 5 fingers on each hand, unary feels a lot easier for that. In a way, it's sort of like sorting algorithms. Merge sort is more effective for really big lists, but it also requires a lot of overhead. So for shorter lists, the less asymptotically efficient insertion sort winds up running faster overall. The main time I can think of that it's potentially useful to display bigger numbers would be something like days, but there are also systems like the Medieval Arabic hand numbering system that do it more easily than finger binary. It's actually even more efficient than finger binary. It encodes one digit on your thumb and index finger and one digit on your other three fingers, so it can go up to 9999 on two hands, as opposed to 1024.
But overall, my main criticism of the video really is that it successfully argues that binary is the least redundant numbering system possible, but not that redundancy is something we need to be avoiding
You should into your TV's settings and turn on either "overscan" or "just scan", or set the aspect ratio to "full" or "just scan". Turning on "game mode" might also fix this.
I really appreciated jan misali's introduction to the advantages of seximal, and have considered myself a convert for many years. You present a strong case here, and it's certainly convenient that your choice happens to be binary. One of the two systems that I have to know anyway. I'd like to see separate (short) videos on some of the topics here. How to write binary numbers. How to speak binary numbers. How to divide binary numbers. Etc.
Personally as programer I chose binary with hexadecimal compression because it is the best of both worlds. Your number to long, just start compressing into hexadecimal, need to do math decompress it. It just allows better writing efficacy and in low level programming you just do base two operations on hex nums anyways.
I am also a programmer, I love the idea of having two different ways of writing numbers, I previously only thought about hexadecimal and even made my own writing system for it wich also uses a sub-base of 4, 2*2 = 4, and 4*4=16, I love the symmetry
Binary for math operations and hexadecimal for showing numbers!
59:00 4 bits is sometimes called a "nibble" (in addition to being the amount of information in a hexadecimal digit, which is why you called it "hex").
missed opportunity to title it "The best way two count"
smart!
The music argument in the early chapter is interesting. Our notation is specified in fractions of powers of two - but they are fractions of four beats per measure. We also work just as often in three (triplets). However, we count in groups of 4 or 6, and often 8 (dance) and 12 (triplets over 4 beat measure). When dealing with irregular numbers like 5 or 7, the rhythm is usually done in syncopated emphasis of groups of three and two. For example, 5 is usuall 3+2 or doubled at 3+3+2+2; 7 is usually 3+2+2. Syncopated division of 8 is common with 3+3+2 (typical of latin and jazz) and 12 with 3+3+2+2+2 in flamenco compas. You further get the complication of one instrument playing a phrase in triplet over the rest of the band playing in duple.
The final pattern is that phrases tend to also be groups of four measures, a prime example being the three sections in a twelve bar blues. This is super internalized, and my brain stops and pays attention if a phrase isnt that multiple.
4-bar phrases and 16-bar sections are common enough that when I'm counting measures of rest or something, it usually feels most natural to count in a base 4 system.
On each hand I use my 4 non-thumb fingers and with 2 hands I can count up to 16.
For example:
1 = ...|
2 = ..||
3 = .|||
4 = |||| (1-4 had implicit |... right)
5 = ...| left, ||.. right
6 = ..|| left, ||.. right
7 = .||| left, ||.. right
8 = |||| left, ||.. right
9 = ...| left, |||. right
etc.
On both hands I start from my index finger, but with each one visualized palm-facing down, that results in the left hand counting right-to-left and the right hand counting left-to-right, but the point is that it feels right in the music.
The best base is base 10
Get it? Not base "ten", but base "one zero" where the base of that number is whatever you want. For me that's base "two" too.
don't explain the joke
explain the joke
@@mrosskne 10 is 2 in binary
@@mrosskne 10 is 6 in base 6
@@mrosskne10 is 10 in base 10
My first reaction was "wait, binary??", especially as i watched the jan Misali's videos before, but then it turned into "oooh, thats how we can do it", and then "it's beautiful as heck"
This really is exceptional content. As a nerd I really appreciate the work that was put into this video. Great work. My only note is the pacing was too fast for me to grasp some of the details that were claimed to be “immediately obvious”.
that's what the footnotes are for! this video would've been astronomically long otherwise :)
@@thebestwaytocount Says the uploader of an over I hour long video responding to an I ..I. minute long one
@@theramendutchman Dude an hour and a half is very short, stop watching tiktok
@@matheuscabral9618 First, there's no need to make assumptions and attack people. Please refrain from it and maintain a civil discussion.
Secondly, my point is that they said their video would've been longer, which is unexpected to me seeing how it responds to a 10 minute video; this video is already so much longer than the video it responds to!
@@theramendutchman eh
My jaw dropped when I saw that square root algorithm. Fuckin’ black magic 👍
My favorite numerical base is centovigesimal, or base-120, because it is really good for divisibility tests and representation of fractions... not only because 120 is very composite, but its neighbors are also composite (119 = 7 • 17 and 121 = 11²).
I have no idea why youtube didn't recommend this to me sooner this is the rabbithole I live for
Having been in computer science, I and all the computers in the world agree with you, because it can also be used to represent pretty much any concept that can be stored or represented digitally ... until it comes time for a human to interpret what's being said, because it's horribly inefficient! This is why we compromised and use 4-bit and 8-bit, which transfer nicely to hexadecimal, and lead to more efficient ways to represent extremely large numbers, which has become necessary in the age of the gigabyte, terabyte, petabyte, and beyond.
Honestly the only reason why I never considered binary in this nerd debate of "best counting system" was the number length. With tally-mark-like digits this already makes so much more sense, and that's just chapter . in the video.
we even have ascii characters for binary numbers: |.|. looks basically like the marks.
regardless of what you think remember, every base is base 10 :3
What a high quality video! I wonder if I’m really slow or most could keep up without extensive pausing and replaying. Congratulations on such a well made video regardless.
Not just you, lots of parts seemed a little confusing
1:08 "Perhaps that too is the case because 2 is the smallest integer greater than 1."
Yes, I think so. The note an octave above is the first frequency in the harmonic series of a pitch other than the pitch itself, and so an octave can be thought of as the smallest maximum consonance interval. It is interesting to consider systems that use other integer ratios in similar ways to how we use octaves though.
Non-binary music scales are neat to explore but they never seem to really click for me. Any system with a base larger than 2 has to avoid including a 2-octave between any of its notes, otherwise it kinda overpowers the 'real' base of the scale.
I have to do some maths and check how unary does in all metrics mentioned in the video. Surely, it won't score much on fractions, but it might get some absurdly good results wherever logarithms appear.
I think many of your ideas have basis. But i dont think that it works quite as well in practice. I am a fan of base 6, because of jan Misali's video, but as a programmer and computer scientist, i have to acknowledge the advantage of binary for raw computation strength. There is a good reason we use it in computers, and computation is a place where numbers shine. But unfortunately for us humans, we are not easily wired for base 2. It's hard to say how the real world efficiency of human calculations would turn out if we completely switched to binary, but i think that our brains are slightly better wired to understand larger bases. As someone who has poked around with basic arithmetic in both binary and seximal, i can say that binary is easier algorithmically, but if i had to do a lot of math, i would prefer base 6. Part of this is the practical problem of we need to write down the numbers we are doing math with and i feel like bigger numbers are better for cramps. I think it would be interesting to do legitimate experiments teaching people variou bases, and see who in the long run uses the bases most effectively. Seximal, for me, just seems to be a nice balance between small bases, which are easier for math, and big bases, which are easier for brains. Curious to see if jan Misali responds, but i still think i am on the seximal side for the best base for humans.
i ask him in his new super mario game video and no hes not gonna respond
Base 4 + 3i is the optimal base.
i don't know aabout that, but base -1+i is buded the most beutiful base.
I personally prefer base 2pi-i*sqrt(97) because I hate rational numbers (especially integers), but this is an interesting choice as well
I LOVE how you play the corresponding musical interval whenever you mention a fraction
I suspect that, were this to be actually used, it might be easier to use hex (or maybe octal) and convert to binary (which is really easy) every time you need to calculate something. Hex is easier to remember numbers in (try remembering 00101101 vs just 2D), to write down (6 strokes and 5 pencil moves for |.||.| vs 4 strokes and 1 move for 2D) and to input on a keyboard (two keys pressed for 2D vs 6 for the binary).
As someone who played with numerals a lot as a teen, and who's done a lot of programming, I find it VERY telling that we programmers who work with it often will ALWAYS display it as hexadecimal.
The 'best base' is really about what is the best base for humans, and minimizing how it pushes against our mental limits.
I really liked your new binary-based symbols for hexadecimal numbers. I wonder if a society based on binary math and those symbols would work better and learn math easier.
I think the binary naming in this video got really awkward and long to speak. Instead, use the groupings and name those. I think a society would need to use octal or hexadecimal for communication, while doing arithmetic in binary.
I'm entirely convinced about the mathematic side. But I think that goes without saying as the arguments were quite conclusive.
As a programmer myself, during my education (specifically when learning assembly), I always found working with the binary numbers far more intuitive / less cumbersome than working with the hex numbers. Like, at all points, I would think about it in the binary expansion, but then every time I had to actually write the number in the program I would have to group the binary by 4s and convert to hex, its just a mess.
This is not to say hex is bad per se, but that it is necessarily just a layer of translation/abstraction over the actual numbers, which are binary. "x5555" is obscure, needs translation. "101 101 101 101" is the real number, the thing you actually need to work with. Anyways..
Well let’s see. The majority of programming is done staring at a text rendered with a monospace font. Why of course it would be quite a nightmate to juggle literals 32+ characters wide. There’s no real compressibility unless the programming language allows macros or something to use a custom number literal format (or the IDE used allows to work with proportional fonts in comfortably).
I suspect there can be more mundane reasons other than this one.
@@05degrees Yeah, suppose we start with a system like the proposed. I think it wouldn't take much for people to start comming up with symbols and meanings that more readly represents the common quantities and groupings.
Felt compelled to decipher the text at 8:20.
It reads:
"You have got to be about the most superficial
commentator on con-langues since the idiotic
B. Gilson.
Did I miss the one where you said which conlang
you're fluent in and read at least three times a
week and can read new books in every week of
even one year or listen to radio shows in every
week? New radio shows?"
Was a fun challenge, thanks!
I love how the sounds are so entangled with the meanings, impressive.
Very interesting. These are some convincing points for doing arithmetic in binary.
I do think the proposed naming scheme is the biggest weakness however. As the process to convert between the spoken and written forms is relatively involved, it loses a lot of the written form's strengths. You can't very naturally think of the spoken form of a number in terms of whatever sized groups of bits are most convenient (e.g. size 3 groups are tricky); and the extremely simple operations of doubling and halving can completely change how a number is said.
For finger counting it's very nice to just use however many finger i'm comfortable freely manipulating when counting something for myself, but it becomes an issue with communication. If someone shows me some various combination of fingers raised on each hand, how do I know whether each hand represents 3, 4, or 5 bits?
Edit: I have an alternate proposal for naming numbers.
You dismissed the idea of simply reading off the digits; however, that could actually work. If you pick short syllables and ensure the consonants between them can flow smoothly (e.g. no mandatory glottal stops), then you can read off bits quickly. For example, if you pronounce 1 like "wun" or "nun", and 0 as "oh", "no", or "wo" depending on the previous bit, then you can read a number like 1010 0011 like "wunowuno owowunun". (yes this means you'd say "owo". deal with it.)
This way, you can group together digits in whatever sized chunks you want, like with the written form; as well as have easy bitshift operations; and between gaps you can *optionally* insert the "magnitude words" like hex,byte,etc. (like in base 10, the magnitude words are sometimes optional - you can say "three fifty" rather than "three hundred and fifty"). And you could start recognising the pronounciation of short bit sequences as words themselves.
I take some umbrage with the “efficient finger counting” argument, because finger counting is primarily a tool for teaching children. Binary finger counting (and even seximal finger counting) is more complex than simply counting the number of fingers raised. Bases are an arbitrary construct and humans don’t think in terms of them, so a child will never intuitively understand binary finger counting the way they understand “base 10” finger counting (though i would argue it’s more accurately described as base 1 finger counting)
Many cultures have counted in dozenal or hexadecimal on one hand. Many Asian cultures to this day count to ten on one hand and find the western way to count on fingers jarring. Your argument is eurocentric.
I’d argue the mentioned systems are still ultimately base-1 counting systems. They use 12 or 16 naturally ordered positions of the hand. Counting up to 12 with the dozenal thumb-thing is still just moving your thumb through a “number line” of hand positions. It still retains the simple linearity of western counting that’s lost with binary counting.
well actually maybe base infinity would be more accurate but my point is that neither 10 finger counting nor 12/16 one-hand counting have to actually deal with the things that make bases complicated (i.e. multiplying by powers of the base according to the position of the digit)
@@considerthehumbleworm During the Middle Ages doing arithmetics with your fingers was common enough that almost every book on mathematics had a chapter on it. For example in _'Liber Abaci'_ by _Leonardo Fibonacci_ finger counting is the first chapter. Systems to count up to 100.000 were around for quite some time.
Conflating finger counting to what _we_ today in the west teach to toddlers is a mistake. Drawing conclusions from oneself about others rarely works out.
Finger counting is what we teach to toddlers tho. I think it’s a bad idea to offer binary finger counting as a replacement in those contexts. It’s useful to be able to count higher than only 10 in other contexts but it only muddies the water for learning basic arithmetic. Honestly, teaching toddlers systems that go up to 8 or 16 is probably preferable to 10 to ease the transition from “number line counting” to using actual positional binary (even binary finger counting, eh?). That transition’s a bit of a given with 10 finger counting and base 10, so it’s worth considering how that works with binary. Also we don’t have to hate each other because we disagree about numbers lol
Unary is quite interesting
Chap 0) Numbers can be written as a series of Dots that follow some Line. This is obviously hard to read and long if the line stays straight, but if we have the Path the Dots follow not be Straight then we can Group them. We could just Grid the Dots with some set Side Length to group them. I also kinda like the idea of Coiling the path then Dots are all Coiled and can group parts by when the Spiral passes some spot, let's say the rightmost spot. Since as we move outward, if the Distance between points on the line remains the same then the Value Difference of each Layer of the Coil Increases as we move Outward, which is kinda cool. We could also just change the distances between them as we move Outward such that the Layers all represent the same Amount.
Chap 1) Log base 1 of x doesn't work so this breaks here. But, if each digit can only be in one state then all digits contain no information (in the same sense that the first Binary digit contains no information) hence Unary is super efficient?
Chap 2) Counting the Joints and Fingertips I can count to 38 on my hands, using a finger from the other hand to just point at the joint. I like to list it from my Right Thumb Tip to right Thumb Joints moving closer to the hand, then next finger etc. but it really doesn't matter. The only confusion I can think of is when the Index Fingertip is Counted, it might be Ambiguous as to Which Finger is being Counted and which is Pointing, and ig you could have one be on top or something. Other than that it's pretty easy to read and avoids the Recognisabilty issue. If you can't use some fingers, that's fine if ur joints are still there cuz u just point at them, and if ur missing fingers then u just count based on how many joints you have.
Chap 3) Unary Arithmetic is Mindlessly simple.
Addition: Just stack them on End (This might be more annoying with the spiral method to immediately visually see, but pretty easy with the Grid method).
Multiplication: Just stack it onto itself the number of times ur multiplying it by.
Division: Count thru the Dividend and reset every time you count to the Divisor and then ur left with the Quotient with some Remainder.
Square Root: Just re-stack them into a square. U can do this procedurally by starting at the corner and expanding from there until u run out so u don't have to know the sqrt beforehand.
Chap 4) No divisibility tests work as far as I can tell.
Chap 5) I don't think you can do fractional expansions in Unary.
Chap 6) BABABABABABABABABABABABABABABABABABABABABABABABA.... where each BA is a single number.
Chap 7) Unary is obviously the Most Fundamental. Objects just Exist, there isn't really the Non-Existence of things to be Counted. Look to Monism for a Philosophical Exploration of One
This is why we should all revert back to Unary.
🗿
The problem with that is that numbers start to get uncomfortably lengthy at an exponential rate. The number 41 in decimal would be 11111111111111111111111111111111111111111 and good luck trying to compress that
let's just go to base infinity, every number has its own unique symbol
or something like that
i guess
or maybe instead of base 2 let's go with base 1 so every number you just have to count out like tally marks
base infinity truly is the chaotic neutral of bases
base -1
Base -1-i
Arqam ftw
59:18 I feel like using kibi, mebi, gibi, etc. for powers of 10n would be good for familiarity's sake, since it's common to approximate 2^10n to 10^3n, and the words already exist.
"Single hand is chiral". My favorite sentence that says so much while just saying that hand is a hand :)
(RE: Section ||.) I find it quite difficult to actually understand what is going on with these "three four hex four three two one"- stuff. As others already have pointed out in the comments that these names for numbers are just too long and complicated to parse intuitively, I came across the fantastic hint that the human brain is capable and moreover the natural language does evolve to number names that exceed the number base (base-ten). E.g. eleven and twelve in English, which are not "oneteen" or "twoteen". For this, it has been proposed to just use a hexadecimal naming convention for this way to count. I sat down and get more thoughts on that till I came up with something comparable to that, what natural languages like English or German had done. A second thing I implemented is a distinction between numbers in higher or lowers orders/register, i.e., we use another name for some partial numbers if they are going big. Or for the multiplier numbers. (we'll come to that later, you'll see what I mean) The first number that has been significantly changed through my system is the number after twelve, which in decimal is call'd 13 (thirteen), but in binary shall be "five-eight" (Note that in binary the 1101 consists of 1000 "eight" and 101 "five"). I now present to you the further titling of the numbers in binary:
0: zero
1: one
10: two
11: three
100: four
101: five
110: six
111: seven
1000: eight
1001: nine
1010: ten
1011: eleven 1100: twelve
1101: fiveeight "five-eight"
1110: sixeight
1111: seveneight
1 0000: hex
1 0001: hex one...
1 1111: hex seveneight
10 ----: bihex... (pseudo-latine-graeco prefixes for distinction)
11 ---- : trihex...
100 ----: tetr(a)hex...
101 ----: penthex...
110 ----: seshex (probably the tongue twister one lol)
111 ----: septhex
1000 ----: octhex
1001 ----: nonhex
1010 ----: dechex
1011 ----: elhex (shortened eleven)
1100 ----: twelv'hex
1101 ----: penteighthex (from fiveeight)
1110 ----: seseighthex
1111 ----: septeighthex (from seveneight)
1 ---- ----: a byte.... and so on
As you please, you can replace the interjacent hexes with nybbles or (or do both) use plural 's'. Final test: The number 1111 1111 1111 1111 looks like this:
septeightnybbles-seveneight bytes septeighthex-seveneight
Now you would argue, this is LOOONG. But the same problem applies to the decimal base with numbers like 999,999, which is:
Nine hundred ninety-nine thousand Nine hundred ninety-nine.
What do you think?
I feel like this system could be improved by using unique labels in spoken language for every permutation of 4 bits, effectively treating it like hexadecimal when talking. You already concede that spoken binary would create some very long number labels and offer quartery as a compromise, so you might aswell embrace the pseudo-hexadecimal nature of grouping 4 bits together by adding a horizontal line at the bottom by covering these groupings with a single label each.
Exactly! And since lower numbers don't have decimal based names (compare "eleven" to "twenty-one") we can start by borrowing those. In spanish we have unique names up to 15 (it's "quince" as opposed to "diecicinco") so we already have the firt 4 digits covered. From that point, I'd say the next 4 digits should have a prefix (so, I.II. would be something like hexasix), but we would need to find something that rolls of the tongue (I love stack for 64, but it sounds clunky as a prefix)
Numberphile convinced me a decade ago that dozenal was the best base. Then jan Misali comes in and shows how seximal was better. After being fascinated about the topic and thinking about it on and off for years, I thought I knew almost all there was to know about number bases, and was convinced for so long that factors of the base are the biggest determining factor for what makes a base good. I never thought anyone else could show me otherwise.
This video showed me just how wrong I was, and that there was a surprising amount of unexplored depth to the topic of number bases. It's videos like these that I love the most, and I wish for this video to explode in popularity. Maybe binary is the best base after all.
Babe wake up found new banger channel