4 ms·

IBM, sonic delay lines, and the history of the 80×24 display (2019)

97 points by rbanffy 8mo ago

lysace 8mo ago
No idea if this was a factor, but 80x25 on the IBM PC allows for showing 80x24 plus that extra line of function key labels: https://en.wikipedia.org/wiki/IBM_BASIC#/media/File%3AIBM_Cassette_BASIC.png https://en.wikipedia.org/wiki/IBM_BASIC#/media/File%3AIBM_Ca... (IBM BASIC screenshot)
- bluedino 8mo ago
  Imagine when edit.com came out and QBASIC used it for the editor. You lost two more lines of valuable code space! https://en.wikipedia.org/wiki/MS-DOS_Editor#/media/File%3AMSDOS_Edit.png https://en.wikipedia.org/wiki/MS-DOS_Editor#/media/File%3AMS...
  - jtarrio 8mo ago
    You know, this is funny because QBasic did not use EDIT.COM. Instead, QBasic was the editor and EDIT.COM was a simple program that called "QBASIC /EDIT" :-)
    - bluedino 8mo ago
      It was basically the same thing. That's my point.
  - lysace 8mo ago
    I recently went back to my 1993 Turbo Pascal code (mostly 2D VGA and Sound Blaster game engine experiments) on period correct hardware. I was surprised by how claustrophobic it felt to only see 21 lines of code in e.g. Turbo Pascal 7.0. Still didn’t like the squashed 80x43 mode. https://winworldpc.com/screenshot/c38a28c3-84c3-ba28-1011-c3a4e284a2ef/c10ab705-6138-11ea-8c4a-fa163e9022f0 https://winworldpc.com/screenshot/c38a28c3-84c3-ba28-1011-c3... Then I remembered how larger displays and xterm felt like such a liberation a few years later.
    - rasz 8mo ago
      Power users had superEGA with 132 column 40/43/44/60/66 row modes.
      - lysace 8mo ago
        I think mostly just Lotus 1-2-3 (the standard textmode spreadsheet application of the time) supported that?
    - rbanffy 8mo ago
      Shorter displays end up making programs more modular with shorter functions. I always encourage my fellow Python developers to stick to 80 columns for readability.
hanfoo 8mo ago
Deeply fascinated by these historical threads. It is precisely the various design choices made throughout history that have shaped the computer systems we use today.
II2II 8mo ago
Tangentially related: is there a history covering IBM's development of microcomputers? It is clear that the traditional story of the development of the IBM PC leaves out many important details. There the 5100/5110/5120, which goes back to the mid-1970's and reflects the stereotype of IBM. There is also the System/23 DataMaster, where the hardware seems to be the basis of the IBM PC. This seems to go against the traditional story that the IBM PC was some sort of renegade project. (If anything, they appear to be companion projects. The main difference being the DataMaster's focus upon IBM firmware/software.)
- kens 8mo ago
  Like I need another big project :-) The IBM Datamaster is an interesting system, but it was doomed. It had an 8-bit Intel 8085 processor, cost $9000, and came out in July 1981. The IBM PC had a 16-bit 8088 processor, cost $1565, and came out a month later. So there was no reason to buy a Datamaster There's a good description of Datamaster in "A Personal History of the IBM PC" by Dave Bradley (one of the PC's designers). Unfortunately, it's paywalled.[1] [1] https://doi.ieeecomputersociety.org/10.1109/MC.2011.163 https://doi.ieeecomputersociety.org/10.1109/MC.2011.163
  - II2II 8mo ago
    That sounds like a great starting point. There's a university library 500 meters from where I work, so bypassing the paywall ought to be easy. And thank you for all of your articles over the years. They border so close to applied physics they are fascinating reads.
  - pinewurst 8mo ago
    https://annas-archive.gl/scidb/10.1109/mc.2011.163/ https://annas-archive.gl/scidb/10.1109/mc.2011.163/
  - rbanffy 8mo ago
    > Like I need another big project :-) You know you want it. > So there was no reason to buy a Datamaster It did use a screen font closer to the 3270 series. That is a big pro ;-) IBM also was studying home computers, and was talking to Atari. There are other gorgeous design studies in “ Delete - a Design History of Computer Vaporware”, a lovely book. I want to 3D print half of those designs.
veltas 8mo ago
From a linked article on shift registers: > To avoid these astronomical prices, some computers used the cheaper alternative of shift register memory. Might be a direction for 2026 too?
thakoppno 8mo ago
One theory I saw argued the punch card size was the reason for 80x24. But why were punch cards that size? They were designed off of the cards used for the census. Why were the census cards that size? Because they were modeled after the dollar bill size. I do love thought experiments like this but do believe they’re insatiably unresolvable.
- ronsor 8mo ago
  In the end, all reasons resolve to either "it's what we had at the time" or "someone thought it looked good."
  - theamk 8mo ago
    Not always, for example original CD disks had capacity of 74 minutes to accommodate Beethoven's Ninth Symphony.
    - rasz 8mo ago
      That one also turned out a myth :) CD size was determined by Cassette tape dimensions (diagonal, human can still hold one in one hand) and that combined with conservative pits/lands/track pitch choice drove the play time. thus CD runtime was derived from something "what we had at the time".
      - antiterra 8mo ago
        The story as told may be inaccurate but it wasn’t simply ‘what we had at the time’ either. The 74 minute length resulted from Sony rejecting the Philips 60 minute 11.5cm diameter “Pinkeltje” disc size in favor of a 12cm diameter. It’s quite possible that Sony’s Norio Ohga simply argued that the 9th symphony or various operas fitting would be enough of an advantage for the slight size increase without meaningfully decreasing portability.
        rzzzt 8mo ago
        Meanwhile the LP crowd was flipping sides like it was Ultima VIII (slight exaggeration). Why would it be critical for a new format to do away with multi-disc releases if the customer base has already grown accustomed to them?
BirAdam 8mo ago
Man. I love the design of old terminals, computers, and such. I am, also, extremely glad that these form factors were abandoned. Having an old terminal, it is possibly the least ergonomic machine I have ever used.
- TacticalCoder 8mo ago
  Same. And you may like this one: https://design.tel/olivetti-tvc-250/ https://design.tel/olivetti-tvc-250/
  - rbanffy 8mo ago
    I’d make it my daily driver ergonomics be damned. Though I’d probably go with an IBM 3278 or 3290 for practicality.
Animats 8mo ago
The PARC crowd thought displays should have the form factor of a sheet of paper. Hence the Alto display.[1] That never caught on. [1] https://www.righto.com/2018/01/xerox-alto-zero-day-cracking-disk.html https://www.righto.com/2018/01/xerox-alto-zero-day-cracking-...
- locallost 8mo ago
  I see people doing that today.
  - rbanffy 8mo ago
    They are correct of course.
thangalin 8mo ago
The linage can be traced back to Basile Bouchon's paper tape invention in 1725. The article doesn't mention the role of punched cards in The Holocaust, though, which my blog post goes into: https://dave.autonoma.ca/blog/2019/06/06/web-of-knowledge/ https://dave.autonoma.ca/blog/2019/06/06/web-of-knowledge/
pezezin 8mo ago
Fascinating article, I really like knowing where the old standards came from. But I am extremely curious the first picture in the "The IBM 2260 video display terminal" section. All the other pictures show the typical extremely round CRT of the era, but that one is the characteristic cylindrical tube of Trinitrons, a technology released several years later. I am trying to find some information about it to no avail.
- kens 8mo ago
  The manual for the IBM 2260 describes the CRT in detail but I don't think it has the information you want. My guess is that if you're IBM, you can get the CRT in whatever shape you want. [1] https://www.bitsavers.org/pdf/ibm/2260/Y27-2046-3_2260_2848_FETOM_Mar68.pdf#page=45 https://www.bitsavers.org/pdf/ibm/2260/Y27-2046-3_2260_2848_...
  - pezezin 8mo ago
    Wow, that is a very detailed manual. I am trying to find more information to no avail. Actually the only picture of an IBM terminal with a cylindrical tube is the one in your article, there is nothing else on the entire Internet. I will keep investigating.
    - kens 8mo ago
      Oh, I see the difference now. In the photo of the IBM 2260 terminal that I used, the bezel is rectangular and flat. In every other photo of the 226, the bezel is sunken and the screen is oval-shaped. I'll ask around and see if I can find out why. Maybe the photo that I used is a later version with a better CRT?
      - rbanffy 8mo ago
        I wonder what the knob on the side was for. Vertical position seems absurd. That’s also a very large screen compared to how much text it’s displaying. The 2260 was a very odd contraption.
  - 8mo ago
useftmly 8mo ago
[dead]
ndiddy 8mo ago
I always assumed the reason the PC had an 80x25 display was to allow for emulating the 3270 (80x24 for content and then an extra line at the bottom for the status information) so it's interesting to see that wasn't the case. > Also, 3270 emulation was not a requirement or goal for the IBM PC, and didn't lead to the 25th line. I would much prefer to have a tidy story where the 3270 led to the PC's display, but unfortunately that's not the case. I talked to two of the original IBM PC engineers to check on this. The said the IBM PC team felt absolutely no need to be compatible with other IBM products. In particular, several features of the PC made 3270 compatibility harder: the use of ASCII instead of EBCDIC, little-endian words, and 10 function keys instead of 12.
tracker1 8mo ago
No mentioned are the PC terminal apps from BBS era dialup connections that tended to use one of the 80x25 lines as a status bar, leaving 24 lines for the remote display... not to mention avoiding or embracing wrap issues that sometimes left you from using column 80 reliably either. It's kind of nice to see a lot of resurgence in terminal technologies and remote terminal apps these days... I think AI is letting a lot of people that have been nostalgic pick up projects that would have been too daunting otherwise. Myself included.

97 points by rbanffy 8mo ago

No idea if this was a factor, but 80x25 on the IBM PC allows for showing 80x24 plus that extra line of function key labels: https://en.wikipedia.org/wiki/IBM_BASIC#/media/File%3AIBM_Cassette_BASIC.png https://en.wikipedia.org/wiki/IBM_BASIC#/media/File%3AIBM_Ca... (IBM BASIC screenshot)

bluedino 8mo ago
Imagine when edit.com came out and QBASIC used it for the editor. You lost two more lines of valuable code space! https://en.wikipedia.org/wiki/MS-DOS_Editor#/media/File%3AMSDOS_Edit.png https://en.wikipedia.org/wiki/MS-DOS_Editor#/media/File%3AMS...
- jtarrio 8mo ago
  You know, this is funny because QBasic did not use EDIT.COM. Instead, QBasic was the editor and EDIT.COM was a simple program that called "QBASIC /EDIT" :-)
  - bluedino 8mo ago
    It was basically the same thing. That's my point.
- lysace 8mo ago
  I recently went back to my 1993 Turbo Pascal code (mostly 2D VGA and Sound Blaster game engine experiments) on period correct hardware. I was surprised by how claustrophobic it felt to only see 21 lines of code in e.g. Turbo Pascal 7.0. Still didn’t like the squashed 80x43 mode. https://winworldpc.com/screenshot/c38a28c3-84c3-ba28-1011-c3a4e284a2ef/c10ab705-6138-11ea-8c4a-fa163e9022f0 https://winworldpc.com/screenshot/c38a28c3-84c3-ba28-1011-c3... Then I remembered how larger displays and xterm felt like such a liberation a few years later.
  - rasz 8mo ago
    Power users had superEGA with 132 column 40/43/44/60/66 row modes.
    - lysace 8mo ago
      I think mostly just Lotus 1-2-3 (the standard textmode spreadsheet application of the time) supported that?
  - rbanffy 8mo ago
    Shorter displays end up making programs more modular with shorter functions. I always encourage my fellow Python developers to stick to 80 columns for readability.

hanfoo 8mo ago

Deeply fascinated by these historical threads. It is precisely the various design choices made throughout history that have shaped the computer systems we use today.

II2II 8mo ago

Tangentially related: is there a history covering IBM's development of microcomputers? It is clear that the traditional story of the development of the IBM PC leaves out many important details. There the 5100/5110/5120, which goes back to the mid-1970's and reflects the stereotype of IBM. There is also the System/23 DataMaster, where the hardware seems to be the basis of the IBM PC. This seems to go against the traditional story that the IBM PC was some sort of renegade project. (If anything, they appear to be companion projects. The main difference being the DataMaster's focus upon IBM firmware/software.)

kens 8mo ago
Like I need another big project :-) The IBM Datamaster is an interesting system, but it was doomed. It had an 8-bit Intel 8085 processor, cost $9000, and came out in July 1981. The IBM PC had a 16-bit 8088 processor, cost $1565, and came out a month later. So there was no reason to buy a Datamaster There's a good description of Datamaster in "A Personal History of the IBM PC" by Dave Bradley (one of the PC's designers). Unfortunately, it's paywalled.[1] [1] https://doi.ieeecomputersociety.org/10.1109/MC.2011.163 https://doi.ieeecomputersociety.org/10.1109/MC.2011.163
- II2II 8mo ago
  That sounds like a great starting point. There's a university library 500 meters from where I work, so bypassing the paywall ought to be easy. And thank you for all of your articles over the years. They border so close to applied physics they are fascinating reads.
- pinewurst 8mo ago
  https://annas-archive.gl/scidb/10.1109/mc.2011.163/ https://annas-archive.gl/scidb/10.1109/mc.2011.163/
- rbanffy 8mo ago
  > Like I need another big project :-) You know you want it. > So there was no reason to buy a Datamaster It did use a screen font closer to the 3270 series. That is a big pro ;-) IBM also was studying home computers, and was talking to Atari. There are other gorgeous design studies in “ Delete - a Design History of Computer Vaporware”, a lovely book. I want to 3D print half of those designs.

veltas 8mo ago

From a linked article on shift registers: > To avoid these astronomical prices, some computers used the cheaper alternative of shift register memory. Might be a direction for 2026 too?

thakoppno 8mo ago

One theory I saw argued the punch card size was the reason for 80x24. But why were punch cards that size? They were designed off of the cards used for the census. Why were the census cards that size? Because they were modeled after the dollar bill size. I do love thought experiments like this but do believe they’re insatiably unresolvable.

ronsor 8mo ago
In the end, all reasons resolve to either "it's what we had at the time" or "someone thought it looked good."
- theamk 8mo ago
  Not always, for example original CD disks had capacity of 74 minutes to accommodate Beethoven's Ninth Symphony.
  - rasz 8mo ago
    That one also turned out a myth :) CD size was determined by Cassette tape dimensions (diagonal, human can still hold one in one hand) and that combined with conservative pits/lands/track pitch choice drove the play time. thus CD runtime was derived from something "what we had at the time".
    - antiterra 8mo ago
      The story as told may be inaccurate but it wasn’t simply ‘what we had at the time’ either. The 74 minute length resulted from Sony rejecting the Philips 60 minute 11.5cm diameter “Pinkeltje” disc size in favor of a 12cm diameter. It’s quite possible that Sony’s Norio Ohga simply argued that the 9th symphony or various operas fitting would be enough of an advantage for the slight size increase without meaningfully decreasing portability.
      - rzzzt 8mo ago
        Meanwhile the LP crowd was flipping sides like it was Ultima VIII (slight exaggeration). Why would it be critical for a new format to do away with multi-disc releases if the customer base has already grown accustomed to them?

BirAdam 8mo ago

Man. I love the design of old terminals, computers, and such. I am, also, extremely glad that these form factors were abandoned. Having an old terminal, it is possibly the least ergonomic machine I have ever used.

TacticalCoder 8mo ago
Same. And you may like this one: https://design.tel/olivetti-tvc-250/ https://design.tel/olivetti-tvc-250/
- rbanffy 8mo ago
  I’d make it my daily driver ergonomics be damned. Though I’d probably go with an IBM 3278 or 3290 for practicality.

Animats 8mo ago

The PARC crowd thought displays should have the form factor of a sheet of paper. Hence the Alto display.[1] That never caught on. [1] https://www.righto.com/2018/01/xerox-alto-zero-day-cracking-disk.html https://www.righto.com/2018/01/xerox-alto-zero-day-cracking-...

locallost 8mo ago
I see people doing that today.
- rbanffy 8mo ago
  They are correct of course.

thangalin 8mo ago

The linage can be traced back to Basile Bouchon's paper tape invention in 1725. The article doesn't mention the role of punched cards in The Holocaust, though, which my blog post goes into: https://dave.autonoma.ca/blog/2019/06/06/web-of-knowledge/ https://dave.autonoma.ca/blog/2019/06/06/web-of-knowledge/

pezezin 8mo ago

Fascinating article, I really like knowing where the old standards came from. But I am extremely curious the first picture in the "The IBM 2260 video display terminal" section. All the other pictures show the typical extremely round CRT of the era, but that one is the characteristic cylindrical tube of Trinitrons, a technology released several years later. I am trying to find some information about it to no avail.

kens 8mo ago
The manual for the IBM 2260 describes the CRT in detail but I don't think it has the information you want. My guess is that if you're IBM, you can get the CRT in whatever shape you want. [1] https://www.bitsavers.org/pdf/ibm/2260/Y27-2046-3_2260_2848_FETOM_Mar68.pdf#page=45 https://www.bitsavers.org/pdf/ibm/2260/Y27-2046-3_2260_2848_...
- pezezin 8mo ago
  Wow, that is a very detailed manual. I am trying to find more information to no avail. Actually the only picture of an IBM terminal with a cylindrical tube is the one in your article, there is nothing else on the entire Internet. I will keep investigating.
  - kens 8mo ago
    Oh, I see the difference now. In the photo of the IBM 2260 terminal that I used, the bezel is rectangular and flat. In every other photo of the 226, the bezel is sunken and the screen is oval-shaped. I'll ask around and see if I can find out why. Maybe the photo that I used is a later version with a better CRT?
    - rbanffy 8mo ago
      I wonder what the knob on the side was for. Vertical position seems absurd. That’s also a very large screen compared to how much text it’s displaying. The 2260 was a very odd contraption.
- 8mo ago

useftmly 8mo ago

[dead]

ndiddy 8mo ago

I always assumed the reason the PC had an 80x25 display was to allow for emulating the 3270 (80x24 for content and then an extra line at the bottom for the status information) so it's interesting to see that wasn't the case. > Also, 3270 emulation was not a requirement or goal for the IBM PC, and didn't lead to the 25th line. I would much prefer to have a tidy story where the 3270 led to the PC's display, but unfortunately that's not the case. I talked to two of the original IBM PC engineers to check on this. The said the IBM PC team felt absolutely no need to be compatible with other IBM products. In particular, several features of the PC made 3270 compatibility harder: the use of ASCII instead of EBCDIC, little-endian words, and 10 function keys instead of 12.

tracker1 8mo ago

No mentioned are the PC terminal apps from BBS era dialup connections that tended to use one of the 80x25 lines as a status bar, leaving 24 lines for the remote display... not to mention avoiding or embracing wrap issues that sometimes left you from using column 80 reliably either. It's kind of nice to see a lot of resurgence in terminal technologies and remote terminal apps these days... I think AI is letting a lot of people that have been nostalgic pick up projects that would have been too daunting otherwise. Myself included.