I was working at PARC when Pike was apparently there and his description of the infrastructure isn’t really correct.
There were a mix of D machines, not just Dorados (which were ECL machines that needed machine room cooling, as well as the earlier generation Dolphin machines, both using the patch panel setup he described. Also Dandelions (sold as the Xerox Star) and a few Altos. The D machines could be booted into any of the three primary environments (Interlisp, Smalltalk, or Cedar/Mesa). I never touched a disk pack (did the Dorados even have them? I think the Dolphins did. The Dandelions had 8 inch floppies — I still have a set with some custom microcode I wrote)
We had a distributed network mail system (grapevine), distributed networked filesystem you mounted directly,etc. The way he describes editing a document it sounded like he was either using an Alto or booted one of these machines in Alto mode.
No terminal rooms? Well yes, this was personal computing, not timesharing.
The speed issues he described suggest he wasn’t using a dorado. My job used Interlisp-D but I did boot into Smalltalk and Cedar sometimes to check them out. The dolphins in particular were particularly underpowered.
It’s often hard to evaluate something on its own terms rather than on the terms you are used to.
What Rob wants is a personal computing environment that isn't restricted to a singular personal computer. I use plan 9 and the concept of your ENTIRE computing environment being portable across any number of machines, at the OS level, is priceless.
Which was actually the computing environment of the early 1980s, at places like PARC, MIT (Athena), CMU (don’t remember the name of their system) as well as commercial systems like Apollo. None of which he ever mentions when touting Plan 9.
After 15 years of computing I was shocked when I realized that Sun machines kept all their data locally, ran sendmail etc. It seemed like such a huge step backwards in time.
Our research group used Apollo's for several years and the feeling that they were all one big system (which never failed) was wonderful. Not being pure Unix was the main drawback.
Hobby use. Though I do have a little nuc at work running 9front to write small bits of automation (IEC 61131-3, ST) and a Basic dialect of CNC RS-274 G code in acme and a little Sam here and there. I'm slowly learning it. I also write a little c code as the plan 9 c library and tooling is really nice. And I love the concurrency library, thread(2). I access the little server via drawterm from both a Linux machine and a Windows 10 machine. I can sit at either computer and have access to the same user files, configuration, etc. And drawterm gives the plan 9 session access to the host (terminal in this case) resources including the TCP stack (you can do really cool routing and tunneling tricks with this method)
I highly recommend giving it a look (I highly recommend 9front), and read nemo's intro.
*The nice thing about the Alto is that it doesn't run any faster at night*
This was when timesharing machines were super-slow during the day, so if you wanted decent response time, you had to come in at night. The more things change...
Well, that's not actually what he wrote. You might want to read that section again. He said his dream setup is simpler than carrying three computers, cameras, an ipod, and other oddments. What's wrong with that?
To achieve that end, he essentially says he wants all his resources somewhere else (hosted by a third party who does the hard work of keeping it online and backed up, etc) and to interact with those resources through a uniform interface via thin-clients that only use local storage and processing power as a cache.
If you still owned all the hardware in this picture, you could argue it's still "personal computing", but once you've offloaded the processing and storage to a third party, it's more "timesharing on a mainframe" than "personal computer".
There are many advantages to this approach, but it does mean giving up something.
I left Xerox May 1983 (from SDD, not PARC), and never had a Dorado (they were reputed to be so big and hot that you didn't want one in your office). You can read what it was like, with hindsight NEVER allowed to the characters, in my book [1].
This is a remarkably detailed critique of one very specific subset of the world of PARC 1984. People in the rest of PARC, and in SDD, were using Cedar, Tajo, and the Mesa Development Environment, not Smalltalk, and not on Dorados, which added up to a much, much different experience than Rob describes.
His last paragraph begins "A few years ago, PARC probably had most of the good ideas. But I don’t think they ran far enough with them, and they didn’t take in enough new ones." is spot on.
There was some point, well before 1984, where Xerox should have recognized that the rest of the world had not been standing still. Expect a series of posts soon where hindsight IS allowed and we look at what Xerox should have done.
Your book has been on my wish list for a while, I just placed the order. This is a romanticized period of technological development. I was born a year after you left Xerox. In a lot of ways 2021 is my 1983. I look forward to the hindsight in your upcoming posts.
Just ordered your book not been so excited to read computer history book since the folklore book about the Macintosh. Thank you for taking the time to document such a pivotal moment.
Thank YOU. I hope you don't expect it to be nothing but computer history! There are characters in there who don't spend every waking minute thinking about work.
I visited Xerox PARC in the 1980s. I was already coding on a Macintosh of some kind, pre-Mac II (color), and in Postscript. The topic of the day for me was print halftoning algorithms. The researcher I spoke with was patient, and interested in new ideas.
It is hard to describe the different information environment at that time, as off-beat print, major newspapers, television and film were active and very influential, yet so different in practice than today.
I came away from my visit to Xerox PARC with a sense of participation despite being self-trained, and respect for those working there (I was pretty young); good times and inspirational overall.
> The type structure is somewhat polymorphic, but breaks down. For example, rectangles cannot have both scalars and points added to them, because the message is resolved by the left operand of the message, not the left and right.
This seems very mildly ironic coming from Rob Pike, given Go’s stance on function overloading.
There were a mix of D machines, not just Dorados (which were ECL machines that needed machine room cooling, as well as the earlier generation Dolphin machines, both using the patch panel setup he described. Also Dandelions (sold as the Xerox Star) and a few Altos. The D machines could be booted into any of the three primary environments (Interlisp, Smalltalk, or Cedar/Mesa). I never touched a disk pack (did the Dorados even have them? I think the Dolphins did. The Dandelions had 8 inch floppies — I still have a set with some custom microcode I wrote)
We had a distributed network mail system (grapevine), distributed networked filesystem you mounted directly,etc. The way he describes editing a document it sounded like he was either using an Alto or booted one of these machines in Alto mode.
No terminal rooms? Well yes, this was personal computing, not timesharing.
The speed issues he described suggest he wasn’t using a dorado. My job used Interlisp-D but I did boot into Smalltalk and Cedar sometimes to check them out. The dolphins in particular were particularly underpowered.
It’s often hard to evaluate something on its own terms rather than on the terms you are used to.