Xerox PARC – Office Alto Commercial (1979) [video]
youtube.com
youtube.com
Over lunch a few years later Alan told me about trying to get Xerox to sell Altos. There was a bake-off back in NY between the Alto and a orange-screen fixed-character machine (perhaps running CPM?) with no mouse, no network, no bitmapped display, just a small screen. No laser printing. Anyway the latter won.
A few years later Alan was on a plane and when talking to his seat mate found they’d both worked for Xerox — in fact that guy had made the decision or at least was in on it! So Alan asked how the decision had been made and the guy shrugged and said “well they were basically the same, so we went with the one that was cheaper to make”.
> “well they were basically the same, so we went with the one that was cheaper to make”
1 - in those days executives, and even other white collar “professionals”, typically male, didn’t type. They had secretaries who took dictation, typed things up, made corrections, and then used, Xerox photocopiers to make copies which would be distributed. So the use case wasn't clear.
2 - for all of those reasons they probably looked at specs (screen? Yep. keyboard? Yep. Network? What's that?) and thought they were making a shrewd decision. And perhaps they were: nobody even knew they needed a WYSIWYG editor much less mail (and this was Xerox, a paper management company -- why would they embrace electronic mail?). Microsoft swooped up Simonyi and he rewrote his Bravo editor as the original version of Word (the only part of office originally written by Microsoft?). The Alto wasn't as expensive as the later Star (also a failure) but I think each one cost about 10K in 1980 dollars.
As far as I know the only Altos and Dovers outside PARC were MIT, Stanford and the White House. Probably CMU too.
What got my attention was how slowly text was painted on the screen, in the ad. I wonder if that was artificial. People used to be weirded out by computers being "too fast". They thought it must be taking a shortcut and not processing everything it ought to. Coders had to insert delay loops so it would look like something was happening. Yes, really true.
My employer in 1982 bought a Xerox CP/M machine, thinking it ought to have some PARC mojo. Probably that one.
If you read Butler Lampson's paper on the design of the Alto you'll see it was an amazing machine. Super screen-oriented (i.e. human-oriented): the bus clock was 3/2 the screen refresh speed, so the majority of the bus bandwidth was spent refreshing the screen! Absurd and crazy at its time, but of course now we realize that it was brilliant. All built out of a handful of MSI and SSI TTL parts IIRC.
Actually the paper is by Thacker, McCreight, Lampson and Sproull. Really a great read.
As much as people complained about its performance, I would have been proud to have designed it.
I'm sure you don't believe that, if you've even heard of that company. They were huge in the 70s, a public company, and they had a terminal that did not require a mainframe connection, i.e. a PC.
Their concept of a PC, though, was "You can do your data entry offline, and then when you're connected, you upload the data! It's magic!" No third-party apps, no "apps" of any kind, and of course not for retail sales.
They also had a LAN, ARCNet, which was a 4MB token ring that ran over regular twisted pair. It was solid, too.
I've probably still got some ISA arcnet cards kicking around in a box someplace.
Apple, meanwhile, had AppleTalk over RS-422. Was it 1/2 Mbps?
Arcnet was a token bus (like IEEE 802.4 from General Motors' MAP) and not a token ring (like IEEE 802.5 from IBM's Zurich research).
They were too far ahead of their time. The component costs needed to come down first.
In 1981, the "starter kit" for their office system consisted of one Xerox Star for a user, a second Star to act as a file and print server, and a laser printer for about a quarter of a million dollars in today's money.
Each additional networked Star that you added was about $50,000 in today's money.
Even at those prices, the system wasn't exactly snappy. I do wonder what would have happened if they had kept iterating on the product until faster, cheaper components arrived.
The IBM story is a bit of an outlier, but Microsoft saw the same potential Apple did.
So the Alto was never properly productised. It was essentially a boutique product at a big-mini price. That made it competitive with DEC and IBM minis, but not with budget the $10k S100 systems that were just about - barely - powerful enough for a small business.
To be fair to Xerox, six figure systems weren't unusual for business customers at that time. Which could be why the company never understood that it had the key to popular budget semi- and non-pro commodity computing.
DEC died for much the same reason. DEC management were used to selling to fellow nerds and business people, and couldn't imagine super-affordable commodity products.
MS, Apple, and the IBM clone makers saw the new market for what it was. So did some of the new software houses.
IBM itself almost did, but couldn't quite shake its monopoly-oriented culture. So MS and the clones killed it in the commodity space.
Things we expect to have dedicated hardware for -- network interface, display screen painter, disk interface, laser printer driver -- were just interrupt routines all on the one CPU. Programs competed for CPU time with feeding pixels out to the CRT-scanning electron beam, and with bits moving directly to and from disk-head R/W coils.
It did switch between those tasks pretty quickly, not like most machines then or today: there was no foolishness about saving and restoring registers. Each interrupt task was assigned a couple of registers permanently, including its own program counter, so an interrupt only ever just flipped which program counter was in charge.
The app-level instruction interpreter got to use what few registers remained.
So, on an interrupt, it would execute a half-dozen instructions to move one machine word, and yield. There were 16 PCs that ran in strict priority order, with the user-program instruction interpreter last in line. Each "yield" put the current PC back to the entry point for the next word to be processed, and the next live lower-priority PC determined the instruction to run next.
The OS was just one of those user programs.
It resembled the I/O processor on CDC mainframes of the time. It just ran the programs, too, not only the I/O.
But it really is amazing what they got it to do. Later machines e.g. Dorado and Dolphin had more stuff done in hardware so could be faster.
What they couldn't wrap their heads around was: no one cared anymore. The world had moved on.
Just Teletype machines and then "glass teletypes": a CRT terminal that merely mimicked a Teletype with scrolling text, but with the disadvantage that if something scrolled off the screen you couldn't reach behind it and pull up the pile of paper to look at your old code. It was gone.
They were quiet though.
Later I discovered a "programmable terminal" in a side office/storage room that no one used. It might have been a Datapoint 2200 or something similar. Whatever it was, it had a processor and a character mode screen buffer that you could address directly.
Naturally, I immediately wrote Conway's Game of Life for it in assembly language.
When my manager saw this, he was fairly gentle: "Mike, that must have been fun to program. But we have business needs. Can you please forget about this useless machine and get back to work?"
I had more success with another programmable terminal: a Texas Instruments DS990/1. I found this in the back room of our Texas office, after I flew down to design and implement our usual Teletype "print a question, you type in the answer" interface. Then I saw that I could put the entire UI on an interactive screen, and stayed up all night implementing it.
Our sales team and customers loved it, so this one made us some money.
Not long after that, we hired Douglas Engelbart. Yes, that Douglas Engelbart.
Obviously, someone in the company had taken an interest in what Doug was doing, but he never got any resources to build his vision of the future.
I worked in the same building and often walked by his corner cubicle. With my own interest in more human-oriented interfaces, I sometimes wonder what might have happened if I'd introduced myself. But I was a shy young geek and walked on by.
Sometimes it's just the wrong place, at the wrong time.
https://www.computerhistory.org/collections/catalog/10265798...
They had no idea what to do with this stuff and no real desire to understand it, nor sell it.
If it'd been 1/10th of the price, that probably wouldn't have mattered too much, but...
Plus Clayton Christensen’s innovator’s dilemma.
The arrival of Visicalc transformed the toy personal computers into legitimate business tools and set Apple on its course for what would be the largest IPO in US history up to that point. No one saw it coming until it happened.
Once it happened, the Xerox brass in New York question how they missed it, and start to demand a return on the money they've been pouring into PARC. This is what led them to seek a partnership with Apple and the infamous Alto demos for Steve Jobs.
Meanwhile, the PARC folks start to realize that the rest of this now exponentially expanding industry is going to catch up to what they're doing faster than they can scale down their stuff into a commercially viable product.
Xerox takes a huge swing with the Xerox Star. It's a market failure for a number of reasons. IBM takes a little swing with the PC, and it's a huge success, but they don't control the software. The PARC folks leave to join the rest of the industry, taking what they learned with them.
Apple Lisa was $10,000. Same, otherwise. ("Clascal", compiled to bytecodes interpreted by 5 MHz 68k.)
Mac was less, but a toy. A little later, Mac 512 (i.e. 1/2MB RAM) was kinda usable, with a pair of 140kB floppies. $2000? Had a monochrome 1bpp paint program, no WP, no spreadsheet. You bought MacWrite separately ($125?), and Multiplan (a proto-Excel) from MS.
The LaserWriter came out pretty quickly after, making it radically more useful. That had a computer inside much beefier than the Mac driving it; they could have put a keyboard, mouse, and monitor connector on the Laserwriter and had a much more useful product much sooner.
For B-school analysis, this isn't your book, but if you want to know what it was like to live through it without knowing the outcome, this is it.
I really tried to get the numbers to back up Dave Liddle's assertion that Xerox made enough from laser printing, all by itself, to make up for all the costs of PARC and commercialization. That data's long gone. Anyhow, they did commercialize laser printing pretty effectively, although obviously not well enough.
https://www.nytimes.com/1972/12/13/archives/xerox-monopoly-o...
There IS a connection, though: Xerox settled the suit by licensing its patents, which led management to think "holy shit, we're losing the monopoly, we'd better do something" and start up PARC.
Really? I'm not saying it didn't happen, but I don't recall that.
“Following these years of record profits, in 1975, Xerox resolved an anti-trust suit with the United States Federal Trade Commission (FTC), which at the time was under the direction of Frederic M. Scherer. The Xerox consent decree resulted in the forced licensing of the company's entire patent portfolio, mainly to Japanese competitors. Within four years of the consent decree, Xerox's share of the U.S. copier market dropped from nearly 100% to less than 14%.[31]”
Wikipedia page: https://en.wikipedia.org/wiki/Knowledge_Navigator
At that time, being seen actually using anything technical -- most especially, typing -- would have made him a clerk, working-class, and not white-collar enough for anybody making budget choices to identify with.
There is no middle class anymore.
Mid-range and higher monitors often have stands that allow for 90 degree rotation of the screen, so the wide screens could shift to vertical; sadly a good way to openly communicate that back to the GPU and OS hasn't been a priority.
I think there was some Swedish Unix 68k system that had a pivot CRT rather early on, too.
This 1979 Xerox innovative white-screen computer commercial, pitching the corporate executive eating a pastry at desk while forwarding memos to the different divisions.
The 1984 Apple innovative white-screen computer commercial pitches very differently. (And Apple's message actually more in line with what I've read that PARC researchers were interested in.)