Trends in keyboard design: Who started what?

tigpha

27 Aug 2019, 17:27

I don't see what the matter is with that beam-spring layout :-) Having said that, my daily driver is an IBM Bigfoot, both at home and work. Love that size, noise, heft and shape. Funky layouts don't deter me, I reconfigure my own layouts and the multifarious standards be damned ...

User avatar
depletedvespene

27 Aug 2019, 17:30

It's (probably) not about the physical layout, but about the logical ("data entry") layout placed on top.

User avatar
mr_a500

27 Aug 2019, 17:32

vometia wrote:
27 Aug 2019, 17:19
Trends and beam-springs. Happily it seems this beam-spring layout didn't catch on. I'm not going to say "I'm sure there was some logic behind it" because it's clearly absurd.

Image
That layout did catch on. It was used on keypunch machines from the late 1940's until keypunch machines became obsolete in the late 1970's. I've got a few keyboards with this layout.

User avatar
depletedvespene

27 Aug 2019, 17:38

mr_a500 wrote:
27 Aug 2019, 17:32
That layout did catch on. It was used on keypunch machines from the late 1940's until keypunch machines became obsolete in the late 1970's. I've got a few keyboards with this layout.
Keypunch machines did NOT have lowercase letters, right?

tigpha

27 Aug 2019, 17:47

Unlikely that character-case was restricted: the 80 column punchcards seem to have at least eight bits per character. Some punchcards have twelve bits per column, with four bits presumably for error checking/correcting. Only paper tape with five-bit data characters would conceivably have such a restriction to upper-case only, since there's only 32 available patterns to represent the alphabet and punctuation.

User avatar
depletedvespene

27 Aug 2019, 17:52

tigpha wrote:
27 Aug 2019, 17:47
Unlikely that character-case was restricted: the 80 column punchcards seem to have at least eight bits per character. Some punchcards have twelve bits per column, with four bits presumably for error checking/correcting. Only paper tape with five-bit data characters would conceivably have such a restriction to upper-case only, since there's only 32 available patterns to represent the alphabet and punctuation.
Considering this keyboard seems to have exactly 63 characters available (+ space == 64), I'm gonna go with six-bit data and, indeed, no lowercase letters at all.

Note how comma, period and minus are repeated, and why, for some reason, A and Z don't have any associated symbol on their Shift layer.

tigpha

27 Aug 2019, 18:00

Curious. It is likely that the ancient IBM systems used EBCDIC, and not ASCII, and that was specified as having an eight bit character set.

tigpha

27 Aug 2019, 21:37

Oh, hang on, it appears that my naïve assumption about binary character encoding on 80 column punch cards is probably very mistaken:
ibmcard.gif
ibmcard.gif (238.21 KiB) Viewed 293 times
So much waste!

User avatar
vometia
irritant

28 Aug 2019, 05:17

depletedvespene wrote:
27 Aug 2019, 17:26
Well, this IS a "data entry" layout, which has its own set of rules. I've seen it in other keyboards as well, not just in beamsprings.
I think it was the arrangement of function keys that got to me more than anything else. But even if there's a good reason for the remainder, as much as I keep harping on about the virtues of sphericals with nice big legends, I think that'd be my keyset from hell!

User avatar
vometia
irritant

28 Aug 2019, 05:21

tigpha wrote:
27 Aug 2019, 17:47
Unlikely that character-case was restricted: the 80 column punchcards seem to have at least eight bits per character. Some punchcards have twelve bits per column, with four bits presumably for error checking/correcting. Only paper tape with five-bit data characters would conceivably have such a restriction to upper-case only, since there's only 32 available patterns to represent the alphabet and punctuation.
Didn't five-hole tape use shifts access alternative character sets? I think there were one or two extra usually, but I am most certainly not an expert on the subject. That's another area where the ordering of the characters tended to be quite bizarre (to me, at least).

tigpha

28 Aug 2019, 09:52

I suppose that the number of symbol bits is not a limitation: A paper tape could record data in a serial pattern of single bits, and encode the symbols with Huffman coding. He published the paper in 1952. Huffman coding information theory is contemporary with paper tapes and cards.

I went down a rabbit hole yesterday reading about the Shannon Limit and Gallager Codes, and Turbocodes etc. Way over my head. Information theory reaches hermetic levels of obscure magic at those levels!

User avatar
kps

13 Sep 2019, 18:01

tigpha wrote:
27 Aug 2019, 21:37
Oh, hang on, it appears that my naïve assumption about binary character encoding on 80 column punch cards is probably very mistaken:
ibmcard.gif
So much waste!
Cards were processed electromechanically for decades before computers existed. Machines worked row-by-row, top-to-bottom, not by column/character. The simplest adding machine simply has a counting wheel for each column, and advances the wheel one step for each row until a hole turns it off.

User avatar
kps

13 Sep 2019, 19:00

vometia wrote:
28 Aug 2019, 05:21
That's another area where the ordering of the characters tended to be quite bizarre (to me, at least).
That's because you're seeing it laid out wrong. e.g. Wikipedia shows the ITA 2 teletype code arranged according to a binary numeric interpration of the signals/holes, like this:

Code: Select all

    NUL E   LF  A   LS  S   I   U   CR  D   R   J   N   F   C   K  
    T   Z   L   W   H   Y   P   Q   O   B   G   FS  M   X   V   DEL 
    NUL 3   LF  :   LS  '   8   7   CR  ²⁄  4   ⁷⁄  −   ⅟   (   ⁹⁄  
    5   .   /   2   ⁵⁄  6   0   1   9   ?   ³⁄  FS  ,   £ ␣ )   DEL 
But nothing cared about the binary number interpretation a hundred years ago. The code was ordered by letter frequency, with the most common symbols having the fewest holes:

Code: Select all

  ┌─────┐ 
  │01234│   Let  Fig
  │     │   ---  ---
  │•    │   E    3
  │    •│   T    5
  │••   │   A    :
  │ ••  │   I    8
  │  •• │   N    −
  │   ••│   O    9
  │• •  │   S    '
  │ • • │   R    4
  │  • •│   H    ⁵⁄
  │•  • │   D    ²⁄
  │ •  •│   L    /
  │•••  │   U    7
  │ ••• │   C    (
  │  •••│   M    ,
  │• •• │   F    ⅟
  │••  •│   W    2
  │• • •│   Y    6
  │ •• •│   P    0
  │•  ••│   B    ?
  │ • ••│   G    ³⁄
  │ ••••│   V    )
  │•••• │   K    ⁹⁄
  │••• •│   Q    1
  │• •••│   X    £
  │•• ••│   FIGS FIGS
  │•••••│   DEL  DEL
  │•   •│   Z    .
  │•• • │   J    ⁷⁄
  │     │   Idle Idle
  │ •   │   LF   LF
  │   • │   CR   CR
  │  •  │   LTRS LTRS
  └─────┘ 
Why were the digits in this bizarre order?

Code: Select all

  ┌─────┐ 
  │•    │   E    3
  │    •│   T    5
  │ ••  │   I    8
  │   ••│   O    9
  │ • • │   R    4
  │•••  │   U    7
  │••  •│   W    2
  │• • •│   Y    6
  │ •• •│   P    0
  │••• •│   Q    1
  └─────┘ 
Just look at it on a keyboard:

Code: Select all

  ┌─────┐ 
  │••• •│   Q    1
  │••  •│   W    2
  │•    │   E    3
  │ • • │   R    4
  │    •│   T    5
  │• • •│   Y    6
  │•••  │   U    7
  │ ••  │   I    8
  │   ••│   O    9
  │ •• •│   P    0
  └─────┘ 

User avatar
vometia
irritant

Yesterday, 13:29

kps wrote:
13 Sep 2019, 19:00
That's because you're seeing it laid out wrong. e.g. Wikipedia shows the ITA 2 teletype code arranged according to a binary numeric interpration of the signals/holes, like this:

[snip]

But nothing cared about the binary number interpretation a hundred years ago. The code was ordered by letter frequency, with the most common symbols having the fewest holes: [...] Why were the digits in this bizarre order? Just look at it on a keyboard [...]
Awesome, thanks for that! Now it makes sense. I've been using binary for so long that I'd completely forgotten that there was a world before that was The Standard (similar to how a co-worker looked at me like I'd been dropped on my head when I tried describing 36-bit computers to him: "that's impossible!" on the basis that anything that wasn't a power-of-two or at least constructed from 8-bit bytes seemed so implausible, though he was also a turd) and of course I only have to cast my mind back as far as e.g. Morse code. Pity my grandfather is no longer around to talk to about this stuff, it would've fascinated him as he was fluent in both Morse and typewriters, being both an RAF comms guy in WWII and then working as a BBC script-writer post-war. He was fascinated by all this stuff.

I recall learning about letter frequencies in my childhood from The Secret Agent's Handbook, an amusing take on espionage and so forth; they were part of its codebreaking section. Perhaps not so useful nowadays, at least not from that point of view, but it gave me an understanding of what it is and why it might be useful elsewhere too.

I understand that the computers at Bletchley Park used the standard paper tape codes for expedience as that way they could use existing equipment to transcribe and print messages: not surprising since the budget was limited and the computers themselves were cobbled together from the standard bits and pieces used to construct telephone exchanges. Besides which, I guess there wasn't much point in creating a brand new character set since it was binary anyway, even though its non-alphabetic arrangement makes me somewhat comically reminded of Eric Morecambe's protestation that "I am playing the right notes! ...just not necessarily in the right order."

Post Reply

Return to “Keyboards”