Hack of the year: 1980

« previous post | next post »

I recently stumbled on this 5/10/2023 Medium article by David Brock, "A Backup of Historical Proportions" — which reminded me of the Xerox Palo Alto Research Center ("PARC") and the Xerox Alto. Those were the people and the machine that invented interactive GUIs on bit-mapped displays, the computer mouse, and so on — though it took Steve Jobs to "borrow" the ideas and turn them into a social (and business) success.

But as a speech person, I always thought it was odd and unfortunate that the Alto had no provision for audio input or output — and I was impressed by the hack that Henry Thompson used to get around the audio output problem for his 1980 Berkeley thesis, "Stress and Salience in English: Theory and Practice".

I remembered that he managed to turn a printer port into a digital-to-analog converter, in order to enable the interactive waveform measurement app that he wrote — but I'd forgotten the details, and the dissertation text doesn't mention them. So I wrote to Henry and asked him about it. His response:

As I recall I just soldered one end of 4 resistors in a 1:2:4:8 ratio ohms to a paperclip and stuck them into the low-order 4 pin sockets of the Alto's RS232 port, then did some trivial signal
processing to convert the 8-bit audio I had into 4-bit.

The RS-232 interface has been used for many things, but not this :-)…

I also remembered that he had to do the digitization on a different computer — since it only had to be done once, he didn't need to hack the Alto to enable the analog-to-digital end. Which was a Good Thing, because it would have been harder. I also forgot the details of the digitization arrangements, and the dissertation only says (p. 184)

This section describes the computer based interactive system for the display and reproduction of speech in digital form which I created and used to obtain the duration data discussed in
 chapter 3. As was the system described in C.1, it was written in Interlisp and makes use of video display facilities developed at Xerox PARC.

Figures 1 and 2 on the following pages are snapshots of the display screen showing what the speech looked like. The system displayed a digitized version of the time-amplitude waveform of the data segment, sampled at 7659 eight bit samples per second. 

There's been a lot of discussion about why Xerox never managed to make the Alto (or the commercial version the Star/Dandelion) into a successful product. Along with the high price, the main reason was presumably a failure of vision on the part of the company's leaders. And there were some unfortunate (and very weird) aspects of the software, like the fact that the text editor stored documents as a dump of the program's runtime heap — or so I was told. The lack of audio probably wouldn't have been a critical factor until much later — but given that the original IBM PC had audio output, and the Apple II had both audio output and input, I wonder why the innovators at PARC never saw audio I/O as worth the (minimal) trouble.




  1. Tom Ace said,

    May 21, 2023 @ 10:10 am

    RS-232C is a serial protocol. The story about the four resistors makes sense if it was a Centronics parallel port (often used for printers of the era).

  2. MattF said,

    May 21, 2023 @ 10:16 am

    Also, Xerox had no idea what was going on. The lab I was working in had a Xerox GUI machine that had broken down, so a repair person was called in. He picked up the mouse and asked “What’s this?”

  3. Gregory Kusnick said,

    May 21, 2023 @ 11:10 am

    the text editor stored documents as a dump of the program's runtime heap — or so I was told.

    I worked at PARC in the 1970s but I don't recall this detail. The two document editing systems from that era I'm most familiar with were Bravo and its successor BravoX, and neither did that.

    Bravo stored text files as text, with formatting information appended as a text string that had to be parsed each time the document was opened. BravoX stored formatting as binary data, in the same disk sector as the text it applied to, so the file was not readable as plain text without piping it through a filtering stream to strip out the formatting.

    Larry Tesler's Gypsy was a fork of the Bravo code base and (as best I can recall) inherited Bravo's text-based file format.

    I suppose it's possible that other groups in the lab used a different editing tool with the property you describe, but if so I'm not remembering it.

  4. Taylor, Philip said,

    May 21, 2023 @ 11:45 am

    Agreed that the connector was more likely to have been a Centronix, given that Henry refers to "st[icking] [four resistors] into the low-order 4 pin sockets", but I am forced to ask "why did he not stick resistors into all eight (D0 to D7) [data] pin sockets" ?

  5. Isaac Wingfield said,

    May 22, 2023 @ 12:34 am

    "why did he not stick resistors into all eight (D0 to D7) [data] pin sockets?"
    As you add more resistors (each a larger power of two), the precision of those resistors must get better and better to guarantee that the output voltage is monotonic with count.

  6. Taylor, Philip said,

    May 22, 2023 @ 2:19 am

    Thank you Isaac, that explanarion makes sense. So using 1% accuracy resistors, for example, one would be able to achieve no more than 7 bit resolution — is this correct ?

  7. Andrew Shields said,

    May 22, 2023 @ 4:01 am

    The story I heard in the mid-1980s from somebody who worked at Xerox PARC went like this: The designers went to Xerox HQ in New York state (Rochester?) with a prototype (or perhaps several) and gave a presentation to the higher-ups. The higher-ups themselves were unimpressed, but the wives of some of them were there, and as former secretaries, they were very impressed. This made the higher-ups think the thing was just a glorified typewriter, so they were even less interested.

    (This might be entirely apocryphal, but as I said, the guy who told me the story worked at PARC.)

  8. Gregory Kusnick said,

    May 22, 2023 @ 11:06 am

    Not entirely apocryphal. My version goes like this:

    Around 1978, a group (including me) split off from PARC with the specific objective of placing prototype systems at various real-world test sites. Xerox HQ in Stamford CT was one such site, as was Jimmy Carter's White House. The Xerox corporate executives, being unskilled at actually using a keyboard (and apparently unwilling to learn), declined to have these systems on their own desks, but agreed to put them on their secretaries' desks as a sort of pro forma gesture of support for the concept. In fact the secretaries loved them (as the PARC people expected they would), but somehow that still wasn't enough to turn the execs into technology evangelists. That lack of vision ultimately led to a diaspora of PARC people to companies like Apple, Microsoft, Adobe, and so forth.

  9. Andrew Shields said,

    May 22, 2023 @ 1:43 pm

    @GregoryKusnick – Thanks for sharing your version of the story! I can see how it might have developed into the version that I heard.

  10. mg said,

    May 22, 2023 @ 4:24 pm

    I was an MIT student in the late '70s and a lot of my friends spent time at PARC during summers or semesters off as well as for jobs after graduation. It was heartbreaking to see the waste of all that brilliant creativity.

  11. John F. Carr said,

    May 26, 2023 @ 2:00 pm

    I heard that secretaries liked even pre-graphical Unix systems for the ease of text editing. No more white-out. No more retyping the whole page to make a minor change.

    Unfortunately until some time in the 1990s the price of anything with a CRT was more than most companies would spend on a secretary. Wikipedia says the VT220 launched at $1,295 in 1983 dollars.

RSS feed for comments on this post