GE Bag-Biters

[larsbrinkhoff]

Aka Datanet 760.

EDIT: See information below. The 760 isn't the Bag-Biter.

EDIT again: Or maybe it is!

[larsbrinkhoff]

Source: http://www.histoireinform.com/Histoire/+infos2/chr4infd.htm

[larsbrinkhoff]

Source: http://www.bitsavers.org/pdf/computersAndAutomation/196512.pdf

[larsbrinkhoff]

https://ia800803.us.archive.org/22/items/TNM_Video_Data_Management_System_GE_DATANET-760_-_20170915_0074/TNM_Video_Data_Management_System_GE_DATANET-760_-_20170915_0074.pdf

[larsbrinkhoff]

[larsbrinkhoff]

Spotted on film:

https://gitlab.com/larsbrinkhoff/aifilm/blob/master/aifilms/82-screen.mp4

[larsbrinkhoff]

Hello @david-moon,

Here are pictures of the GE Datanet 760, which I understand to be the Bag-Biters. There's also a link to a film.

[larsbrinkhoff]

Mr Moon informed me the Datanet 760 isn't the Bag-Biter and sent me this link:
http://www.bitsavers.org/pdf/ge/terminals/GE_Data_Editing_Display_Dec64.pdf

[larsbrinkhoff]

My information on the Datanet 760 comes from AI memo 358:

Another appendix on "Inserting lower case letters into the TECO buffer" was due to the lack of lower case terminals, and the need for a trick to get lower case into the TECO buffer from our model 33 and GE "bagbiters" (Datanet 760's).

[larsbrinkhoff]

A frame from the AI film.

Since this is supposedly from MIT but Moon remembers another terminal, I'm guessing they had both Datanet 760s and Data Editing Displays.

Greenblatt says this is Carl Engelman.

[larsbrinkhoff]

More of those in the background here?
https://archive.org/details/AIfilms/45-mac.mp4

[larsbrinkhoff]

Greenblatt calls them GE consoles. It's at 8:20 in the video, and 7:45 in the audio.

https://archive.org/details/AIfilms/81-mathlab.mp4
http://projects.csail.mit.edu/films/aifilms/Podcasts/81-mathlab.mp3

---

Another one by the Minsky-Bennett arm. From https://www.youtube.com/watch?v=JuXQPdd0hjI

[larsbrinkhoff]

TK wrote to ITS-LOVERS in 2000:

Ah, but the Datapoints were a GREAT improvement over the GE bag biters, which had a screen of 13 lines of 26 characters each (uppercase only, of course), and a model 35 teletype keyboard.

[larsbrinkhoff]

AI memo 161A, ITS 1.5 Reference Manual, has this information:

Devices T11 through T14 (the two digits are treated as an octal number) are GE Datanet 760 terminals with a character only display and keyboard.

T11 GE console by plotter. T12 GE console on 8th floor. T13 GE console by air conditioner. T14 GE console by ham rig (in far corner).

[larsbrinkhoff]

ITS 1.4 Reference Manual has the GE consoles at T15-T20.

[larsbrinkhoff]

AI memo 190 from March 1970 says

currently there are four Model 35 teletypes, four General Electric remote keyboards with character-scope displays, one ARDS console, and three telephone data-sets

[larsbrinkhoff]

I showed the pictures here to Richard Greenblatt, and he identified this as a GE console. But he didn't recognize this. I related how this was contrary to Moon's recollections, but RG couldn't say anything more.

[larsbrinkhoff]

I asked Larry Krakauer about his recollections, and he wrote me this:

I'm with Greenblatt on this. The terminals had blue metal cases, and I recall them being identified as being made by GE. I don't think the keyboard was attached to the terminal (except by a cable), but I'm not sure of that.

The text and picture in the second image on the page you linked to describes (and shows) the Datanet-760 as being able to display graphics. The GE terminals I recall absolutely could not do that - they were character mode only. I recall them as displaying 25 lines of 80 characters per line. I don't remember ever hearing the name "Datanet-760", but that doesn't mean anything.

These were not standalone terminals that connected directly to an interface in the computer. They were wired to a controller cabinet, about the size of a refrigerator. A cable ran from each terminal to an interface board in the controller, which I THINK was capable of controlling 8 displays. The computer had a digital interface to the controller.

If you think back to the mid sixties, memory was quite expensive. The 25 X 80 = 2,000 sixbit characters for each display were stored in an acoustic delay line in the controller cabinet. When one of these delay line boards broke, we took it apart. The delay line was a spiral of wire mounted on standoffs above a board. My understanding was that the characters propagated down the delay line using Manchester Encoding or some similar protocol. When received at the far end of the delay line, the signal was cleaned up, re-synchronized, and sent back into the delay line. The propagation of the characters through the delay line was synchronized with the horizontal scan of the CRT in the corresponding terminal, so each line came up just as it was needed. I could go into this a bit more if you have any interest - I'm leaving out a lot of details.

If anyone banged his fist on the controller cabinet, the screens on ALL of the displays were cleared by the acoustic shock. As the various programs running in ITS re-wrote text on each screen, the text would have holes in it - blank character positions. This is because ITS (the Incompatible Time Sharing system) kept track of what was in each screen position. If the software wanted to write a character, let's say an "e", in a particular screen position, and ITS knew that there was already an "e" in that position, it skipped writing the character. But if the screen had in fact been cleared by some transient problem, and that position was in fact blank, the "e" would not appear. The reason for the above was that writing characters to the acoustic delay line memory was rather slow. The software put a character and its address in a buffer, but then nothing could happen until that character position came around under the write-transducer (in that sense, it was sort of like a disk). That's why the software was coded to avoid writing a character if at all possible.

There were a lot of young students in the lab, and they were bothered by the high-pitched noise that came out of these CRT displays. They were, in effect, television sets, with a horizontal scan frequency of 15,750 Hz. But it seemed that many of the flyback transformers used to generate the horizontal scan had loose laminations, and hence generated sound at that frequency. You may recall that when one is young, one can hear sound up to as high as 20,000 Hz. The noise from the transformers was apparently extremely loud, and it drove some of the young students crazy. Some of them stuffed foam rubber into the terminals to try to dampen the noise. I was 24 when I came back to the AI Lab in 1966, and already could barely hear sounds at that frequency, so it didn't bother me.

[larsbrinkhoff]

AI memo 164A from June 1970 suggests the "GE DATANET" is 25 lines by 44 columns.

[larsbrinkhoff]

The two PDFs linked above both say 46x26.

[larsbrinkhoff]

GE 760 mentioned in TK; PDP6 POOP12:

        The PDP6 Time Sharing System is a core sharing system using
a 256k Fabri-Tek core memory.  Its secondary storage consists of three
IBM 2311 disk drives, two Data-disks, and four DEC micro-tape drives.
Additional peripheral equipment consists of several model 35 teletypes,
four GE 760 scope and keyboard combinations, a papertape reader
and punch, a Data Products line printer, a DEC 340 CRT display,
a Calcomp plotter, and various robot oriented items such as
mechanical hands and eyes.

[larsbrinkhoff]

From AI memo 253, about eye tracking.

[larsbrinkhoff]

Datanet 760 information on pages 6321:00-12.

https://usermanual.wiki/Document/AuerbachDataCommunicationsReportsVol21970.1075835479

[larsbrinkhoff]

I think this patent describes the terminal controller:
https://patentimages.storage.googleapis.com/55/e0/8f/683af82a65989e/US3500338.pdf

[larsbrinkhoff]

Emulator vs photos. I think it's close.

[larsbrinkhoff]

http://www.bitsavers.org/pdf/ge/GE-6xx/CPB-1643A_Series_600_GECOS_Time-Sharing_System_General_Information_Manual_Jul1971.pdf

Page 51 and forward.

[larsbrinkhoff]

PHW and maybe GE console? From https://people.csail.mit.edu/bkph/phw_copy_demo.shtml

[larsbrinkhoff]

Character set and control codes, from the Auerbach report.

Maybe this is why ITS defaults to wrapping the cursor from bottom to top rather than scrolling.

[david-moon]

Maybe this is why ITS defaults to wrapping the cursor from bottom to top rather than scrolling.

It's really hard to read text that is scrolling out from under your gaze. Remember back in the old days transmission from the computer to the terminal was slow, it's not like the screen would scroll for a fraction of a second then stop at --MORE-- to give you a chance to read it. At 2400 baud the screen would jump 3 or 4 times a second if it was scrolling; it would be nauseating and unreadable. That is probably the reason for wrapping around rather than scrolling.

But the decision is before my time.

-Moon

[larsbrinkhoff]

ITS "1.0" listing being processed: https://github.com/PDP-6/ITS

It has support for GE consoles, "GETY".

[larsbrinkhoff]

I documented the GTYI/GTYO interface used by ITS 138:
https://gunkies.org/wiki/MIT_AI_lab_IO_bus_interface_for_GE_Datanet_760