daemon@ucbvax.BERKELEY.EDU (The devil himself) (01/24/86)
I attended this conference because I was fortunate to have worked on
a several of these systems: ARPAnet (1973 on, at UCSB [3rd node]) the
Culler-Fried OLS, Altos (1978 at Caltech), and of course H-P
calculators. There appears to be a lot of relevance to many news
groups: net.cog-eng, net.lang, net.nlang, net.arch, net.unix and so
forth, but I will only post to net.works and net.text.
The purpose of the conference is to make some type of historical
record of the history of computing. Several years ago there was a
similar conference on the history of programing languages. The next
such conferene will be held later this year on this history of
scientific computing [to be held as LANL]. The Conference was video
taped and a book is forthcoming.
Gordon Bell (keynote) - Towards a History of Personal Workstaations
Bell develops his memory versus cost curves: this was covered a tad
in the Portland Usenix meeting and in Science. There is some "fear"
that workstations something of a soon to die breed like 4-function
calculators. Noted were "The era of timesharing has ended..." in the
late 1970s. Bell described developments at PARC, Bell Labs, Apollo,
and other sites (IBM notably missing). Special note: Bell described
on paper, for the first time, his vision of the VAX homogeneous
environment call E consisting of the following bus components: DEC E
(Environment) CI -- Computer Interconnect, CI780, HSC-50 NI --
Network Interconnect, Ethernet BI -- Bus Interconnect, ? II --
Interchip Interconnect, never made
Doug Ross -- SofTech/MIT
The distinction of Work Stations versus Workstations is important to
keep in mind. Ross described the early SAGE system which may have
had the first Work Stations. The nature of Work must be analysed and
remembered. Work Stations are not just computers. The light gun and
light cannon were described before the light pen. The light cannon
became the basis for the earliest computer graphics when Ross got the
idea of using the feedback tracking on the light cannon's CRT to
trace out his signature. The ideas of programming languages were
new, so Ross and others developed Gestalt programming based on the
idea of a conversation. The Whirlwind was described in some detail
[it now sits in the Smithsonian]. Ross gave Bush's Memex paper as a
classic for all to read about workstations, but also gave a lesser
known paper by G. Price in Fortune [1956, pp. 150] on How to Speed
Invention.
J.C.R. Licklider reflected on early history and spoke in generalities
about the issues: analogy versus digital, various I/O devices, and
software.
Larry Roberts spoke about the early ARPAnet. Noteable was that IBM
and CDC both thought that a heterogeneous wide-area network of
computer would be prohibitively expensive (impossible). Roberts
described early justifications, and successes as well as failings.
Glen Culler described Culler-Fried system for scientific computing.
The emphasis again was to make a system for non-computer people,
physicists, to use without really needing to know the internal grit.
No control-keys were evident on the first system. Instead a two
keyboard system was chosen using fixed function keys like SIN. A
signal processing application was illustrated.
Doug Englebardt talked about the Augmentation Research Center. He
brought the first (wooden) mouse constructed by Bill English. In a
joke, (he described) how a person walked up to him, picked the mouse
up by the tail, hanging it, saying, "Oh, it's female." He showed his
1969 movie. Slide included a knee-based pointing device, and he
talked about foot-based pointing devices. NLS never really caught
on. The paper and the discussion pointed to several vaguely
illustrative failures. Many people came up to Doug afterward and
expressed their surpise that NLS did not catch fire, but it had
contributed greatly to the concept of workstations. This is one of
the beginnings of WYSIWYG and the integration of text and figures.
Alan Kay spoke for three hours as the dinner time entertainment. Kay
elaborated on his initial work on the FLEX and the Reactive Engine.
He had admiration for several early minicomputers such as the DG Nova
and mainframes such as the 6600. He gave his vision of the Dynabook,
the problems of the Mac/Lisa. He showed his Inner Tennis movie and
well as clips of Adele Goldberg (ponytails) and other children
working with Smalltalk. He had good words to say about the Alto
(which many regard now as slow).
Chuck Thacker and Butler Lampson spoke on the Alto, hardware and
software. The first Altos had 128 KB memory which was inadequate.
The small address space (512KB) as well as the development of micros
out paced the Alto, but the Alto with it's 170 ns micro cycle was
quite fast for its day. The Alto was a derviative of the DG Nova and
one of the distant descendents was Rob Pike's Blt from Ingall's
BitBlt which in turn came from the DG [all noted in the paper]. This
is pretty cursory. Other descendents of the Alto at PARC were
described: the Dorado, the Dolphin, the Dandelion, the Star, Mac,
LISA, and a later paper by Stu Card has a picture of a DandeIris.
The software covered editors, print servers, drawing programs,
languages, and themes of software. In some ways, I was disappointed
there was not more.
The LINC was described by Wesley Clark. The LINC was a project in
"portable" laboratory process control. The work was funded by
NIH/NASA. Lots of interesting pictures of assembly and one of a cat
hooked up to an experiment via electrodes. The first DECtapes can
easily be seen.
Chuck House described H-P developments. H-P development of
calculators from the HP9100A to the HP-35. This covered individual
development of components such as CRTs. No individuals distinguished
themselves, the author pointed out, because that is the nature of the
H-P organization. One developer was in the audience, and he was a
virtual unknown compared to Gordon Bell and others.
I had to leave for a plane to Usenix, so I missed the description of
PROMIS. This appears to be another medically funded project. There
are some interesting pictures of equipment (50 MB of disk taking huge
boxes...)
The last paper was Stu Card of PARC on human factors work in pointing
and pondering. There is lots of empirical work at PARC which appears
not used by most. The pondering is more difficult. They have
appeared to have bogged down in the same area as Englebardt.
I've left lots out, the proceedings are available from the ACM in NY.
From the Rock of Ages Home for Retired Hackers:
--eugene miya
NASA Ames Research Center
{hplabs,ihnp4,dual,hao,decwrl,allegra}!ames!aurora!eugene
emiya@ames-nas.ARPA