9:21 "Something about how I'm wired makes it so that when something is glaringly different with no explanation, it sticks out like a sore thumb and it bugs me. I've gotta know the how and the why." This perfectly encapsulates why 80% of my brain is dedicated to trivia nobody but me will ever care about. You put the smallest mystery in front of me and it will dominate my life for DAYS. In fact, the moment Dave mentioned the digits being KSNJFL I was about to start looking it up, but then remembered I was on a Dave video and I can just wait lol
"We do not use valuable scientific computing instruments to do clerical work!" ...let's not tell von Neumann what we use our pocket supercomputers for now.
Just that one computer second today can do so very much more than quite a number of computer hours then. And one of our computer seconds has a totally silly low cost. It costs soooo much less to pay 1 second of electricity, than to pay the food needed to have the developer (or a significant amount of helping clerks) perform the translations manually. I don't think von Neumann would be upset of our assemblers and compilers after having seen the actual - current - costs to have the machines perform the tasks. However, he might suffer some serious stroke if he knew the amount of CPU time and RAM consumed by having or smartphones run Java applications where the technical calculator may gobble 50-100 MB RAM to start. There is such a huge difference in CPU and RAM consumption between many "modern" applications and what the computation needs were to do the same thing on a MS-DOS machine or maybe an Amiga or C64 or ZX Spectrum. The shift in tools and strategies to write applications today may save developer time, but also requires quite a number of extra nuclear (and coal) power plants.
I remember my professor of computer engineering telling us in class "why do we say decimal where 'dec' is of Latin origin and 'hex' which is Greek origin and not 'sex'? Why it's simple, programmers don't have sex!"
Fun Fact: A guy that used to work for IBM told me it was rumored that HAL was supposed to be IBM, but the company would not give them permission to use the name (like Pan Am had for the spaceship). Notice that H is the letter before I, A is before B, and L is before M. Years later I saw where someone got a Laserdisc of 2001 and it had high enough resolution to see "IBM" on a wrist read out on one of the suits. Apparently the prop was made before they got the word that they did not have permission.
There's also an IBM logo on the cockpit, if I recall. Also, the "H" before "I", "A" before "B", etc. was called out in the novel _2010: Odyssey Two_ when Dr. Chandra was asked if he named the computer "HAL" to be one step ahead of IBM. Dr. Chandra angrily replies that no, "HAL" stands for "Heuristic ALgorithm." Dr. Floyd quips that he could actually hear the capital letters in Dr. Chandra's reply.
@@JamesQMurphyWOW! I had forgotten about that ... HAL was also a 10,000 in the book. The book wasn't even finished before they started working on the movie ... but, Mr. A.C. Clarke was there to assist, and fill in the gaps.
This was fabulous! I never knew that! The story I learned while in college during the 70's about hexadecimal vs sexadecimal had an IBM slant. At the time, IBM was the largest manufacturer of computers in the world. I was a novitiate computer student learning to program big iron. My instructor taught us multi base numbering schemes and mentioned that the actual term for base 16 should actually be SEXadecimal. However IBM was a very conservative company and would *never* market anything with S-E-X in it. So they invented the term HEXadecimal meaning that it was 6 + 10 = 16 and thus avoided having to market a sex machine. However, since they used the term HEX, they inadvertently caused all programmers to curse their programs. I know I did. Many, many, times. I've always been fascinated with computer history, having lived much of it myself, and I thank you for the deep dive into this very interesting topic!
Yes hexadecimal mixes greek and latin, while sexadecimal was all latin. And since this mixed up term hexadecimal seems to have originated somewhere 1952, it's quite likely it was the scary word "sex" that motivated mating (pun obviously intended) greek and latin to create a child-safe alternative.
@@perwestermark8920 That reminds me of the joke about the linguistics professor that is appalled when a student mentioned that they were poly-amorous. "That's disgusting! Mixing Latin and Greek like that. You are poly-eros!"
Wow, you're getting on a bit, love (you only have 6 years on me)... Yes, I learned Pascal on a CDC 7600. Much later, I looked into it and: yep! Transistors....
Some years ago I worked in the Operations Center at UofI. Every Christmas we would put up a tree adorned with tech from the past. Rainbow ribbon cable, coax, serial to parallel adaptors, terminators, media converters, a small (maybe 1k) of core memory and at the top of the tree was a single vacuum tube from Illiac.
I don't have one of those, but I do have a tube from EDSAC 2 (1958), Cambridge University's second digital computer, which I was given in 1968. Both EDSACs were designed by Prof Maurice Wilkes, head of the Cambridge University Maths Department, and based on the von Neumann architecture. EDSAC 1 ran its first program in May 1949, a few months before von Neumann's own EDVAC, and was the second stored program digital computer to be operational. The Manchester University "Baby" was the first - partly because some of the Manchester team had worked on the Colossus project at Bletchley Park. That was still top secret at the time so they could not talk about it, but what they had learned from building it was still in their heads. Back then, the UK was a leader in digital computing. Still is in some areas - your phone is almost certainly running on the ARM architecture, designed in Cambridge.
Just to put those tubes in perspective, I heard tell once of a professor at the University of Toronto who had on his desk a metal block holding three vacuum tubes. When asked what it was, he would answer "a bit".
Right to left, uses hammers: 3:2 6:S 7:311:4 14:F 15:519:623:7 26:J 27:8 28:M 30:K 31:9 34:L 35-0 40-1 Choice was based on some teletext machines where numbers and letters are shared on a single hammer via a shift register, and those which have dedicated keys. Between the two designs, the only character choices without overlap with numbers are: A, B, C, D, F, G, H, I, J, K, L, M, N, S, V, W, X, Y Out of those, some remaining characters have visual similarity, so were rejected to improve readability. BD, CG, AVWYX That leaves the following characters as options: F, H, I, J, K, L, M, N, S Of those, 0, 1, are separated from the other numbers on the left, K, L just happened to be in the same area and follow the same pattern allowing for most and least significant bit groupings. The remaining are displayed alphabetically right to left based on their hammer positions, just like 2 through 0.
like that idea, I can only relate to the semi-modern ASCII, but if you forget about the Numbers for moment , the ASCII alphabet ascii values in binary form "a" is 97 or 01100001 , "A" is 65 or 01000001 of if you where really get the most out or memory, when memory that small you be using the smallest amount of words as possible so just adding or subtracting 32 from the ascii value would change the selected letter from the lower case to the uppercase, and in binary it only moving one bit, jump from a letter from a to z, and would almost save binary bits from A to Z , to only adding one extra bit to the lower case letter or uppercase letter, so that could be extended the hole of the selectable characters, including numbers 01234567 as say lower case and then 89abcdef as say the upper case you would still only be using the same extra special bit, as the alphabet uses, saving even more memory, into the bargain also, that one extra bit could also be almost hard wired in to the machine like the to act like a modern keyboard [Caps-lock] shift key, you could also save on real buttons (KEY's) on the keyboard also, if it had the buttons doubling up as say 0/8 , 1/9 , 2/9 , 3/a , 4/b , 5/c , 6/d , 7/f with shift bottom to jump from the lower to the upper as required but if didn't use a QWERTY keyboard, maybe the sexdecimal version was just what lined up when the shift button (key) was pressed, as so stuck because it made the computer system use less memory which is always good, without having to repeat thinks, also early type writer where not to keen on the number 0 (zero) so you'd have to type a letter O instead that would just make the computer as made as HAL9000, so maybe the letter Key of O is wired to the Number 0 to keep things compatible :-)
@@dh2032 It definitely did use a Qwerty keyboard, 1 to 0 are the same as Q to P (the letters directly under the numbers on a modern keyboard). It's only the rest of the keyboard that's scrambled.
Well Dave, As an altruistic gentleman, who does this for the likes & subscribes, I truly appreciate you. Being a subscriber, I don't like every video as, I don't like having one watch videos on my like list, but, the ones I want to re-reference in the future. I loved this video. Thank you. Period.
It relates to the required modifications to the teletype. The manual "Modification of Teletype equipment for use with Illiac" describes the reason. Figure 1B shows that KSN and JFL line up with the index fingers of both hands. As detailed in paragraph II "key levers are rearranged ... to provide convenient single hand operation for punching sexadecimal digits..."
Awesome! Figure 1B in that manual tells the story - the numbers are in the center of the keyboard (shifted) with K,S,N immediately to the left of the numbers and J,F,L on the right.
This really seems to be an arrangement of keys after the ILLIAC encoding was invented. In particular, it's dated 1954, at least 2 years after the fact.
Well, your reasons seem better than my guesses. I was guessing that maybe Nixie tubes were involved and the characters were chosen because of things like drafting print standards where you don’t use 0 and O (zero or “O”) on the revised prints to avoid confusion.
I know maybe 11% of what my laptop can do. I have never seen the HAL 9000 movie you used clips from. I'm a retired US Navy Chief Petty Officer, worked as a EOD tech. I qualified as a Master Training Specialist. I really did enjoy this video, your prestation was good and I learned something new. Useless but new.
My mother worked in one of the offices at University of Illinois in the 50's and typed papers for many of the students, my oldest sisters husband graduated from U of I and worked at TRW. When he was talking to another person at work they mentioned that he had also attended U of I and he mentioned the little lady in the office that made their papers look good, it turned out that mom had also typed his papers. She was always proud to mention that she had been around Illiac, although I don't think she was directly involved. Thanks for sharing your insights with us, I always learn something from your content.
Love the channel, Dave. Good to hear from someone who was behind the scenes for so much of PC history, and who has a vey clear way of explaining code. Too often we just see programmers as unit that turn coffee into code.
It's great to see someone who cares about all of these mundane areas that are forgotten by most people. The history is much easier to understand when you know the roots they grew from. Keep up the good work.
3 года назад+100
I’m a simple man, if I see a Dave’s video I click.
Great Video Dave and quite a journey! I must say that having come up through machine-code (not even having access to an assembler) it was always a joy to figure out how pack both the program, data and scratchpad in as little as 256 bytes! You would sit there and have a program that wouldn't fit, then all of a sudden you'd get this crazy idea (that worked) on how to represent data and save a bunch of memory!. I built my first computer out of raw-TTL gates and used 7489's as my memory; one-bank for instructions and one-bank for data. I was so freaked-out that I could both read instructions, execute them and (as conditions warranted) change the data in the 7489's memory bank.
Very interesting and well-researched history, Dave. My first computer was the Librascope LGP-30, which had a Flexowriter (with paper tape) for I/O. The first shot in your video of the Flexowriter has the "LGP-30" nameplate on it. Their codes for the hex digits over 9 were FGJKQW as I remember, probably for the same reason as the Ordvac codes. I wrote a one-pass assembler for the LGP-30, and it was greeted with a resounding chorus of yawns.
Eat your heart out LTT this is the real deal here. Dave, thank you for your relentless curiosity. I've learned much from you over the past year, and I sincerely look forward to learning more. 🍻 Cheers!
@@concernedcitizen6313 Because one source is a true expert with deep technical knowledge. And the other source is a clumsy, squeaky, greedy entertainer waving his wallet around.
Thank you for reminding me why I could never work at Microsoft (or any computer company.) I was a successful business programmer in finance. I learned what I needed to know to make OS360 obey me. But it’s a pleasure to watch you and how your brilliant mind works. It’s hypnotic.
I'm not sure how this ended up in my RUclips suggestions TODAY, but I got a bit exited seeing it & even more watching it. You are VERY close on the reason these characters were chosen, but I'll pass on the unknown bit (pun intended) why these were chosen. As a precocious kid, I happened to become friends thru a customer of my Dad whom did home remodeling & repair, namely a Mr. H. Isabell, who was friends with Albert Einstein. I asked Mr. Isabell about the pattern for keyboards (mind you I was 8), and we went thru the debate of slowing down typists, and we got into mechanics. The characters of Sexadecimal (Mr. Isabell got so much amusement of me turning red every time he used the word...) the unit that they developed it on had been repaired on so those keys were chosen because in the repaired unit those keys were 'redone' or remapped and were the only ones reliable for any time due to some solenoid over heating.
My father was a senior electronics engineer when I was a child in the 1950's. One of his responsibilities was the Telex and punch paper tape network to enable NZ wide communications for Civil Aviation radars, ATC, weather and Nav Aids etc.I seem to recall that he could run the punched tape thru his fingers and read the info from it. He had been a ham radio developer since his childhood and was so fast at sending and understanding Morse. Pity he is long gone as his knowledge would have been of great interest today.
I can, in full honesty say, that I have not for many years since I started getting sucked into the RUclips universe, found a channel so interesting and so almost scaringly aligned with my own interests. I truly love watching your videos, and I can also say that the only times I haven't seen a video all the way to the end it is because I was interrupted. My father was a programmer at IBM in the 1960's (he passed 6 years ago now, but I never forget when he took me to technical museums in Europe that had various analogue representations of logical circuits, illustrating with handles, rolling balls and magnets combined (and other skillfully engineered gadgets), how the various logical circuits work illustrating in a very visual way to a 9 year old how the AND, NAND, OR, NOR, XOR, and other variations of logic arithmetic operators works. (I'm sorry if my phraseology is off, but English is not my first language). A moment I never forget is how my father - who was also a very emotional man who was moved to tears by classical music, the beauty of art, etc. - came across one of the first IBM mainframes he had been developing software routines for i one museum in Munich, Germany. He simply just stopped and looked at this (to him) marvel, that had been a milestone in his life from the days there were no formal education of computer engineers, software engineers, programmers, etc. so he was self-taught (sort of, as he had a university degree in mathematics, and therefore had a natural understanding of different based number-system, such as the binary, octal, hexadecimal, etc. as these systems was in use long before the first digital computers). He simply looked at this giant IBM Model 7000 and a S/360 next to each other and tears just started to run down his cheeks. This sound so strange to others - even to me when writing it - but I do have a tiny sense you know where he was coming from. He even told me - with great pride - that he had been allowed to program a very special variant of the S/360 called model 44 (that had a special scientific purpose, i memory serves), and it should have been the only one ever shipped to Denmark of that particular model. (My family are from Denmark, and my Dad worked for IBM from years before I was born in 1971 and until 1977). The last project he was involved in was programming for the Model 1 series. My father was a very skilled Assembler and Fortran programmer, and later on while I studied programming languages and their history, I fully understood how valuable it had been to have the basic understanding of the arithmetics of a computer when your aim is to make efficient programs. I also understood how he so easily learned new programming languages as they appeared on the scene, such as Algol, C, C++, COBOL, etc. Around that time my father became a professor at the university and remained there till he retired, but he never forgot his years with IBM. He still had a suitcase of memorabilia from these days when he died, and when I opened it I took a long time studying every little object in there from his small silver lapel pins, gold lapel pins (for senior programmers), and his IBM issued pocket calendar from 1969 with their motto on the from in gold letters on the black leather simply forming the word "THINK". I love that motto, and I still believe it is one of the most brilliant an catchy mottos of all I've seen. It is so simple and to the core, while not being pushy (like "Work harder" or something like that). Just - THINK. I love it. I also found the hilarious employees handbook, which is something completely different than an employees handbook today. This went to lengthy details to describe the kind of fabric that was allowed in the suits all male employees were expected to wear, the color of shoes allowed, that jacket and trousers were not allowed to be of different fabric or color, and how ties should always be in dark and discreet tones matching the suit and shoes. How belt and shoes should have the same colour of leather and how shirts were only allowed to be white, off-white or light blue. HOw male jewellery was limited to an engagement ring, wedding band and - if applicable - a family signet ring. This had never been accepted by anyone today. Back then it was the norm - at least if you worked at IBM. Those were also the days of the height of the hippie movement, and there must have been a clash between the moore liberal thinkers amongst the staff and the more conservative types who probably didn't all have the same degree of acceptance of such rigid demands. This comment became much longer than I had expected, but I simply just want to thank you for all these wonderful and enlightening anecdotes from the early days of the computer age, and - not least - for unknowingly reminding me of my father who also taught me programming from a very early age. (He started by teaching me PASCAL as that is a very forgiving language when it comes to forgetting to define your types, constants and variables as well as freeing up memory, which - if I remember correctly - isn't even a topic covered in the early PASCAL definitions. I had already experimented with BASIC on my own when I was 7 or 8, but my dad frowned upon such an unstructured programming language that would teach me all sorts of bad habits with its ability to jump "illegally" out of loops or the bad handling of sub routines, so he demanded that I at least honed my programming skills on a programming language that was somewhat structured and operated with function and procedure calls with the optional passing of arguments, so I was more prepared for the "real" programming languages such as C and C++ later on. Today I myself am a retired software engineer, and I even had a few years in the beginning of my career working for IBM, programming a system that was used as a cashiers terminal that could copy parts of display memory and present it to the costumer on a 40x2 character display on the checkout counter (as is now the lowest of standards in every little corner shop, but back then it was a system ordered by a nationwide chain of book stores), while every item was scanned with a handheld barcode reader. It was really nitty-gritty programming making sure that the characters that should be presented to the customer always ended up on the same two lines on the 80x25 screen that the cashier had pointing toward him/her, until the entire bill came out on paper on a 40 char 9-pin printer connected to the terminal through a Centronix interface. It feels like a lifetime ago but this is actually only a little less that 30 years ago I sat there in my office at IBM and programmed this in C++.
I read all this! IBM had a very strict dress code as its corporate identity. In the '60s there were a number of competing computer manufacturers. By 1980 IBM had utterly destroyed its competition to the point where when they entered the PC market almost all other PC manufacturers were restricted to making their PCs "IBM compatible". That dress code had been a key part of their strategy.
@@howardchambers9679 honestly, the ability to comprehend even a few paragraphs has obviously become too much strain on the brain for way too many from today's generation. Until I retired from my full time job as an IT team lead, I still subscribed to two major newspapers daily - which I actually found the time to read. Today, most people don't have the cognitive span to comprehend a couple of paragraphs it would take literally 2-3 minutes to read. I am in no way bashing on people suffering from dyslexia (In fact, my most rewarding job experience was working with dyslexic students and their teachers, developing solutions for people with reading challenges and children in school needing speech to text - and text to speech - recognition tools to help them keep up with their classmates), but people without such challenges have simply forgotten the value of reading a book, an article, or any other written material taking an in-depth approach to any kind of topic. Being used to getting everything served to the masses in headlines, "shorts", "reels", Tweets, etc. is simply going to backlash on society with a generation that isn't capable of reading even a basic article, or comprehend simple written instructions. What I wrote in my previous comment, would take an average reader 2-3 minutes to read, despite my horrible English skills. If that is more time than a person can spare, then it's a clear warning sign that person leads a very unhealthy lifestyle that is likely to - at best - end in a stress related illness. What is even more sad is that that the same persons who hasn't time to read what is within the limit of the commented length set by RUclips still has the time to tell the rest of the world that, not only them, but the entire world population has too little time to read two minutes worth of reading. Worst of all, though, is the fact that people writing comments like that are so little in touch with the general state of the world around them that they make claims such as "NO ONE has time" which was immediately contradicted by at least 10 people who read the comment and gave it a like. So it is not only a comment showing a person who is out of touch with the surrounding society, but also an outright fallacy. I also can't help wonder how people who cannot spend two minutes reading a comment ever got through University, let alone high school. I know that I am a ancient museum artefact who studied decades ago, but I do see many young men and women coming to the University each morning near the place I live and they still carry a heavy backpack full of books with them to all lectures, study groups, etc. At the same time they make up the demographic most active on social media, so it IS possible for some people to find TWO minutes to read a comment. Furthermore, it is beyond bold to talk on behalf of the entire world, claiming that "NO ONE" has the two minutes it take to read SO LITTLE, just because your own deficiencies keeps you from staying focused for the two minutes it takes to read a few lines (less than 40, I guess), like my previous comment. It is truly remarkably telling for a generation where most have no problem binge-watching series or watching a 2-3 hour long movie, but reading a book for a couple of hours seems totally unrealistic to the same group of people, when - in reality - one shouldn't be harder than the other. "Honestly", it is time to get AFK ... 🤷♂️🙂🙏
@@kalburgy2114 I think you make a very valid point. I'm also certain that even the dress code was a very conscious strategic choice. IBM even had their own Song Book (Calles "the IBM song-book", used at special occations and gatherings. It was almost cult-like in some areas, and today such corporate culture sounds almost eerie or spooky to most people. Thank you for your feedback.👍
I took a look at that Illiac manual, also available on Bitsavers and the Internet Archive. A page later than the table you showed us, the full tape code is shown. It's structured a lot like ITA 2. They didn't use Q and P because in the "letters" position, Q corresponds to the "figures" code for 1, and P corresponds to zero, just like on an ITA 2 keyboard. K, N, S, J, F, and L print as those six letters in either figures or letters mode; they figured they needed ten fewer special characters than ITA 2 had. This doesn't solve the mystery, but it's a tiny step forwards from what you covered in the video.
I didn't see a reference to the Frieden Flexowriter in that book. The illustrations are clearly a Model 19 Teletype. But what is really interesting is this comment on page 9-9 under "Keyboard Perforator Unit": "The sixteen sexadecimal keys that will be use most are in the center of the keyboard and arranged to be operated with one hand." The letters KSNJFL (not to mention the numbers) are *not* in the center of a QWERTY keyword, and stock Model 15/19 teletypes had QWERTY keyboards. Did they re-arrange the keys on the keyboard, too?
I am guessing that knsjfl are more or less on the "home row" for touch typists. Although n is on a lower row, it is the exception but easy to do with the right forefinger. On a standard qwerty keyboard. I suspect it is for easy typing.
I am by no means a programmer, so I must admit that some of what you talk about is entirely lost on me. Barring that, I absolutely love your work and documenting things that I hadn't even heard of. Please, keep up the great work and thank you for doing such!
I feel this is an opportunity, that if you like, you can ask questions and we can try to clarify. Then again if it's about the depths of C++, we'd probably need to write a book, and a better book has probably been already written.
I wasn't able to pay attention for the first few minutes, because I was thrown off by the lack of a "High, I'm Dave" salute. But once I recovered and restarted, this was indeed one of the most fascinating episodes so far, in that it was all completely new to me.
Urbana, IL resident here. I work on the UIUC campus. It's neat to see the whole "the web browser was invented here!" and other Internet-related firsts plaques and such all over the place, but there's very little mention of ILLIAC and the like. Thanks for sharing a bit of the history that's not well-known and almost never talked about around here.
I was a civilian employee of the US Army at one time, and I used the BRLESC computers some during their final few years of operation (mid-late 1970s). The mnemonic for KSNJFL used at that time was King-Sized Numbers Just For Laughs. However, by the time I worked there most of the pioneers had retired and no one (that I talked with) seemed to know the reason for that choice of letters. Now I know or at least have heard a good, plausible theory.
Even before you suggested it, my guess about "why these characters" immediately went to "I bet it has something to do with a mechanical limitation of an output device." I like your guess.
Colossus was the first digital computer built by Alan Turing at Bletchley Park in WW2. It doesn't get the fame it deserves because its existence was kept secret after the end of WW2 for many years. We didn't want the world to know that we had cracked the Enigma encoder.
It was good to hear Dave mention it. Whether he is able to find enough to study is a different matter. Maybe if he joined GCHQ but then he would not be allowed to talk.
Looking at the Flexowriter hardware manual there are some interesting things related to how the hardware encodes a letter. While it has 8-bits, it uses them in a very an interesting way. First off, it has odd parity. This means that (number from 8 to 1) bit 5 is usually used as a the parity bit. The numbers 1-9 are encoded as binary 1-9, with bit 5 set as needed. The only exception is 0, which has only bit 6 set (as space was encoded as only bit 5). What's really interesting is that the lower 4 bits never go above 9, meaning everything is in groups of ten with gaps between them. So there are no specific keys that would generate the right bottom 4 bits. In addition everything is in alphabetical order. If anything it seems like they were picked because the bottom 6 bits (yes, including the parity) wouldn't map to a real number so there would be no way to accidentally interpret them as 0-9.
On that theory, no numbers have bit 6 or 7 set. Letters A to I and S to Z all have bit 6 set, so no chance of confusion using only bits 1 to 6. Letters J to R all have bit 7 set, meaning they'll always have opposite parity from the numbers, so again not chance of confusion because parity fails. There's no reason to pick any particular letters if accidental interpretation is driving your decisions. Also, the ILLIAC encoding seems to have nothing to do with the Flexowriter's internal encoding. The only overlap between them are the numbers, both binary encoded.
@@christophertstone yes it seems to me that made since from my old days using various paper tape encoding systems. all holes punched and no holes punched = null 5 level teletype used a shift sequence for letters and figures and that hand similar checks built into the character set
At first I wasn't sure your channel was my cup of tea, I mostly am here “you tube” to learn more about my trade, I’m a Geek. Yes I recognize that you have several videos with that in mind but I find myself recently as interested in you teaching the history of geek world as much or more entertaining than the pure tech learning. Thank You.
I had a good friend who was an Honeywell engineer in the 40s, 50s, and 60s . He had a VERY large collection of truly old computers, they included a low serial pdp, a 4 bit bendix tube type, and a whole bunch of other stuff. Being an engineering student in the early 70s made this very interesting to me. He was the first who explains sexadecimal to me, and showed me how it was 'wired' into the bendix io circuits. As I recall the bendix had a rudmentary ecc circuit that compared computed results to expected benchmarks to detect failing tubes. This was directly related to the 4 bits nature of the io data, sadly, when this passed away, his family, who had no idea of the treasure trove he had, sold ALL of the "stuff" for scrap.
You see, the 4 bits cannot hold ascii... so its impossible to write something with it. So... its purely the numeric bit patterns... All that matters is that the computer reads the value as decimal 10,11,12,13,14,15... ending up with the matching binary in its circuits.. There is no difference between a sexadecimal circuit and a hexadecimal circuit.. The Ordvac was sexadecimal only because the type writer was left with the KJNSLF keys, the other keys cut away... this was due to the practicalities of modifying the keyboard !!!..
This sparked a memory- I used to do data mapping for a fairly large company that somewhat dictated how other departments needed to do their work. I produced a ton a literature on best practices and style guides sister companies informally use. I did a career pivot, but spent about a year training my replacement. Things go pear shaped several months after I'm 100% in my new role and I'm asked to step in to figure out why things aren't working. I started looking at how the system was set up and everything was basically a bizarro world version of my work. My replacement basically choose to reject all of the previous work just to do things differently, just for the sake of being different. As time has gone on, I come across more and more instances of people just wanting to be different, "revolutionize" things, etc. with no clear motivator as to why other than to just be different. Why are so many things on an Iphone just the opposite of an Android? The answer could just be as silly as this being the first instance of creating an arbitrary difference for no reason other than that.
A lot of things are different on Android because Apple PATENTED the way they do things on the iPhone. Kind of like Zilog (or Motorola?) Patenting the Op-Code alpha representations and Intel having to come up with their own version (or vice-versa, I don't remember who was first).
One thing I love about our professional field...almost everyone is only a couple degrees of separation from a famous person in the field. Though, a connection to Von Neumann is a cool one.
In most fields, the foundations were laid centuries or millennia ago. In computing, the equivalent people were working less than a human lifetime ago. It's honestly incredibly cool that we can still talk to many of them today.
That's the basic idea behind the "Six Degrees of Kevin Bacon". It doesn't only apply to actors. It's essentially a result of the "small world phenomenon" which can be seen almost everywhere when you look close enough.
You're right. Among other connections I've made, my former boss went on to work with Douglas Engelbart before he passed away. I was in the audience of like 40 people when Roy Fielding introduced the original REST API concept at ApacheCon 2000. I also talked to Ken Thompson through email once about the origins of the 'rm' command.
This is so true. I used to play in a weekly home poker game with a woman whose supervisor in the Navy was Grace Hopper. One of the professors in my son's department in college is Brian Kernigan. It is surreal to me either of these is even possible.
I gathered that in the first few seconds, Dave. 00:13 Fascinating history, I liked the code table. 15:53 And your speculative conclusion does make sense, working around the limitations of a mechanical device.
I was just thinking about 2010 during this. I haven't really cracked open any of Clarke's works in a long time, but I have 2001, 2010, 2061 and 3001 in my collection. I should read them through again this winter.
Fantastic deep dive, I was taught hexadecimal for the UYK20 and 642 B computers in the NTDS system in the navy. Was on the Eisenhower in the early 80s. Thanks for great content.
An amazing bit of sleuthing Dave, it sure drew me in. Actually, this is the first reference I have ever seen to "sexadecimal" and I've been around as long as some of the vacuum tube devices (not computers tho). My first introduction to computers was in studying solid-state electronics in the very early '60s.
I appreciate the historical information re: sexadecimal. My earliest days of computer programming was on CDC Cyber-series computers that used 60-bit words so the common encoding format was 20 octal (3-bit) digits. I guess the satement that the referenced computeres were "programmed" in sexadecimal is accurate in the sense that when you would type a program into a paper tape, you would enter numbers in sexadecimal format. However, sexadecimal was just a convient numeric encoding format that aligned with the paper tape readers of the day, not anything fundamental to the computer architecture. In other words just an 'input' encoding for numbers. But whether they used 10 digit sexadecimal, 10 digit hexadecimal, or 40 digit binary representations of words for input really had nothing to do with the computer architecture. It is only an artifact of the input/output designs for those computers.
I am a why and a how person too. In my field (music) almost nobody knows the why and how for piano and its history. Very frustrating experience. Thankfully lots of academic articles exist, just takes someone to actually read them.
It should be noted that according to the ILLIAC 1 character set (page 9-5 of the manual), when it is in FIGs shift, the sequence printed is actually 0123456789+-NJFL because the K and the S were not in the FIGs shift, but only in the LTRs shift. For logical reasons, and unlike the common practice for teleprinters, they duplicated the letters NJFL into the FIGs shift. But for who knows what reasons, they chose to set the + and - signs into the positions of the symbols for 10 and 11, which should correspond to the K and the S letters. They could have also placed the K and the S in there, but they would have had to sacrifice 2 other symbols, so they obviously were pressed into compromise. Interestingly they also had a lowercase x fitted into the FIGs shift.
I seriously enjoy these vids Dave, finding out about something so old nowadays is really interesting and actually, we do share that one trait, if there's something that needs an answer as to why, I get an itch too.
Correction: ENIAC was definitely not the first general purpose programmable computer. This honour belongs to Konrad Zuse and his Z3, completed in 1941. Zuse also invented the first general purpose programming language and the world's first chess program.
There is a case to be made for honoring the brilliant work of Konrad Zuse especially for innovation in the area of the Plankalkül structured programming language, but it is quite a stretch to label this a correction. The reason being that the Z series were prototype designs incrementally building out Zuse’s ideas, making the Z4 comparable to ENIAC in many ways. However the scarcity of the materials Konrad Zuse desired for his research required reinvention and work-around after work-around delaying progress and recognition by the competing research and development projects under Luftwäffe funding. Combined with allied bombing, the invasion of Germany not only slowed progress but prevented the Z-series from contributing substantially to the war effort of the Third Reich. This in no way diminishes the brilliance of Zuse, but comparatively ENIAC did make contributions for assisting the US Army with artillery targeting tables and creation of targeting tools that were noteworthy in redefining how artillery targeting was performed at the end of the war and beyond into the Korean War and the 1950s. Additionally, the Kuse hardware and research had to be spirited away to avoid the Red Army onslaught, resulting in work being lost for a time and little known outside of those that worked with Zuse in Germany. Although Z4 components, research and executable code punched on recycled celluloid film were discovered post war in Bavaria, Jude’s work was not restored to operational condition til the ‘50s. Again, solely for comparison, the ENIAC was put into effective general purpose computing by the end of ‘45 and the US Army kept ENIAC running in continuous operation from ‘47 to ‘55 at the Aberdeen Proving Grounds. Like Zuse’s Z series, the ENIAC was funded by wartime research, and they both focused primarily on calculations. Zuse solved many challenges numeric processing challenges in innovative ways, but even if the devastation of the war had not stifled his progress, the Z series would have had a similar obsolescence as ENIAC, Colossus, et al, with the general purpose computing and miniaturization leaps brought about by Von Neumann architecture that was researched concurrently with the creation of these prototypes that were each milestones in their own right. There were several examples of these early computers, and each was brilliant in their own ways. Sadly Konrad Zuse and his innovations are not well known outside of Germany and like Von Braun’s rocketry advances they were tainted by development under the Hitler regime. Much like the development of the first jet engine for use in aircraft, their are multiple players in the first general purpose computer that get obscured by the war which enshrouded their development. My attempt here was not to correct you in regard to Zuse and the Z series, but rather indicate that it’s not a cut and dry answer to “which is the first general purpose computer?”, eh? Given Dave’s remarks in this case were a side note (as he could do a video on this subject) that would detract from the central topic of this video. For my part, I would tend to say both Z3/Z4 from Zuse and the ENIAC could be considered the first from a certain point of view. I am more interested in the contribution of Kuse for starting with binary out of the gate and creating the workable structured language Plankalkül that could be translated into machine code (though this vision was not actually realized in the case of the Z series) and having the processor completely controlled by the code compared to rivals like Colossus and ENIAC. I believe that Zuse would have been more globally acknowledged in Computer Science of his early Z series prototypes and design documents had survived the war. Just as Turing (and his peers) were in acknowledged due to military secrecy for decades, Zuse is also largely under-recognized for his work due to being overseen by Hitler’s Germany. If it had come about prior to ‘33 or after ‘45 - Konrad Zuse’s Z series Computer Science contributions would have greater exposure historically.
@@CaptainBakerJasonit's very useful to call out the problem with Eniac For a long time it was touted as the first computer. Americans were happy with that title for a long time, but interacting with the literally dozens of not Americans on the web has made that title harder to hold. Each time another computer is pointed to, there's always as excuse. "We meant digital. We meant programmable..." Now Wikipedia comes with all these adjectives: "programmable, electronic, general-purpose digital" "general purpose" doesn't point to another page. It doesn't really mean anything
@@geoffjones5421arguably this isn't entirely correct either, as Charles Babbage (whom Ada Lovelace worked under) had a machine that could do mathematical calculations, albeit entirely mechanical but still programmable and digital which was even earlier, and that's not counting the (theorised to be) ancient planetary computer that is believed to track the star positions and is thousands of years old. American's are right to be proud of their achievements, they are many, varied, and transformative of the world we live in today for better or worse, but their work is built on the shoulders of giants, many of whom they deliberately ignore in their attempt to prove their exceptionalism. :(
@@SeanOfEarth Much of the video does sound like a scripted narratve, but it is still some interesing information even if "literal first" does not properly apply. Fact is, the literal first computer might have been created down in a root cellar somewhere but never got publicity before a landslide buried it and its creator. The only way to know the "first" with certainty is to know every action ever taken. And even the supercomputers of today cannot reach that level - and never will. Yes, even computers cannot become God :-0)
I absolutely love this channel and your mindset regarding it. I don't know the path I took to discover this channel, but I am glad I did. Keep up the excellent work Dave 😊
This is great stuff! We're still early-enough in the history of computers to have access to the people who were there; it's great that you're preserving this history for future generations!
I knew I should've paid more attention when dealing with 70s mainframes, and later, PC-DOS. Really enjoy your videos, and how you share your knowledge. Maybe it will make me pull all the punch cards saved from decades past. Great job.
Video idea: program a floating-point binary number class in C++ that is fully compatible with the standard double type so that we can understand how it works!
I tried doing that (in a different language but that's irrelevant) and very quickly realised the daunting task of printing and parsing a floating-point. There have been papers written about it and dedicated libraries for doing this fast and accurate (than glibc's parsing and printing function).
It is easy if you just convert your number to the language's primitive floating-points and let them handle the printing and parsing, but it's not as fun then.
From my first week of computer science I remember Zuse Z3 as the first programmable Computer. It was built 3 years earlier than ENIAC, could process binary numbers and was (theoretical, with some tricks) touring complete.
I heard somewhere that in some versions of the tape computers, the letter codes for tape were chosen so that the most common letters used would require the least amount of holes, thus increasing the lifespan of the machine that needs to punch them out and read the tape. Not sure if this is true though.
In Morse code, the most common letters used, are represented with the shortest codes to improve efficiency and to increase the ‘lifespan’ of the telegraphers wrist. The close relation between telegraph, teletext, and tape computers might explain why the same rationale was projected onto the tape and the tape machines.
Boy, this took me back. I haven't thought much about the history of computers and hexadecimal since my first computer science classes in 1982. Yes, Dave, I'm older than you are. I'm very much enjoying your videos. My wife comes into my office, asks me what I'm what I'm watching, and when I try to explain it she simply says, "oh, nerd stuff." Then she walks out again. I don't bother reminding her that the "nerd stuff" is what pays my salary.
I see hexadecimal an sexadecimal as interchangable, and see the different letters as just different things used before it was standardised, just like different names were. Some decided to go u,v,w,x,y,z. Some people suggested entirely separate character sets, including different characters for 0-9. And you don't need computers at all. People were thinking about different bases (including base 16) before we had computers.
The ICL mainframe I started programming on had a 6-bit byte and hence a 24-bit word, with the digits 0-9 having binary values of 0-9. The OpSys, George 3+, remains one of the best I have ever used.
My first job after leaving the RAF in 1962 was at GEC Electronics in Coventry, as a final test engineer on the ICT 1301 manufacture. That machine used four bits plus parity in twelve digit words. I later came across bytes, half words, words and double words when I joined IBM and worked on System 360/370 and 3030 series.
At 9:54 you mention "big, old analog transistors"--but actually, all transistors are analog. If you look at the various "digital" waveforms in a modern computer, you won't find a square edge anywhere. When transistors have to switch on and off at gigahertz frequencies, the time to switch on and off is a significant fraction of the clock period.
Analog/digital is a characteristic of a circuit not of an electronic component. A transistor, being an electronic component, is not analog nor digital. But it can be used to build both analog and digital circuits.
Plus for FETs we still often call the linear region, where varying the gate voltage has a linear response like a resistor, the triode region since the behavior is like a triode vacuum tube.
As were valves and long before that the beads on an abacus. Even the original _digital_ counting (ie using fingers) involves ignoring the finer details of the exact position of the finger: pinkie is either up or down, and halfway between is a firmware error in the finger driver. On the abacus a bead has to be at one end of the free space or the other. Circuits for digital counting likewise have to be designed to force the valve or transistor to flip quickly through the intermediate states between one end of its range and the other. Only when we get to quantum physics do we get physical systems that are actually confined to give discrete results (binary in the case of a spin-half particle, more digits for some other quantised effects). The joke is that the analogue behaviour of a transistor is actually the aggregate of a lot of quantum subsystems: as are valves, beads, and fingers.
An analog transistor will have a smooth characteristic across the range of voltages applied to the gate. A digital one is optimized to switch as fast as possible between extrema, what happens in the forbidden band in between be damned.It's a difference in technology. You could just as well say "solar cells are just LEDs" - and while technically true, it's misguiding, because they thoroughly suck at emitting light when powered, and are optimized for collecting it and producing electric current.
herd of a company slang, calling their darlington transistors (amplification chains) just digital transistors as their processor GPIOs would often drive them directly. FETs are also more sort of digital ones due to their very useable fully-open/closed states.
I remember in school poking around in cabinets of equipment for ILLIAC maintenance team in room L510 of The Digital Computer Lab building, or in the storage room under the building extension. There was so much early computer history just piled up most just waiting to eventually go in the dumpster. In the days of processors on desktops exceeding 200MHz and 128megs of ram, everyone viewed all those historical systems garbage. Today one could have built a world renowned museum from those rooms alone.
Something you didn't mention as a possibility. Do the letters relate to related people's initials somehow, KSN and JFL? Or KS, NJ, FL Or wait, maybe they are states?
I don't think so. We are on longer solving problem "why those 6 characters for sexadecimal", we are solving problem "why are 30+ characters on that teletype ordered in seamingly random order". You will not order that many characters in this particular way just because six of them have some meaning for you.
@@petrkubena Why wouldn't you. It would help in remembering the order. Kansas, New Jersey, and Florida are easier to remember than the random letter of KSNJFL. And, when explaining to someone on how to memorize this set, they could refer to how the correlate with the states.
Yes! Like in the movie Brazil... The ghost in the machine: EREIAMJH, an anagram of JEREMIAH. "It's as though he's there still speaking to me - 'ere I am, J.H.'"
@@jonathanreal8018 Yes, but those are all on one side of the keyboard. Spreading them out, would allow for faster typing. Now choosing which letters, and a easy way to memorize them would be a possible concern. But who knows. They could have chosen the letters using darts.
Hi Dave, great video. Always a pleasure to watch you. I ordered 2 of your coffee mugs. One for me and one for my son; who BTW is also named Dave. Thank you for your channel.
I always giggle like a madman when thinking of those tube monstrosities needing half a powerplant of their own, while holding a fricking Arduino in the hands that runs circles around those things :D
Very interesting subject. And I think you are the first person I've watched that's not in it for the money. Also, I too get bugged when something is different for no reason and I set out to find out why. I think I might like your channel. This is the first time I've come across it, so I will be checking out more of your videos. You have a great voice for talking and keeps it interesting.
Fact: I am very interested in any further research you do on this topic as my instructors never went any deeper than to mention the alternate sexadecimal scheme in passing. Theory: I believe that I am not the only person in your audience that shares this interest. Thank you very much for all the time and effort you put into this topic, like all of the other videos that I've seen from you this is top shelf!
Interesting. KSNJFL are all middle row letters except for N, using 2 on the left & 4 on the right side of the keyboard leaving ADGHK free. F & J are the locater keys so it makes sense to use them. Why the others were picked may as you say have been lost to time but I think you're on the right track with hammer jams & readability. Maybe even the punch tape was more reliably read with certain sequences.
My first thought was it was for outputting to segmented displays... with regular hexadecimal, it's almost impossible to differentiate an 8 from a B! However, N can't easily be represented. Very intriguing video!
nice idea, but you could use lower case b and lower case n, if display on a 7 segment was a concern. Don't think 7-segment displays were already invented at that time.
This is possible, but highly unlikely. Designing and implementing these systems was crazy difficult, arranging the keys by your initials would make it harder yet.
You certainly get my like. I'm also going to flag this to my brother in law - he worked on the Eniac, so this will be a walk down memory lane for him. - *_core_* memory!
You might enjoy the "Seximal" numbering scheme by jan misali then. There's a quite interesting argument that base-6 would be the best base for common use, instead of base 10 or 12
This is all way above my understanding, I may have to watch it again instead of just listening to it. I like finding things that are hard to understand because I also have the incessant curiosity.
9:21 "Something about how I'm wired makes it so that when something is glaringly different with no explanation, it sticks out like a sore thumb and it bugs me. I've gotta know the how and the why."
This perfectly encapsulates why 80% of my brain is dedicated to trivia nobody but me will ever care about. You put the smallest mystery in front of me and it will dominate my life for DAYS. In fact, the moment Dave mentioned the digits being KSNJFL I was about to start looking it up, but then remembered I was on a Dave video and I can just wait lol
Thank you, Dave, for having "the talk" with us that our parents did not.
You're the best!!
Yeah, I had to learn about sexidecimal in the back of an old Chevy van! ;)
LMAO, I had that talk with my parents. I was doing the talking! (though it was hex!)
"We do not use valuable scientific computing instruments to do clerical work!"
...let's not tell von Neumann what we use our pocket supercomputers for now.
We still do a lot of ballistics calcs, but they're mostly for flappy birds!
Just that one computer second today can do so very much more than quite a number of computer hours then. And one of our computer seconds has a totally silly low cost.
It costs soooo much less to pay 1 second of electricity, than to pay the food needed to have the developer (or a significant amount of helping clerks) perform the translations manually.
I don't think von Neumann would be upset of our assemblers and compilers after having seen the actual - current - costs to have the machines perform the tasks.
However, he might suffer some serious stroke if he knew the amount of CPU time and RAM consumed by having or smartphones run Java applications where the technical calculator may gobble 50-100 MB RAM to start.
There is such a huge difference in CPU and RAM consumption between many "modern" applications and what the computation needs were to do the same thing on a MS-DOS machine or maybe an Amiga or C64 or ZX Spectrum. The shift in tools and strategies to write applications today may save developer time, but also requires quite a number of extra nuclear (and coal) power plants.
Look up the LEO series of computers.. Always think they should be more famous
holy shit it's MrCheeze
@@DavesGarage If we're talking ballistics, I think Angry Birds is more fitting.
“Fact or theory please mark it“ this channel is a wonderful hold out from a kinder, gentler time on the Internet
I remember my professor of computer engineering telling us in class "why do we say decimal where 'dec' is of Latin origin and 'hex' which is Greek origin and not 'sex'? Why it's simple, programmers don't have sex!"
WTF? 'dec' is of greek origin as well as 'hex'. Did your professor know their ancient langages or were they making stuff up for comedic purposes?
@@Meta11axis well, yes and no.
Dec is from Latin. The Latin dec comes from the Greek dek/deka.
deni didi dici
Anglicizing words from dead languages into jargon is art not grammar. It becomes its definition.
@@rjonboy7608 whatever works
Fun Fact: A guy that used to work for IBM told me it was rumored that HAL was supposed to be IBM, but the company would not give them permission to use the name (like Pan Am had for the spaceship). Notice that H is the letter before I, A is before B, and L is before M. Years later I saw where someone got a Laserdisc of 2001 and it had high enough resolution to see "IBM" on a wrist read out on one of the suits. Apparently the prop was made before they got the word that they did not have permission.
The spaceship was Pan Am
There's also an IBM logo on the cockpit, if I recall.
Also, the "H" before "I", "A" before "B", etc. was called out in the novel _2010: Odyssey Two_ when Dr. Chandra was asked if he named the computer "HAL" to be one step ahead of IBM. Dr. Chandra angrily replies that no, "HAL" stands for "Heuristic ALgorithm." Dr. Floyd quips that he could actually hear the capital letters in Dr. Chandra's reply.
@@JamesQMurphyWOW!
I had forgotten about that ... HAL was also a 10,000 in the book.
The book wasn't even finished before they started working on the movie ... but, Mr. A.C. Clarke was there to assist, and fill in the gaps.
Word salad 🤣🥕
Arthur C. Clarke always denied this. He was adamant that it simply abbreviated "Heuristically programmed ALgorithmic Computer"
This was fabulous! I never knew that! The story I learned while in college during the 70's about hexadecimal vs sexadecimal had an IBM slant. At the time, IBM was the largest manufacturer of computers in the world. I was a novitiate computer student learning to program big iron. My instructor taught us multi base numbering schemes and mentioned that the actual term for base 16 should actually be SEXadecimal. However IBM was a very conservative company and would *never* market anything with S-E-X in it. So they invented the term HEXadecimal meaning that it was 6 + 10 = 16 and thus avoided having to market a sex machine. However, since they used the term HEX, they inadvertently caused all programmers to curse their programs. I know I did. Many, many, times.
I've always been fascinated with computer history, having lived much of it myself, and I thank you for the deep dive into this very interesting topic!
It's plausible to me! It seems like the earliest stuff is sexadecimal, then it switches to hex over time in the 50s...
Yes hexadecimal mixes greek and latin, while sexadecimal was all latin. And since this mixed up term hexadecimal seems to have originated somewhere 1952, it's quite likely it was the scary word "sex" that motivated mating (pun obviously intended) greek and latin to create a child-safe alternative.
@@perwestermark8920 That reminds me of the joke about the linguistics professor that is appalled when a student mentioned that they were poly-amorous.
"That's disgusting! Mixing Latin and Greek like that. You are poly-eros!"
Wow, you're getting on a bit, love (you only have 6 years on me)...
Yes, I learned Pascal on a CDC 7600.
Much later, I looked into it and: yep!
Transistors....
But the advantage of Hex over Sex was that it set up a nice pun for Terry Pratchett's Discworld computer which ran on magic, mice, and cheese...
Some years ago I worked in the Operations Center at UofI. Every Christmas we would put up a tree adorned with tech from the past. Rainbow ribbon cable, coax, serial to parallel adaptors, terminators, media converters, a small (maybe 1k) of core memory and at the top of the tree was a single vacuum tube from Illiac.
Was the tube illuminated?
@@microdesigns2000 Not that we did not think about doing that but that would have melted that little plastic tree.
I don't have one of those, but I do have a tube from EDSAC 2 (1958), Cambridge University's second digital computer, which I was given in 1968. Both EDSACs were designed by Prof Maurice Wilkes, head of the Cambridge University Maths Department, and based on the von Neumann architecture. EDSAC 1 ran its first program in May 1949, a few months before von Neumann's own EDVAC, and was the second stored program digital computer to be operational. The Manchester University "Baby" was the first - partly because some of the Manchester team had worked on the Colossus project at Bletchley Park. That was still top secret at the time so they could not talk about it, but what they had learned from building it was still in their heads.
Back then, the UK was a leader in digital computing. Still is in some areas - your phone is almost certainly running on the ARM architecture, designed in Cambridge.
Just to put those tubes in perspective, I heard tell once of a professor at the University of Toronto who had on his desk a metal block holding three vacuum tubes. When asked what it was, he would answer "a bit".
Right to left, uses hammers: 3:2 6:S 7:3 11:4 14:F 15:5 19:6 23:7 26:J 27:8 28:M 30:K 31:9 34:L 35-0 40-1
Choice was based on some teletext machines where numbers and letters are shared on a single hammer via a shift register, and those which have dedicated keys.
Between the two designs, the only character choices without overlap with numbers are: A, B, C, D, F, G, H, I, J, K, L, M, N, S, V, W, X, Y
Out of those, some remaining characters have visual similarity, so were rejected to improve readability.
BD, CG, AVWYX
That leaves the following characters as options:
F, H, I, J, K, L, M, N, S
Of those, 0, 1, are separated from the other numbers on the left, K, L just happened to be in the same area and follow the same pattern allowing for most and least significant bit groupings.
The remaining are displayed alphabetically right to left based on their hammer positions, just like 2 through 0.
I was wondering if it used the "type box format" similar to how the
Selectric type ball worked
Do you have any sources for any of this??
like that idea, I can only relate to the semi-modern ASCII, but if you forget about the Numbers for moment , the ASCII alphabet ascii values in binary form "a" is 97 or 01100001 , "A" is 65 or 01000001 of if you where really get the most out or memory, when memory that small you be using the smallest amount of words as possible so just adding or subtracting 32 from the ascii value would change the selected letter from the lower case to the uppercase, and in binary it only moving one bit, jump from a letter from a to z, and would almost save binary bits from A to Z , to only adding one extra bit to the lower case letter or uppercase letter, so that could be extended the hole of the selectable characters, including numbers 01234567 as say lower case and then 89abcdef as say the upper case you would still only be using the same extra special bit, as the alphabet uses, saving even more memory, into the bargain also, that one extra bit could also be almost hard wired in to the machine like the to act like a modern keyboard [Caps-lock] shift key, you could also save on real buttons (KEY's) on the keyboard also, if it had the buttons doubling up as say 0/8 , 1/9 , 2/9 , 3/a , 4/b , 5/c , 6/d , 7/f with shift bottom to jump from the lower to the upper as required
but if didn't use a QWERTY keyboard, maybe the sexdecimal version was just what lined up when the shift button (key) was pressed, as so stuck because it made the computer system use less memory which is always good, without having to repeat thinks, also early type writer where not to keen on the number 0 (zero) so you'd have to type a letter O instead that would just make the computer as made as HAL9000, so maybe the letter Key of O is wired to the Number 0 to keep things compatible :-)
That's a pretty interesting insight, thanks!
@@dh2032 It definitely did use a Qwerty keyboard, 1 to 0 are the same as Q to P (the letters directly under the numbers on a modern keyboard). It's only the rest of the keyboard that's scrambled.
Well Dave,
As an altruistic gentleman, who does this for the likes & subscribes,
I truly appreciate you.
Being a subscriber, I don't like every video as, I don't like having one watch videos on my like list, but, the ones I want to re-reference in the future.
I loved this video.
Thank you.
Period.
It relates to the required modifications to the teletype.
The manual "Modification of Teletype equipment for use with Illiac" describes the reason.
Figure 1B shows that KSN and JFL line up with the index fingers of both hands.
As detailed in paragraph II "key levers are rearranged ... to provide convenient single hand operation for punching sexadecimal digits..."
Which manual did you spot that in? I'm not home at the moment but couldn't spot it in the programming manual?
@@DavesGarage RUclips won't let me post a link, but it's called "Modification of Teletype Equipment for use with Illiac"
Awesome! Figure 1B in that manual tells the story - the numbers are in the center of the keyboard (shifted) with K,S,N immediately to the left of the numbers and J,F,L on the right.
This really seems to be an arrangement of keys after the ILLIAC encoding was invented.
In particular, it's dated 1954, at least 2 years after the fact.
Well, your reasons seem better than my guesses. I was guessing that maybe Nixie tubes were involved and the characters were chosen because of things like drafting print standards where you don’t use 0 and O (zero or “O”) on the revised prints to avoid confusion.
I know maybe 11% of what my laptop can do. I have never seen the HAL 9000 movie you used clips from. I'm a retired US Navy Chief Petty Officer, worked as a EOD tech. I qualified as a Master Training Specialist. I really did enjoy this video, your prestation was good and I learned something new. Useless but new.
Man, returning home to watch some new video straight out of Dave's Garage. Now that's life.
My mother worked in one of the offices at University of Illinois in the 50's and typed papers for many of the students, my oldest sisters husband graduated from U of I and worked at TRW. When he was talking to another person at work they mentioned that he had also attended U of I and he mentioned the little lady in the office that made their papers look good, it turned out that mom had also typed his papers.
She was always proud to mention that she had been around Illiac, although I don't think she was directly involved.
Thanks for sharing your insights with us, I always learn something from your content.
Love the channel, Dave. Good to hear from someone who was behind the scenes for so much of PC history, and who has a vey clear way of explaining code. Too often we just see programmers as unit that turn coffee into code.
It's great to see someone who cares about all of these mundane areas that are forgotten by most people. The history is much easier to understand when you know the roots they grew from. Keep up the good work.
I’m a simple man, if I see a Dave’s video I click.
He's got a system!
{
Int dave = 1 , vid = 1;
If
(
dave + vid = 2
)
then
print(click);
}
return 0;
Dave, I am totally lost when it comes to this stuff. I'm just glad you are on our side.
Great Video Dave and quite a journey! I must say that having come up through machine-code (not even having access to an assembler) it was always a joy to figure out how pack both the program, data and scratchpad in as little as 256 bytes! You would sit there and have a program that wouldn't fit, then all of a sudden you'd get this crazy idea (that worked) on how to represent data and save a bunch of memory!.
I built my first computer out of raw-TTL gates and used 7489's as my memory; one-bank for instructions and one-bank for data. I was so freaked-out that I could both read instructions, execute them and (as conditions warranted) change the data in the 7489's memory bank.
Hmm, that's pretty hard core....
And now I have 40TB of videos and no idea what to watch!
As someone who likes to go down rabbit holes myself, I enjoyed watching Dave go down one and listening to his thought process.
Very interesting and well-researched history, Dave. My first computer was the Librascope LGP-30, which had a Flexowriter (with paper tape) for I/O. The first shot in your video of the Flexowriter has the "LGP-30" nameplate on it. Their codes for the hex digits over 9 were FGJKQW as I remember, probably for the same reason as the Ordvac codes. I wrote a one-pass assembler for the LGP-30, and it was greeted with a resounding chorus of yawns.
HAL inspired me greatly. Thank you for this piece of historic value.
Eat your heart out LTT this is the real deal here. Dave, thank you for your relentless curiosity. I've learned much from you over the past year, and I sincerely look forward to learning more. 🍻 Cheers!
And eat your heart out Chris tutus tech!
I mean, why not enjoy and learn from both?
@@concernedcitizen6313 Because one source is a true expert with deep technical knowledge.
And the other source is a clumsy, squeaky, greedy entertainer waving his wallet around.
Thank you for reminding me why I could never work at Microsoft (or any computer company.) I was a successful business programmer in finance. I learned what I needed to know to make OS360 obey me. But it’s a pleasure to watch you and how your brilliant mind works. It’s hypnotic.
I'm not sure how this ended up in my RUclips suggestions TODAY, but I got a bit exited seeing it & even more watching it. You are VERY close on the reason these characters were chosen, but I'll pass on the unknown bit (pun intended) why these were chosen. As a precocious kid, I happened to become friends thru a customer of my Dad whom did home remodeling & repair, namely a Mr. H. Isabell, who was friends with Albert Einstein. I asked Mr. Isabell about the pattern for keyboards (mind you I was 8), and we went thru the debate of slowing down typists, and we got into mechanics. The characters of Sexadecimal (Mr. Isabell got so much amusement of me turning red every time he used the word...) the unit that they developed it on had been repaired on so those keys were chosen because in the repaired unit those keys were 'redone' or remapped and were the only ones reliable for any time due to some solenoid over heating.
My father was a senior electronics engineer when I was a child in the 1950's. One of his responsibilities was the Telex and punch paper tape network to enable NZ wide communications for Civil Aviation radars, ATC, weather and Nav Aids etc.I seem to recall that he could run the punched tape thru his fingers and read the info from it.
He had been a ham radio developer since his childhood and was so fast at sending and understanding Morse. Pity he is long gone as his knowledge would have been of great interest today.
What do you mean by NZ?
the inner immaturity in me wants to laugh at dave highlighting “sex” in the thumbnail but i’m not going to
I am going to
@@lennymclennington so am I
I already did
mhmmhmmhm
I used bright red, that ought to help!
I can, in full honesty say, that I have not for many years since I started getting sucked into the RUclips universe, found a channel so interesting and so almost scaringly aligned with my own interests. I truly love watching your videos, and I can also say that the only times I haven't seen a video all the way to the end it is because I was interrupted. My father was a programmer at IBM in the 1960's (he passed 6 years ago now, but I never forget when he took me to technical museums in Europe that had various analogue representations of logical circuits, illustrating with handles, rolling balls and magnets combined (and other skillfully engineered gadgets), how the various logical circuits work illustrating in a very visual way to a 9 year old how the AND, NAND, OR, NOR, XOR, and other variations of logic arithmetic operators works. (I'm sorry if my phraseology is off, but English is not my first language).
A moment I never forget is how my father - who was also a very emotional man who was moved to tears by classical music, the beauty of art, etc. - came across one of the first IBM mainframes he had been developing software routines for i one museum in Munich, Germany. He simply just stopped and looked at this (to him) marvel, that had been a milestone in his life from the days there were no formal education of computer engineers, software engineers, programmers, etc. so he was self-taught (sort of, as he had a university degree in mathematics, and therefore had a natural understanding of different based number-system, such as the binary, octal, hexadecimal, etc. as these systems was in use long before the first digital computers). He simply looked at this giant IBM Model 7000 and a S/360 next to each other and tears just started to run down his cheeks. This sound so strange to others - even to me when writing it - but I do have a tiny sense you know where he was coming from. He even told me - with great pride - that he had been allowed to program a very special variant of the S/360 called model 44 (that had a special scientific purpose, i memory serves), and it should have been the only one ever shipped to Denmark of that particular model. (My family are from Denmark, and my Dad worked for IBM from years before I was born in 1971 and until 1977). The last project he was involved in was programming for the Model 1 series. My father was a very skilled Assembler and Fortran programmer, and later on while I studied programming languages and their history, I fully understood how valuable it had been to have the basic understanding of the arithmetics of a computer when your aim is to make efficient programs. I also understood how he so easily learned new programming languages as they appeared on the scene, such as Algol, C, C++, COBOL, etc.
Around that time my father became a professor at the university and remained there till he retired, but he never forgot his years with IBM. He still had a suitcase of memorabilia from these days when he died, and when I opened it I took a long time studying every little object in there from his small silver lapel pins, gold lapel pins (for senior programmers), and his IBM issued pocket calendar from 1969 with their motto on the from in gold letters on the black leather simply forming the word "THINK". I love that motto, and I still believe it is one of the most brilliant an catchy mottos of all I've seen. It is so simple and to the core, while not being pushy (like "Work harder" or something like that). Just - THINK. I love it.
I also found the hilarious employees handbook, which is something completely different than an employees handbook today. This went to lengthy details to describe the kind of fabric that was allowed in the suits all male employees were expected to wear, the color of shoes allowed, that jacket and trousers were not allowed to be of different fabric or color, and how ties should always be in dark and discreet tones matching the suit and shoes. How belt and shoes should have the same colour of leather and how shirts were only allowed to be white, off-white or light blue. HOw male jewellery was limited to an engagement ring, wedding band and - if applicable - a family signet ring. This had never been accepted by anyone today. Back then it was the norm - at least if you worked at IBM. Those were also the days of the height of the hippie movement, and there must have been a clash between the moore liberal thinkers amongst the staff and the more conservative types who probably didn't all have the same degree of acceptance of such rigid demands.
This comment became much longer than I had expected, but I simply just want to thank you for all these wonderful and enlightening anecdotes from the early days of the computer age, and - not least - for unknowingly reminding me of my father who also taught me programming from a very early age. (He started by teaching me PASCAL as that is a very forgiving language when it comes to forgetting to define your types, constants and variables as well as freeing up memory, which - if I remember correctly - isn't even a topic covered in the early PASCAL definitions. I had already experimented with BASIC on my own when I was 7 or 8, but my dad frowned upon such an unstructured programming language that would teach me all sorts of bad habits with its ability to jump "illegally" out of loops or the bad handling of sub routines, so he demanded that I at least honed my programming skills on a programming language that was somewhat structured and operated with function and procedure calls with the optional passing of arguments, so I was more prepared for the "real" programming languages such as C and C++ later on.
Today I myself am a retired software engineer, and I even had a few years in the beginning of my career working for IBM, programming a system that was used as a cashiers terminal that could copy parts of display memory and present it to the costumer on a 40x2 character display on the checkout counter (as is now the lowest of standards in every little corner shop, but back then it was a system ordered by a nationwide chain of book stores), while every item was scanned with a handheld barcode reader. It was really nitty-gritty programming making sure that the characters that should be presented to the customer always ended up on the same two lines on the 80x25 screen that the cashier had pointing toward him/her, until the entire bill came out on paper on a 40 char 9-pin printer connected to the terminal through a Centronix interface. It feels like a lifetime ago but this is actually only a little less that 30 years ago I sat there in my office at IBM and programmed this in C++.
Honestly,no one has time to read that much!
I read all this! IBM had a very strict dress code as its corporate identity. In the '60s there were a number of competing computer manufacturers. By 1980 IBM had utterly destroyed its competition to the point where when they entered the PC market almost all other PC manufacturers were restricted to making their PCs "IBM compatible". That dress code had been a key part of their strategy.
@@howardchambers9679 sure they do! you just _chose_ not to
@@howardchambers9679 honestly, the ability to comprehend even a few paragraphs has obviously become too much strain on the brain for way too many from today's generation. Until I retired from my full time job as an IT team lead, I still subscribed to two major newspapers daily - which I actually found the time to read. Today, most people don't have the cognitive span to comprehend a couple of paragraphs it would take literally 2-3 minutes to read. I am in no way bashing on people suffering from dyslexia (In fact, my most rewarding job experience was working with dyslexic students and their teachers, developing solutions for people with reading challenges and children in school needing speech to text - and text to speech - recognition tools to help them keep up with their classmates), but people without such challenges have simply forgotten the value of reading a book, an article, or any other written material taking an in-depth approach to any kind of topic. Being used to getting everything served to the masses in headlines, "shorts", "reels", Tweets, etc. is simply going to backlash on society with a generation that isn't capable of reading even a basic article, or comprehend simple written instructions.
What I wrote in my previous comment, would take an average reader 2-3 minutes to read, despite my horrible English skills. If that is more time than a person can spare, then it's a clear warning sign that person leads a very unhealthy lifestyle that is likely to - at best - end in a stress related illness. What is even more sad is that that the same persons who hasn't time to read what is within the limit of the commented length set by RUclips still has the time to tell the rest of the world that, not only them, but the entire world population has too little time to read two minutes worth of reading. Worst of all, though, is the fact that people writing comments like that are so little in touch with the general state of the world around them that they make claims such as "NO ONE has time" which was immediately contradicted by at least 10 people who read the comment and gave it a like. So it is not only a comment showing a person who is out of touch with the surrounding society, but also an outright fallacy. I also can't help wonder how people who cannot spend two minutes reading a comment ever got through University, let alone high school. I know that I am a ancient museum artefact who studied decades ago, but I do see many young men and women coming to the University each morning near the place I live and they still carry a heavy backpack full of books with them to all lectures, study groups, etc. At the same time they make up the demographic most active on social media, so it IS possible for some people to find TWO minutes to read a comment. Furthermore, it is beyond bold to talk on behalf of the entire world, claiming that "NO ONE" has the two minutes it take to read SO LITTLE, just because your own deficiencies keeps you from staying focused for the two minutes it takes to read a few lines (less than 40, I guess), like my previous comment.
It is truly remarkably telling for a generation where most have no problem binge-watching series or watching a 2-3 hour long movie, but reading a book for a couple of hours seems totally unrealistic to the same group of people, when - in reality - one shouldn't be harder than the other.
"Honestly", it is time to get AFK ... 🤷♂️🙂🙏
@@kalburgy2114 I think you make a very valid point. I'm also certain that even the dress code was a very conscious strategic choice. IBM even had their own Song Book (Calles "the IBM song-book", used at special occations and gatherings. It was almost cult-like in some areas, and today such corporate culture sounds almost eerie or spooky to most people. Thank you for your feedback.👍
I took a look at that Illiac manual, also available on Bitsavers and the Internet Archive. A page later than the table you showed us, the full tape code is shown. It's structured a lot like ITA 2. They didn't use Q and P because in the "letters" position, Q corresponds to the "figures" code for 1, and P corresponds to zero, just like on an ITA 2 keyboard. K, N, S, J, F, and L print as those six letters in either figures or letters mode; they figured they needed ten fewer special characters than ITA 2 had. This doesn't solve the mystery, but it's a tiny step forwards from what you covered in the video.
I thumbed through the ORDVAC manual, and they labeled the rows on the output tape as A, B, C, and D. Significant, do you think?
I didn't see a reference to the Frieden Flexowriter in that book. The illustrations are clearly a Model 19 Teletype. But what is really interesting is this comment on page 9-9 under "Keyboard Perforator Unit": "The sixteen sexadecimal keys that will be use most are in the center of the keyboard and arranged to be operated with one hand." The letters KSNJFL (not to mention the numbers) are *not* in the center of a QWERTY keyword, and stock Model 15/19 teletypes had QWERTY keyboards. Did they re-arrange the keys on the keyboard, too?
I am guessing that knsjfl are more or less on the "home row" for touch typists. Although n is on a lower row, it is the exception but easy to do with the right forefinger. On a standard qwerty keyboard. I suspect it is for easy typing.
As a computer nerd that grew up from 8 bit to todays fullstack, I found this Extreamly interesting. Well worth the time! Thanks.
I love this. We need more archaic computer history!
Literally fascinating. I slapped myself on the back when I guessed the bit convenience based on old patterns, and was kind of right. Thank you.
I am by no means a programmer, so I must admit that some of what you talk about is entirely lost on me. Barring that, I absolutely love your work and documenting things that I hadn't even heard of. Please, keep up the great work and thank you for doing such!
I feel this is an opportunity, that if you like, you can ask questions and we can try to clarify. Then again if it's about the depths of C++, we'd probably need to write a book, and a better book has probably been already written.
I wasn't able to pay attention for the first few minutes, because I was thrown off by the lack of a "High, I'm Dave" salute. But once I recovered and restarted, this was indeed one of the most fascinating episodes so far, in that it was all completely new to me.
Urbana, IL resident here. I work on the UIUC campus. It's neat to see the whole "the web browser was invented here!" and other Internet-related firsts plaques and such all over the place, but there's very little mention of ILLIAC and the like. Thanks for sharing a bit of the history that's not well-known and almost never talked about around here.
I was a civilian employee of the US Army at one time, and I used the BRLESC computers some during their final few years of operation (mid-late 1970s). The mnemonic for KSNJFL used at that time was King-Sized Numbers Just For Laughs. However, by the time I worked there most of the pioneers had retired and no one (that I talked with) seemed to know the reason for that choice of letters. Now I know or at least have heard a good, plausible theory.
Even before you suggested it, my guess about "why these characters" immediately went to "I bet it has something to do with a mechanical limitation of an output device." I like your guess.
Colossus was the first digital computer built by Alan Turing at Bletchley Park in WW2. It doesn't get the fame it deserves because its existence was kept secret after the end of WW2 for many years. We didn't want the world to know that we had cracked the Enigma encoder.
It was good to hear Dave mention it. Whether he is able to find enough to study is a different matter.
Maybe if he joined GCHQ but then he would not be allowed to talk.
Looking at the Flexowriter hardware manual there are some interesting things related to how the hardware encodes a letter. While it has 8-bits, it uses them in a very an interesting way. First off, it has odd parity. This means that (number from 8 to 1) bit 5 is usually used as a the parity bit. The numbers 1-9 are encoded as binary 1-9, with bit 5 set as needed. The only exception is 0, which has only bit 6 set (as space was encoded as only bit 5).
What's really interesting is that the lower 4 bits never go above 9, meaning everything is in groups of ten with gaps between them. So there are no specific keys that would generate the right bottom 4 bits. In addition everything is in alphabetical order.
If anything it seems like they were picked because the bottom 6 bits (yes, including the parity) wouldn't map to a real number so there would be no way to accidentally interpret them as 0-9.
Ahhh, so it’s an error detection thing. Shaping the encoding scheme to guarantee any possible single-bit error maps to an illegal (in sex) character
On that theory, no numbers have bit 6 or 7 set.
Letters A to I and S to Z all have bit 6 set, so no chance of confusion using only bits 1 to 6.
Letters J to R all have bit 7 set, meaning they'll always have opposite parity from the numbers, so again not chance of confusion because parity fails.
There's no reason to pick any particular letters if accidental interpretation is driving your decisions.
Also, the ILLIAC encoding seems to have nothing to do with the Flexowriter's internal encoding. The only overlap between them are the numbers, both binary encoded.
@@christophertstone yes it seems to me that made since from my old days using various paper tape encoding systems.
all holes punched and no holes punched = null
5 level teletype used a shift sequence for letters and figures and that hand similar checks built into the character set
At first I wasn't sure your channel was my cup of tea, I mostly am here “you tube” to learn more about my trade, I’m a Geek. Yes I recognize that you have several videos with that in mind but I find myself recently as interested in you teaching the history of geek world as much or more entertaining than the pure tech learning. Thank You.
I had a good friend who was an Honeywell engineer in the 40s, 50s, and 60s . He had a VERY large collection of truly old computers, they included a low serial pdp, a 4 bit bendix tube type, and a whole bunch of other stuff. Being an engineering student in the early 70s made this very interesting to me. He was the first who explains sexadecimal to me, and showed me how it was 'wired' into the bendix io circuits. As I recall the bendix had a rudmentary ecc circuit that compared computed results to expected benchmarks to detect failing tubes. This was directly related to the 4 bits nature of the io data, sadly, when this passed away, his family, who had no idea of the treasure trove he had, sold ALL of the "stuff" for scrap.
Oh, that hurts....
You see, the 4 bits cannot hold ascii... so its impossible to write something with it. So... its purely the numeric bit patterns... All that matters is that the computer reads the value as decimal 10,11,12,13,14,15... ending up with the matching binary in its circuits.. There is no difference between a sexadecimal circuit and a hexadecimal circuit.. The Ordvac was sexadecimal only because the type writer was left with the KJNSLF keys, the other keys cut away... this was due to the practicalities of modifying the keyboard !!!..
I have a 2400 baud modem. That's a really old one. Would you like to buy it from me? 😄
Does the friend have any spare thermostats around?
@@BillGreenAZ My first modem was 300 baud. I was thrilled when I could afford to upgrade to 1200 baud.
This sparked a memory- I used to do data mapping for a fairly large company that somewhat dictated how other departments needed to do their work. I produced a ton a literature on best practices and style guides sister companies informally use. I did a career pivot, but spent about a year training my replacement. Things go pear shaped several months after I'm 100% in my new role and I'm asked to step in to figure out why things aren't working. I started looking at how the system was set up and everything was basically a bizarro world version of my work. My replacement basically choose to reject all of the previous work just to do things differently, just for the sake of being different. As time has gone on, I come across more and more instances of people just wanting to be different, "revolutionize" things, etc. with no clear motivator as to why other than to just be different. Why are so many things on an Iphone just the opposite of an Android? The answer could just be as silly as this being the first instance of creating an arbitrary difference for no reason other than that.
A lot of things are different on Android because Apple PATENTED the way they do things on the iPhone. Kind of like Zilog (or Motorola?) Patenting the Op-Code alpha representations and Intel having to come up with their own version (or vice-versa, I don't remember who was first).
Hi Dave! It's Hal speaking. Cool link here. Nibbles and bits. Nibbles and bits.
One thing I love about our professional field...almost everyone is only a couple degrees of separation from a famous person in the field. Though, a connection to Von Neumann is a cool one.
In most fields, the foundations were laid centuries or millennia ago. In computing, the equivalent people were working less than a human lifetime ago. It's honestly incredibly cool that we can still talk to many of them today.
That's the basic idea behind the "Six Degrees of Kevin Bacon". It doesn't only apply to actors. It's essentially a result of the "small world phenomenon" which can be seen almost everywhere when you look close enough.
You're right. Among other connections I've made, my former boss went on to work with Douglas Engelbart before he passed away. I was in the audience of like 40 people when Roy Fielding introduced the original REST API concept at ApacheCon 2000. I also talked to Ken Thompson through email once about the origins of the 'rm' command.
This is so true. I used to play in a weekly home poker game with a woman whose supervisor in the Navy was Grace Hopper. One of the professors in my son's department in college is Brian Kernigan. It is surreal to me either of these is even possible.
@@MrCheeze and yet, there is information that has been lost, as the reasons why these letters were chosen.
I gathered that in the first few seconds, Dave. 00:13 Fascinating history, I liked the code table. 15:53 And your speculative conclusion does make sense, working around the limitations of a mechanical device.
The bloopers 🤣 we love you Dave. Thank you for being here and sharing your stories.
Ps: next novel explains HAL's behavior very nicely
I was just thinking about 2010 during this. I haven't really cracked open any of Clarke's works in a long time, but I have 2001, 2010, 2061 and 3001 in my collection. I should read them through again this winter.
Fantastic deep dive, I was taught hexadecimal for the UYK20 and 642 B computers in the NTDS system in the navy. Was on the Eisenhower in the early 80s. Thanks for great content.
I honestly think that you make great videos Dave.
This is my new favorite channel. Thanks for the wide variety of topics. Also the outtakes were hilarious!
An amazing bit of sleuthing Dave, it sure drew me in. Actually, this is the first reference I have ever seen to "sexadecimal" and I've been around as long as some of the vacuum tube devices (not computers tho). My first introduction to computers was in studying solid-state electronics in the very early '60s.
Your depth of research is incredible! Love this!
I appreciate the historical information re: sexadecimal. My earliest days of computer programming was on CDC Cyber-series computers that used 60-bit words so the common encoding format was 20 octal (3-bit) digits. I guess the satement that the referenced computeres were "programmed" in sexadecimal is accurate in the sense that when you would type a program into a paper tape, you would enter numbers in sexadecimal format.
However, sexadecimal was just a convient numeric encoding format that aligned with the paper tape readers of the day, not anything fundamental to the computer architecture. In other words just an 'input' encoding for numbers. But whether they used 10 digit sexadecimal, 10 digit hexadecimal, or 40 digit binary representations of words for input really had nothing to do with the computer architecture. It is only an artifact of the input/output designs for those computers.
Loved the blooper reel. I always learn so much computing history from watching your channel.
I am a why and a how person too. In my field (music) almost nobody knows the why and how for piano and its history. Very frustrating experience. Thankfully lots of academic articles exist, just takes someone to actually read them.
I love doing my own research and evidently, so do you. Thanks for all the hard work.
Theory: "Hammer Jam" was the name of my 1980s electro party rock band in highschool.
Thank you. I'm glad I stopped by. ❤
It should be noted that according to the ILLIAC 1 character set (page 9-5 of the manual), when it is in FIGs shift, the sequence printed is actually 0123456789+-NJFL because the K and the S were not in the FIGs shift, but only in the LTRs shift. For logical reasons, and unlike the common practice for teleprinters, they duplicated the letters NJFL into the FIGs shift. But for who knows what reasons, they chose to set the + and - signs into the positions of the symbols for 10 and 11, which should correspond to the K and the S letters. They could have also placed the K and the S in there, but they would have had to sacrifice 2 other symbols, so they obviously were pressed into compromise. Interestingly they also had a lowercase x fitted into the FIGs shift.
Bravo! Even a cave-being such as myself understood most of what was said in this video. Yay! There's hope yet.
I seriously enjoy these vids Dave, finding out about something so old nowadays is really interesting
and actually, we do share that one trait, if there's something that needs an answer as to why, I get an itch too.
Dave! Outtakes were hysterical. Hang onto them and make a 10-15 min digital comedy special. Love the programming insights.
Correction: ENIAC was definitely not the first general purpose programmable computer. This honour belongs to Konrad Zuse and his Z3, completed in 1941. Zuse also invented the first general purpose programming language and the world's first chess program.
There is a case to be made for honoring the brilliant work of Konrad Zuse especially for innovation in the area of the Plankalkül structured programming language, but it is quite a stretch to label this a correction. The reason being that the Z series were prototype designs incrementally building out Zuse’s ideas, making the Z4 comparable to ENIAC in many ways. However the scarcity of the materials Konrad Zuse desired for his research required reinvention and work-around after work-around delaying progress and recognition by the competing research and development projects under Luftwäffe funding. Combined with allied bombing, the invasion of Germany not only slowed progress but prevented the Z-series from contributing substantially to the war effort of the Third Reich. This in no way diminishes the brilliance of Zuse, but comparatively ENIAC did make contributions for assisting the US Army with artillery targeting tables and creation of targeting tools that were noteworthy in redefining how artillery targeting was performed at the end of the war and beyond into the Korean War and the 1950s. Additionally, the Kuse hardware and research had to be spirited away to avoid the Red Army onslaught, resulting in work being lost for a time and little known outside of those that worked with Zuse in Germany. Although Z4 components, research and executable code punched on recycled celluloid film were discovered post war in Bavaria, Jude’s work was not restored to operational condition til the ‘50s. Again, solely for comparison, the ENIAC was put into effective general purpose computing by the end of ‘45 and the US Army kept ENIAC running in continuous operation from ‘47 to ‘55 at the Aberdeen Proving Grounds.
Like Zuse’s Z series, the ENIAC was funded by wartime research, and they both focused primarily on calculations. Zuse solved many challenges numeric processing challenges in innovative ways, but even if the devastation of the war had not stifled his progress, the Z series would have had a similar obsolescence as ENIAC, Colossus, et al, with the general purpose computing and miniaturization leaps brought about by Von Neumann architecture that was researched concurrently with the creation of these prototypes that were each milestones in their own right. There were several examples of these early computers, and each was brilliant in their own ways. Sadly Konrad Zuse and his innovations are not well known outside of Germany and like Von Braun’s rocketry advances they were tainted by development under the Hitler regime.
Much like the development of the first jet engine for use in aircraft, their are multiple players in the first general purpose computer that get obscured by the war which enshrouded their development. My attempt here was not to correct you in regard to Zuse and the Z series, but rather indicate that it’s not a cut and dry answer to “which is the first general purpose computer?”, eh? Given Dave’s remarks in this case were a side note (as he could do a video on this subject) that would detract from the central topic of this video. For my part, I would tend to say both Z3/Z4 from Zuse and the ENIAC could be considered the first from a certain point of view. I am more interested in the contribution of Kuse for starting with binary out of the gate and creating the workable structured language Plankalkül that could be translated into machine code (though this vision was not actually realized in the case of the Z series) and having the processor completely controlled by the code compared to rivals like Colossus and ENIAC. I believe that Zuse would have been more globally acknowledged in Computer Science of his early Z series prototypes and design documents had survived the war. Just as Turing (and his peers) were in acknowledged due to military secrecy for decades, Zuse is also largely under-recognized for his work due to being overseen by Hitler’s Germany. If it had come about prior to ‘33 or after ‘45 - Konrad Zuse’s Z series Computer Science contributions would have greater exposure historically.
@@CaptainBakerJasonit's very useful to call out the problem with Eniac
For a long time it was touted as the first computer. Americans were happy with that title for a long time, but interacting with the literally dozens of not Americans on the web has made that title harder to hold.
Each time another computer is pointed to, there's always as excuse. "We meant digital. We meant programmable..."
Now Wikipedia comes with all these adjectives: "programmable, electronic, general-purpose digital"
"general purpose" doesn't point to another page.
It doesn't really mean anything
@@CaptainBakerJason The first was the Colossus in the UK. The first programmer predated this and was, of course, Ada Lovelace.
@@geoffjones5421arguably this isn't entirely correct either, as Charles Babbage (whom Ada Lovelace worked under) had a machine that could do mathematical calculations, albeit entirely mechanical but still programmable and digital which was even earlier, and that's not counting the (theorised to be) ancient planetary computer that is believed to track the star positions and is thousands of years old.
American's are right to be proud of their achievements, they are many, varied, and transformative of the world we live in today for better or worse, but their work is built on the shoulders of giants, many of whom they deliberately ignore in their attempt to prove their exceptionalism. :(
@@SeanOfEarth Much of the video does sound like a scripted narratve, but it is still some interesing information even if "literal first" does not properly apply. Fact is, the literal first computer might have been created down in a root cellar somewhere but never got publicity before a landslide buried it and its creator.
The only way to know the "first" with certainty is to know every action ever taken.
And even the supercomputers of today cannot reach that level - and never will.
Yes, even computers cannot become God :-0)
I absolutely love this channel and your mindset regarding it. I don't know the path I took to discover this channel, but I am glad I did. Keep up the excellent work Dave 😊
This is great stuff! We're still early-enough in the history of computers to have access to the people who were there; it's great that you're preserving this history for future generations!
Try SCCS Interface , Interface Age, Dr Dobbs. Byte, PCW,
I knew I should've paid more attention when dealing with 70s mainframes, and later, PC-DOS. Really enjoy your videos, and how you share your knowledge. Maybe it will make me pull all the punch cards saved from decades past. Great job.
Video idea: program a floating-point binary number class in C++ that is fully compatible with the standard double type so that we can understand how it works!
I tried doing that (in a different language but that's irrelevant) and very quickly realised the daunting task of printing and parsing a floating-point. There have been papers written about it and dedicated libraries for doing this fast and accurate (than glibc's parsing and printing function).
It is easy if you just convert your number to the language's primitive floating-points and let them handle the printing and parsing, but it's not as fun then.
just use a union to map a float over a bitfield struct.
4:50 the Manchester Baby was the first stored program computer. The Manchester TC was the world's first transistorised computer.
From my first week of computer science I remember Zuse Z3 as the first programmable Computer.
It was built 3 years earlier than ENIAC, could process binary numbers and was (theoretical, with some tricks) touring complete.
This is great! Love the history lecture that tie in with the specific subject. Always fun to learn how it was done and why we don't anymore.
I heard somewhere that in some versions of the tape computers, the letter codes for tape were chosen so that the most common letters used would require the least amount of holes, thus increasing the lifespan of the machine that needs to punch them out and read the tape. Not sure if this is true though.
I had learned a similar anecdote, but it stated it had to do with the lifespan of the tape rather than the machine that punches it.
In Morse code, the most common letters used, are represented with the shortest codes to improve efficiency and to increase the ‘lifespan’ of the telegraphers wrist.
The close relation between telegraph, teletext, and tape computers might explain why the same rationale was projected onto the tape and the tape machines.
Dave your presentations are fascinating! Thank you for the time you put into these histories.
Fact: Dave is awsome
Theory: We will never survive an atomic disaster without smart people like Dave.
Counter point, if it wasn't for smart people like Dave, we wouldn't need to worry about an atomic disaster to begin with.
counter point, an atomic disaster close enough is not survivable.
for the people saying:
KS-Kansas NJ-New Jersey FL-Florida
pre 1963 abbreviations were KANS, NJ, FLA
dead end.
If you haven't read it, "Turing's Cathedral" is a fascinating tale of early computer development.
Boy, this took me back. I haven't thought much about the history of computers and hexadecimal since my first computer science classes in 1982. Yes, Dave, I'm older than you are. I'm very much enjoying your videos. My wife comes into my office, asks me what I'm what I'm watching, and when I try to explain it she simply says, "oh, nerd stuff." Then she walks out again. I don't bother reminding her that the "nerd stuff" is what pays my salary.
I see hexadecimal an sexadecimal as interchangable, and see the different letters as just different things used before it was standardised, just like different names were. Some decided to go u,v,w,x,y,z.
Some people suggested entirely separate character sets, including different characters for 0-9.
And you don't need computers at all. People were thinking about different bases (including base 16) before we had computers.
Very interesting indeed. Appreciate all the sleuthing on this! Very cool stuff.
The ICL mainframe I started programming on had a 6-bit byte and hence a 24-bit word, with the digits 0-9 having binary values of 0-9. The OpSys, George 3+, remains one of the best I have ever used.
My first job after leaving the RAF in 1962 was at GEC Electronics in Coventry, as a final test engineer on the ICT 1301 manufacture. That machine used four bits plus parity in twelve digit words. I later came across bytes, half words, words and double words when I joined IBM and worked on System 360/370 and 3030 series.
Brilliant episode. I loved the hunt you had to go on to find the ’answer’ (and the bloopers at the end).
At 9:54 you mention "big, old analog transistors"--but actually, all transistors are analog. If you look at the various "digital" waveforms in a modern computer, you won't find a square edge anywhere. When transistors have to switch on and off at gigahertz frequencies, the time to switch on and off is a significant fraction of the clock period.
Analog/digital is a characteristic of a circuit not of an electronic component.
A transistor, being an electronic component, is not analog nor digital. But it can be used to build both analog and digital circuits.
Plus for FETs we still often call the linear region, where varying the gate voltage has a linear response like a resistor, the triode region since the behavior is like a triode vacuum tube.
As were valves and long before that the beads on an abacus.
Even the original _digital_ counting (ie using fingers) involves ignoring the finer details of the exact position of the finger: pinkie is either up or down, and halfway between is a firmware error in the finger driver.
On the abacus a bead has to be at one end of the free space or the other.
Circuits for digital counting likewise have to be designed to force the valve or transistor to flip quickly through the intermediate states between one end of its range and the other.
Only when we get to quantum physics do we get physical systems that are actually confined to give discrete results (binary in the case of a spin-half particle, more digits for some other quantised effects).
The joke is that the analogue behaviour of a transistor is actually the aggregate of a lot of quantum subsystems: as are valves, beads, and fingers.
An analog transistor will have a smooth characteristic across the range of voltages applied to the gate. A digital one is optimized to switch as fast as possible between extrema, what happens in the forbidden band in between be damned.It's a difference in technology. You could just as well say "solar cells are just LEDs" - and while technically true, it's misguiding, because they thoroughly suck at emitting light when powered, and are optimized for collecting it and producing electric current.
herd of a company slang, calling their darlington transistors (amplification chains) just digital transistors as their processor GPIOs would often drive them directly.
FETs are also more sort of digital ones due to their very useable fully-open/closed states.
I remember in school poking around in cabinets of equipment for ILLIAC maintenance team in room L510 of The Digital Computer Lab building, or in the storage room under the building extension. There was so much early computer history just piled up most just waiting to eventually go in the dumpster. In the days of processors on desktops exceeding 200MHz and 128megs of ram, everyone viewed all those historical systems garbage. Today one could have built a world renowned museum from those rooms alone.
Something you didn't mention as a possibility. Do the letters relate to related people's initials somehow, KSN and JFL? Or KS, NJ, FL Or wait, maybe they are states?
I don't think so. We are on longer solving problem "why those 6 characters for sexadecimal", we are solving problem "why are 30+ characters on that teletype ordered in seamingly random order". You will not order that many characters in this particular way just because six of them have some meaning for you.
@@petrkubena Why wouldn't you. It would help in remembering the order. Kansas, New Jersey, and Florida are easier to remember than the random letter of KSNJFL. And, when explaining to someone on how to memorize this set, they could refer to how the correlate with the states.
Yes! Like in the movie Brazil... The ghost in the machine: EREIAMJH, an anagram of JEREMIAH.
"It's as though he's there still speaking to me - 'ere I am, J.H.'"
@@melissa3920 I like this theory, but if it was just about picking letters easy to remember the order... abcdef is a lot easier.
@@jonathanreal8018 Yes, but those are all on one side of the keyboard. Spreading them out, would allow for faster typing. Now choosing which letters, and a easy way to memorize them would be a possible concern. But who knows. They could have chosen the letters using darts.
It's all Double-Dutch to me but never-the-less I find this engrossing and entertaining, thanks.
The accent's pretty good actually. Greeting from Austria.
Thank-you! :-)
Hi Dave, great video. Always a pleasure to watch you. I ordered 2 of your coffee mugs. One for me and one for my son; who BTW is also named Dave. Thank you for your channel.
I always giggle like a madman when thinking of those tube monstrosities needing half a powerplant of their own, while holding a fricking Arduino in the hands that runs circles around those things :D
Very interesting subject. And I think you are the first person I've watched that's not in it for the money. Also, I too get bugged when something is different for no reason and I set out to find out why. I think I might like your channel. This is the first time I've come across it, so I will be checking out more of your videos. You have a great voice for talking and keeps it interesting.
Dave, i love your challenge to yourself. And editing yourself in 4k is appreciated sir
Fact: I am very interested in any further research you do on this topic as my instructors never went any deeper than to mention the alternate sexadecimal scheme in passing.
Theory: I believe that I am not the only person in your audience that shares this interest.
Thank you very much for all the time and effort you put into this topic, like all of the other videos that I've seen from you this is top shelf!
I will need to watch this at least twice. Well done.
Interesting. KSNJFL are all middle row letters except for N, using 2 on the left & 4 on the right side of the keyboard leaving ADGHK free. F & J are the locater keys so it makes sense to use them. Why the others were picked may as you say have been lost to time but I think you're on the right track with hammer jams & readability. Maybe even the punch tape was more reliably read with certain sequences.
These days we're supposed to use Unicode which I don't know much about. Love the bloopers Dave!
My first thought was it was for outputting to segmented displays... with regular hexadecimal, it's almost impossible to differentiate an 8 from a B! However, N can't easily be represented. Very intriguing video!
no tubes were involved. ;-)
or maybe it was more likely that it would be full letter Nixie tubes.
nice idea, but you could use lower case b and lower case n, if display on a 7 segment was a concern. Don't think 7-segment displays were already invented at that time.
See the various FranLab discussions of 'archaic' character displays for some options...
I must say, brilliant detective work. Thanks for sharing info I never knew I would want to know.
Could KSN and JFL be initials of two individuals involved in the project ?
That was my first thought, too.
This is possible, but highly unlikely. Designing and implementing these systems was crazy difficult, arranging the keys by your initials would make it harder yet.
I was thinking something similar only 3 people KS, NJ and FL, then thought what about if the 3 came from Kansas, New Jersey and Florida?
You certainly get my like. I'm also going to flag this to my brother in law - he worked on the Eniac, so this will be a walk down memory lane for him. - *_core_* memory!
So is sexadecimal a numbering system in base 69?
Nice!
I love it when a mystery, that I didn't even know existed, is solved! Fun video Dave.
"I just like saying SEXadecimal"
Me too, Dave. Me too.
😏
Sextillion
Undoubtetly whoever came up with Hexadecimal knew about Sexadecimal.
You might enjoy the "Seximal" numbering scheme by jan misali then. There's a quite interesting argument that base-6 would be the best base for common use, instead of base 10 or 12
This is all way above my understanding, I may have to watch it again instead of just listening to it. I like finding things that are hard to understand because I also have the incessant curiosity.