Thank you all for an enlightening discussion around the use of diagrammatic approaches. This
leads to some comments and followup questions:
(1) What may have proceeded diagrams? I would think natural language descriptions, tables,
and log books, as suggested. Without a diagram, the processes involved have to be gleaned in
these more traditional formats. Creating a modern diagram is possible but only with a good deal
of research. Apart from Babbage's remarkable diagram (and the 20th tradition of charting as described
in this thread), the practice of "formalizing the process" (of human computing), even in pure natural
language, seems not commonplace.
(2) Brian: you quote Swade's article, which I found most informative, but I had a question. If one
(Mechanical notation), there is a citation labeled 18. This citation is for this article:
18 Babbage, Charles. ‘On the Determination of the General Term of a New Class of Infinite Series.’ Transactions of the Cambridge Philosophical Society 2 (1826): 217-25. Reprinted in Works, Vol. 2, pp. 61-8. See p. 62.
But if one goes to
archive.org, and reads this article (pp. 217-25), the referenced illustration is not present. Here
I might find the paper in the appendix, I also searched there. Do you know where it is published?
(3) Paul: I also like Richardson's book. Also, the quote that you include is a good example of how some, like Richardson,
describe the human process in natural language without the use of tables, lists, diagrams, etc.
(4) David H.: I also have that one but nothing earlier on the use of symbols or diagrams for human interaction for
business or industrial processes. Do you know of other, earlier examples? Perhaps David G.'s suggestion of Leffingwell
is a good place to begin. I'll do a search.
(5) As for Pickering and the Harvard Computers, I believe I have enough from videos, scholarly publications, and online
resources to fabricate an information workflow model consisting of (a) the act of an astronomer taking one of two types of
plates: optical vs. spectral, (b) a cabinet where the glass plates were stored, (c) the explanations of the key women who
made it all work, including the interesting task of overlaying plates to search for Cepheid variables. The information flows are
all there in the historical ether, and only remain to be drawn.
As to why I am doing all of this, it is in preparation for a short lecture on computing, and contextualizing computing within
culture and history. By doing so, to illustrate computing as information science and management (human
computing provides an excellent resource).
-paul
On Apr 19, 2015, at 12:35 PM, Paul N. Edwards <
pne@umich.edu> wrote:
Very interesting.
This calls to mind some period diagrams of computational flows in the pioneering weather forecasts run on the ENIAC. I often show these in talks because the flow included both pre- and post-processing phases done by human beings - e.g. “prepare punch card deck for step 4,” on p. 309 of
Platzman, G. W. 1979. “The ENIAC Computations of 1950 — Gateway to Numerical Weather Prediction.” Bulletin of the American Meteorological Society 60 (4): 302-12.
Another good flow diagram, this time for analysis of incoming weather data, is on p. 334 of Bedient, H. A., and G. P. Cressman. 1957. “An Experiment in Automatic Data Processing.” Monthly Weather Review 85 (10): 333-40.
Best,
Paul
(Apologies if this ends up appearing twice on the list, but I think I may have sent it to the wrong address earlier).
Hello everybody,
One aspect that
hasn't (to the best of my knowledge) been mentioned in discussions of
the origins of the von Neumann/Goldstine notion of flowcharting is the
work that was done in visualizing computational flow as part of the
ENIAC project. (The following comments draw heavily on work done with
Tom Haigh and Crispin Rope as part of the
"ENIAC in Action" project.)
The
famous "plugging and switching" of the ENIAC allowed basic sequences of
operations to be set up (and a bit more than that, but that's a detail)
but the team soon realized - by the end of 1943 - that they would need
ways of combining sequences into structures of arbitrary complexity,
involving repetition and conditional branching. During 1944 they
investigated various ways of doing this "sequence programming" (their
term) before settling on the design of ENIAC's master programmer, the
central unit where loops and branching were controlled. Program
structures were illustrated using "master programmer diagrams" which
showed the basic sequences as unanalyzed blocks linked by simplified
representations of the settings of the master programmer's "steppers".
These diagrams first appeared in public in a report written for the
Applied Mathematics Panel in November 1945 and were later used, for
example, by Douglas Hartree to document his 1946 ENIAC program.
So
the ENIAC modeled general computational flow using complex "switches"
(the steppers) that both counted iterations and provided multi-way
branching. The EDVAC design & code of spring 1945 reduced this
complexity to the more primitive notions of binary decisions and address
modification. I suggest that when they came to try to visualize
programs in 1946/7, von Neumann and Goldstine (who clearly would have
been familiar with the ENIAC notation) performed a similar reduction,
replacing the steppers on the master programming diagrams with the
"alternative boxes" of their flow diagram notation. If you do this, you
have the basic flow diagram notation: vN+G's "operation
boxes ... where no branching or
merger takes place" are functionally equivalent to the basic sequences
appearing on the master programmer diagrams.
So I don't
think it's necessary to look as far afield as chemical engineering or
process modeling to see where the flow diagram notation came from. We
have not found the term "flow diagram" itself anywhere in the ENIAC
literature, however.
Best wishes
Mark
_______________________________________________
This email is relayed from members at
sigcis.org, the email discussion list of SHOT SIGCIS. Opinions expressed here are those of the member posting and are not reviewed, edited, or endorsed by SIGCIS. The list archives are at
http://lists.sigcis.org/pipermail/members-sigcis.org/ and you can change your subscription options at
http://lists.sigcis.org/listinfo.cgi/members-sigcis.org
___________________________
Paul N. Edwards
Terse replies are
deliberate (and better than nothing)
4437 North Quad
105 S. State Street
Ann Arbor, MI 48109-1285
(734) 764-2617 (office)
(206) 337-1523 (fax)
This email is relayed from members at sigcis.org, the email discussion list of SHOT SIGCIS. Opinions expressed here are those of the member posting and are not reviewed, edited, or endorsed by SIGCIS. The list archives are at http://lists.sigcis.org/pipermail/members-sigcis.org/ and you can change your subscription options at http://lists.sigcis.org/listinfo.cgi/members-sigcis.org
Paul Fishwick, PhD
Chair, ACM SIGSIM
Distinguished University Chair of Arts & Technology
and Professor of Computer Science
Director, Creative Automata Laboratory
The University of Texas at Dallas
Arts & Technology
800 West Campbell Road, AT10
Richardson, TX 75080-3021