[SIGCIS-Members] Zuse / binary
Mark Priestley
m.priestley at gmail.com
Sun Jul 9 07:58:43 PDT 2023
Hi Johannah - As far as I know the letter hasn't been published, though
there are references to it in the literature. There's a typed copy in
Herman Goldstine's papers at the APS. Goldstine sent von Neumann's original
manuscript (of the First Draft) back to Valentine Bargmann at Princeton, as
JvN was in Los Alamos at the time. I contacted Princeton last year to see
if there was any sign of it in Bargmann's papers. The archivist I was
talking to said no, but did turn up the autograph of the May 8 letter
instead, which was a nice consolation prize!
All the best
Mark
On Sun, 9 Jul 2023 at 14:45, Johannah Rodgers <johannah.rodgers at gmail.com>
wrote:
> Thank you, Mark! Not having been aware of this von Neumann letter, my
> reading of the "First Draft" and of the "EDVAC Progress Report" with regard
> to the decision to use binary notation would indicate that the technical
> affordances of various components far outweighed consideration of human
> usability and legibility. Now that I know about this von Neumann
> letter--is this in the APA archive or published somewhere?--I will look
> forward to integrating that into the mix.
>
> All best,
>
> Johannah
>
> On Sun, Jul 9, 2023 at 9:26 AM Mark Priestley via Members <
> members at lists.sigcis.org> wrote:
>
>> Hi all
>>
>> It's worth remembering that von Neumann mailed the text of the *First
>> Draft* to Herman Goldstine at the end of April 1945, around two months
>> before the June 30 date that we are familiar with from the mimeographed
>> version. On May 8 he followed up with a long letter which included
>> additional material to be added to the report (it never was). A lot of this
>> material concerns input and output: von Neumann was concerned that long
>> binary strings are unfamiliar to operators, and proposed a binary-coded
>> decimal scheme for input. Instructions were to be punched on a
>> special-purpose keyboard with dedicated symbols corresponding to those in
>> EDVAC's order code.
>>
>> Of course, binary-coded input would require translation into binary, so
>> there are trade-offs involved, the effects of which may not have been
>> obvious a priori. Presumably Eckert and Mauchly are just exploring these
>> trade-offs by considering a binary-coded machine in detail - had anyone
>> done this before? They don't seem to have be committed to one or the other
>> ("a binary system may ultimately be used"). But importantly, they are
>> concerned with the overall economy of using the machine, not just with the
>> narrower question of whether binary is in some sense "simpler". So on p71,
>> for example, they write "the ease with which a problem may be arranged for
>> computation and the desired instructions placed on magnetic wire or tape is
>> a very important factor in the design of the EDVAC". They go on to say that
>> it's important that all the symbols used should appear on a standard
>> typewriter - presumably a direct riposte to von Neumann's proposal!
>>
>> Best wishes
>> Mark
>>
>>
>> Dr Mark Priestley
>> www.markpriestley.net
>>
>> Senior Research Fellow
>> The National Museum of Computing
>> Bletchley Park, UK
>> https://www.tnmoc.org <http://www.markpriestley.net>
>>
>> On Sun, 9 Jul 2023 at 05:19, Brian E Carpenter via Members <
>> members at lists.sigcis.org> wrote:
>>
>>> On 09-Jul-23 05:23, Johannah Rodgers via Members wrote:
>>> > Thanks everyone for this fascinating discussion. While I cannot offer
>>> any contribution to Zuse's decision to use binary notation, I do have an
>>> opinion to offer re: the decision to use binary rather than decimal
>>> notation with the EDVAC. In their 1945 report entitled "Automatic High
>>> Speed Computing: A Progress Report on the EDVAC," Eckert and Mauchly write
>>> that "accuracy is ensured by digital methods" (6) but, there is also an
>>> indication that the reliability of vacuum tubes is greatly increased by the
>>> use of "digital methods."
>>>
>>> Thanks for reminding me of that report. On the binary v. decimal point,
>>> it is interesting to contrast it with the von Neumann report. The progress
>>> report says:
>>>
>>> "In the EDVAC, the various digits specifying a number will be transmitted
>>> serially through a single channel. Switching equipment is consequently
>>> simplified and reduced. Moreover, the storage or memory elements to be
>>> used in the EDVAC are entirely different from those used in the ENIAC."
>>> [page 4]
>>>
>>> "Indeed, the EDVAC will store its operating instruc-
>>> tions in exactly the same manner as it stores numbers." [page 5]
>>>
>>> "For the purpose of this report, it was decided to present plans based on
>>> use of a coded decimal number representation, although a binary system
>>> may
>>> ultimately be used. The computer elements are therefore somewhat more
>>> com-
>>> plicated than they would be for a binary machine." [page 8]
>>>
>>> That was dated almost exactly 3 months later than the von Neumann
>>> report, which clearly states that the machine should be binary. An
>>> interesting discrepancy. I wonder whether the progress report was aimed at
>>> a somewhat conservative management chain? Or is this simply a reflection of
>>> von Neumann's status as a brilliant outsider?
>>>
>>> Regards,
>>> Brian Carpenter
>>>
>>> >
>>> > All best,
>>> >
>>> > Johannah
>>> >
>>> > On Sat, Jul 8, 2023 at 12:43 PM Ceruzzi, Paul via Members <
>>> members at lists.sigcis.org <mailto:members at lists.sigcis.org>> wrote:
>>> >
>>> > Fascinating discussion. Zuse told me that he was taught binary
>>> arithmetic in grade school, but he only learned of Leibniz's essay on
>>> binary (ca. 1680) after Zuse began work on computers. Babbage was an
>>> admirer of Leibniz's work but did not consider binary for his engines. The
>>> Z1, which had a binary calculating unit, did not work well, and therefore
>>> Zuse adopted electromagnetic relays (a technology not available to Babbage)
>>> for his later machines, which did work. Attempts to reconstruct the Z1 for
>>> the Deutsches Technikmuseum using late 20th-Century machine tools also
>>> failed -- please correct me if I am mistaken. But note that Zuse built a
>>> successful mecahnical binary /memory/ unit for the Z4, which not only
>>> worked well but which was in steady service at the ETH in Switzerland for
>>> years. The joke was that the clicking of the Z4 relays was the loudest
>>> noise in Zurich on any given evening. The Z4 is now at the Deutsches Museum
>>> in Munich.
>>> >
>>> > Paul Ceruzzi
>>> >
>>> ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
>>> > *From:* Members <members-bounces at lists.sigcis.org <mailto:
>>> members-bounces at lists.sigcis.org>> on behalf of Jesper Juul via Members
>>> <members at lists.sigcis.org <mailto:members at lists.sigcis.org>>
>>> > *Sent:* Saturday, July 8, 2023 8:05 AM
>>> > *To:* Evan Koblentz <evank at njit.edu <mailto:evank at njit.edu>>
>>> > *Cc:* members at lists.sigcis.org <mailto:members at lists.sigcis.org> <
>>> members at lists.sigcis.org <mailto:members at lists.sigcis.org>>
>>> > *Subject:* Re: [SIGCIS-Members] Zuse / binary
>>> >
>>> > *External Email - Exercise Caution*
>>> >
>>> > In Raúl Rojas' paper on the Z1/Z3(1), he writes that "instead of
>>> using gears (as Babbage had done in the previous century), Zuse implemented
>>> logical and arithmetical operations using sliding metallic rods. The rods
>>> could move in only one of two directions (forward or backward) and
>>> therefore were appropriate for a binary machine." The Z1 is 1936-38.
>>> >
>>> > This doesn't quite explain if rods or binary arithmetic came
>>> first, but there is a footnote to Zuse's book "Der Computer mein
>>> Lebenswerk".
>>> >
>>> > As a side note, I can recommend seeing the Z1 in the Technical
>>> Museum in Berlin if you are near.
>>> >
>>> > Jesper Juul
>>> >
>>> > 1) https://ed-thelen.org/comp-hist/Zuse_Z1_and_Z3.pdf <
>>> https://ed-thelen.org/comp-hist/Zuse_Z1_and_Z3.pdf>
>>> >
>>> > On Fri, 7 Jul 2023 at 18:34, Evan Koblentz via Members <
>>> members at lists.sigcis.org <mailto:members at lists.sigcis.org>> wrote:
>>> >
>>> > What are some good articles addressing why Zuse, Aiken,
>>> Stibitz, etc.
>>> > decided to use binary (or not)?
>>> >
>>> > --
>>> > Evan Koblentz
>>> >
>>> > New Jersey Institute of Technology
>>> > - Senior Writer, Office of Communications and Marketing
>>> > - Adjunct Instructor, Ying Wu College of Computing
>>> > - Faculty/Staff Advisor, NJIT Lego Club
>>> >
>>> > evank at njit.edu <mailto:evank at njit.edu>
>>> > (973) 596-3065
>>> > https://web.njit.edu/~evank <https://web.njit.edu/~evank>
>>> >
>>> > _______________________________________________
>>> > This email is relayed from members at sigcis.org <
>>> http://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/ <
>>> 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 <
>>> http://lists.sigcis.org/listinfo.cgi/members-sigcis.org>
>>> >
>>> > _______________________________________________
>>> > This email is relayed from members at sigcis.org <
>>> http://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/ <
>>> 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 <
>>> http://lists.sigcis.org/listinfo.cgi/members-sigcis.org>
>>> >
>>> >
>>> >
>>> > --
>>> > johannahrodgers at gmail.com <mailto:johannahrodgers at gmail.com>
>>> > www.johannahrodgers.net <http://www.johannahrodgers.net>
>>> >
>>> >
>>> > _______________________________________________
>>> > 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
>>> _______________________________________________
>>> 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
>>>
>> _______________________________________________
>> 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
>>
>
>
> --
> johannahrodgers at gmail.com
> www.johannahrodgers.net
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.sigcis.org/pipermail/members-sigcis.org/attachments/20230709/cb863b6d/attachment.htm>
More information about the Members
mailing list