Loading ...
Sorry, an error occurred while loading the content.

Re: [midatlanticretro] Wired article about John Vincent Atanasoff

Expand Messages
  • Dan Roganti
    I m not one to begin discounting the work of Mauchly and Eckert, no less that of Atanasoff Berry or even Babbage. But when it comes to history of any
    Message 1 of 23 , Nov 7, 2010
    • 0 Attachment
      I'm not one to begin discounting the work of Mauchly and Eckert,  no less that of Atanasoff Berry or even Babbage. But when it comes to history of any engineering/scientific accomplishments, it's rather important to grasp the underlying fundamentals of electronics and computers when attempting to rationalize the differences.

      I think it's known already what were the limitations of the ABC computer. It was not a Stored Program computer - ie. general purpose progammable computer which is capable of storing the instructions and data in a random access memory unit. But neither was the ENIAC -ie. programmed via patch cables versus their 2nd computer the EDVAC, which was. Although the ENIAC _was_ a Turing complete machine and the ABC wasn't - but Alan Turing only conceived this idea in his paper not much earlier than Atanasoff was designing his computer in 1937 to even begin engineering a way to implement Turing's ideas.

      On Sun, Nov 7, 2010 at 9:13 PM, Evan Koblentz <evan@...> wrote:


      I finished reading the book.  It is terrible.  Even the staff of Wired magazine mocked the author in a recent interview: http://www.wired.com/magazine/2010/11/mf_smiley/all/1

      In addition to be an Iowa alum, the author is an acclaimed * novelist * so she knows how to tell a great story.  However non-fiction is not her forte.  All of her "sources" were fellow Iowa people.


      It's one thing to find authors who may have some lack of comprehension on electronics or computer design but taking an unbiased look yourself at the technological evidence is paramount.

       

      The ABC * calculated * with tubes -- or at least added and subtracted; it couldn't multiply or divide.  But it had two big catches:


      Back to mathematics now, Linear Algebraic equations make use of multiplication all the time - which the ABC was designed for. The multiply & divide operation is nothing more than a series of additions and subtractions. This is how you solve an equation, it required a series of repetitive addition instructions to perform a multiply operation between each coefficient and constant of the equation.

       
      - Its input and output was mechanical. ENIAC was electronic end-to-end.


      So were many of the early modern computers - in fact the ABC did use the IBM punched cards for data entry and output - typical of the early modern computers. In fact, it made use of the IBM punched cards to enter the equations - which is far more advanced that stringing patch cables all day.

       
      - It has no program. All steps had to be decided and entered by hand.


      Again, this is where some people need to understand the basics. The ABC computed Linear Algebraic equations >automatically< which means that it would access the variables and constants from it's memory unit,  add, subtract, multiply however many times needed to solve the equation. The data entry involved entering complete equations, no patch cables, and the computer decoded this automatically. While each equation was computed automatically, it did still require some manual steps to start the computation for the next equation - which was still automatic - remember this was the late 30's.

      Nobody realizes the manual work it saved in solving a system of linear algebraic equations in physics or statistics - just try doing that on your adding machine. So these steps were not entirely entered by hand, as one would have with a adding machine or calculator of the day. Given the scope of their budget and the time of their innovation, this was still far more advanced that any present day calculator then.

      The point about having no Program means there was no Stored Program unit internally to the Processor which provides the capability of holding the instruction along with the data in memory. This would allow for the CPU to alter the operation, such as, compute a different set of equations. But guess what, neither was the ENIAC, it is not a Stored Program computer, it relied on patch cables. It was not until their next computer the EDVAC which offered this feature.

       

      Smiley (the author) harps on ABC's use of binary.  Was that novel?  No.  Leibniz, IBM punch card machines, etc. all used binary long before the ABC.
       

      The point was not who instituted the Binary number system - which actually dates back over 2 millennia , but who implemented this number system into a electronic machine capable of using this form. Atanasoff & Berry were first to develope >electronic circuits< that implemented calculations using binary number system - they built the first electronic logic gates - that is, AND gate, INVERTER gates, et cetera  - out of electron tubes -  which allowed it to take advantage of Boolean Logic - George Boole who authored Boolean Algebra. That is what is significant about their development.

      And the ENIAC was not even binary, with their massive budget -- just to save tubes - relatively speaking - instead it was decimal based.

      Leibniz didn't theorize/design any hardware(mechanical, relay or electronic) using binary logic but rather the use of the binary number system as alternate method for arithmetic. The fundamental component of IBM punched cards, i.e. a punched card,  is not a electronic device which is what the point is here.



      She says the ABC is special because it's digital.  She implies that was an innovation.  She states that Babbage's and Aiken's machines were analog.  That is wrong.  They were both digital.


      Again, the poor interpretation by this author is not one to base any conclusion on any technological evidence. We know that Babbage designs were mechanical, Aiken designs is Electro-mechanical, ie. Relay and the ABC is electronic.

       

      She implies that Mauchly had something to do with Cliff Berry's murder!  (Bill Mauchly emailed me and noted, "A thief AND and a murderer.  Pretty cool, eh?")


      I'm not going there, it has *nothing* to do with engineering here.

       

      Dan wrote: >>>  performing every function using electronics alone, no relays, gears, etc.

      That's not true, as I explained above.  Only the memory was electronic.
       

      As I stated above, it important when deciphering historical artifacts to grasp the fundamentals of electronics and computer fundamentals rather than basing it on the subjective viewpoints.
      Again, these are not my viewpoints, but rather significant engineering accomplishment that you simply cannot deny.
      Traditional calculators of the day, both mechanical, relay, were not even close to being as advanced as this.
      1. This machine contained a rudimentary form of a C.P.U. - Central Processing Unit - the basis of modern computers, ie. Arithmetic unit and a Memory unit.
      2. The Arithmetic logic unit contained the electronic circuits designed using Logic Gates, ie. AND gates, INVERTER gates, etc.  made with Electron Tubes for all binary calculations.
      3. The Memory unit used a Capacitive Memory drum, again an electronic circuit with a huge array of capacitors, which was the forerunner of DRAM.
      4. Equations were stored in Memory and computed automatically, ie. CPU read the equation and solved the answer.
      5. Data Entry used punch cards  - instead of switches or patch cables - to type in the equations.

       

      >>> it's very possible that AB would have been able to finish a modern computer

      No, it's not.  A-B were not trying to do so, and even if they had tried, they didn't have the idea to do it electronically end-to-end.  Mauchly himself was well documented in asking anyone who'd listen (paraphrasing), "What was Atanasoff thinking; how could he not have seen the obvious solution?"  This is well documented in numerous articles from IEEE Annals of the History of Computing.


      I think you're quoting somebody else here, it wasn't me.
      But the fact is Atanasoff and Berry were both reassigned shortly after this to work on other military projects for WWII.
      Who knows what they could have developed afterwards given the funding.

       

      >>> the work done by the AB team influenced the ENIAC team as the AB computing device pre-dated the ENIAC. Is that reasonable?


      I guess you're quoting the book here.
      This is always going to be subjective, no matter who the author is. But it's very naive to think that somebody's work, however the slight, has never become influential in some manner - especially when the someone was quite aware of the prior art [hehe, pun intended ]

       

      Bill M. also observed in email:

      >>> The ENIAC worked at the electronic speed of 5000 operations a second.  It could do this because the numbers were "stored" in electrons, not paper tape, rotating drums, or punch cards.  And, ALSO, the calculation units all worked at that speed, with no moving parts.  And, ALSO, and perhaps most important, the NEXT operation could proceed immediately, without waiting for paper tapes or human intervention.  It took some smart men, an urgent need, and a lot of money to solve that problem.


      Again, put this into perspective, we're talking about something that was designed almost 10 yrs apart, David vs. Goliath budgets, any design is going to have major improvements and/or enhancements - as in the Square Root unit of the ENIAC - that's the nature of engineering !

      But don't get me wrong, I'm not going to start discounting the work of Mauchly and Eckert, no more than the ABC or Babbage.

       

      >>> Atanasoff never thought about these things in 1940.  He had no secret sauce that he imparted on John Mauchly.  No concepts of high-speed memory or program flow originated in Iowa.  None of Atanasoff's circuits ever informed Eckert or made its way into some ENIAC module.  Atanasoff was not trying to solve the speed problem.  He was not trying to solve the program sequencing problem.  He was trying out concepts for a caculator with capacitive memory instead of relays.  It took mechanical motion to get the numbers in.  It took mechanics to get them out.  It took, actually, a human, to choose the next operation.  It just plain was not a computer.



      Again, it all boils down to comparing apple and oranges between the ABC designed about 10yrs earlier on a modest budget from a grant proposal to the ENIAC designed a decade later on a huge military budget. The same difference on what I would design between working for some two bit company in the past versus when I was working for a conglomerate like Bell Labs.

      =Dan


    • Evan Koblentz
      ... I sincerely hope you do not think I lack this knowledge. Admittedly, I can t make a logic probe work magic. You have that ability, and I admire it. I m a
      Message 2 of 23 , Nov 7, 2010
      • 0 Attachment
        >>> But when it comes to history of any engineering/scientific accomplishments, it's rather important to grasp the underlying fundamentals of electronics and computers when attempting to rationalize the differences.

        I sincerely hope you do not think I lack this knowledge.

        Admittedly, I can't make a logic probe work magic. You have that ability, and I admire it. I'm a historian, not a technician.

        But I certainly "grasp the underlying fundamentals of electronics and computers"...! Not being a hands-on expert doesn't mean one is a dolt. (You're not a professional writer like I am, though I wouldn't assume you can't "grasp the underlying fundamentals of grammar and spelling.")

        Enough about that. Your assessment of the ABC is too caught up in minutiae. True, ENIAC wasn't (at first) a stored-program computer, although it did get that feature after it was moved to Aberdeen. Regardless, the controversy isn't about ENIAC, it's about the ABC. I read the new book (have you?) and I'm saying for fact that its author is wildly misinformed. This is meant to be helpful, so MARCHins don't waste their money on it. (When Chris first posted about it, he didn't know that the ABC-as-first / Mauchly-as-thief arguments were both discounted by serious historians many decades ago.)

        Atanasoff and Berry built an adder and subtractor by using vacuum tubes for memory, and it was limited to a weakest link of mechanical I/O. Mauchly and Eckert built an all-electronic general-purpose computer.

        Moreover, patents are intended to go to whomever actually invents or substantially improves something. Regardless of the extent to which Atanasoff and Berry used now-modern concepts in their device, Mauchly and Eckert did it in an entirely different (and obviously better) way. That alone shows their patent should've stuck. Even if they stole their ideas from Atanasoff and Berry (which they didn't, and that's documented out the proverbial wazoo), they still had to actually build it in the new electronic way. Atanasoff and Berry never even finished their mechanical/electronic hybrid!

        No reasonably informed person could confuse this. Yet the judge in Honeywell v. Sperry Rand did.

        If you read up on the court case (by which I mean reading real history, not Wikipedia), you'll learn that the judge may not have been * entirely * ignorant. He ruled the ENIAC patent invalid for many legal reasons that have absolutely nothing to do with technology nor Mauchly and Eckert. It was mostly about legal moves on the part of bureaucrats at Penn and Remington/Sperry.

        Friendly counter-jab from this lowly English major: when it comes to history of any engineering/scientific accomplishments, it's rather important to grasp the underlying fundamentals of patents and lawsuits when attempting to rationalize the differences. :)
      • system@great-escape.tmesis.com
        ... What was binary about an IBM punch card?
        Message 3 of 23 , Nov 8, 2010
        • 0 Attachment
          Evan Koblentz <evan@...> writes:

          >Smiley (the author) harps on ABC's use of binary. Was that novel? No.
          >Leibniz, IBM punch card machines, etc. all used binary long before the ABC.

          What was binary about an IBM punch card?
        • Dan Roganti
          ... I read the most important thing when it comes to this I read the manuals Schematics don t lie ;) =Dan
          Message 4 of 23 , Nov 8, 2010
          • 0 Attachment
            On Mon, Nov 8, 2010 at 2:19 AM, Evan Koblentz <evan@...> wrote:
            >>> But when it comes to history of any engineering/scientific accomplishments, it's rather important to grasp the underlying fundamentals of electronics and computers when attempting to rationalize the differences.

            I sincerely hope you do not think I lack this knowledge.

            Admittedly, I can't make a logic probe work magic. You have that ability, and I admire it. I'm a historian, not a technician.

            But I certainly "grasp the underlying fundamentals of electronics and computers"...! Not being a hands-on expert doesn't mean one is a dolt. (You're not a professional writer like I am, though I wouldn't assume you can't "grasp the underlying fundamentals of grammar and spelling.")

            [snip]
             
            Friendly counter-jab from this lowly English major: when it comes to history of any engineering/scientific accomplishments, it's rather important to grasp the underlying fundamentals of patents and lawsuits when attempting to rationalize the differences.  :)


            I read the most important thing when it comes to this
            I read the manuals
            Schematics don't lie ;)

            =Dan

          • B Degnan
            ... ENIAC had wiring panels (not papertape), and although it was electronic, it recognized patterns of instructions set up in advance on wiring panels.
            Message 5 of 23 , Nov 8, 2010
            • 0 Attachment
              Evan Koblentz wrote:
              > The ABC * calculated * with tubes -- or at least added and subtracted;
              > it couldn't multiply or divide. But it had two big catches:
              > - Its input and output was mechanical. ENIAC was electronic end-to-end.
              > - It has no program. All steps had to be decided and entered by hand.
              ENIAC had wiring panels (not papertape), and although it was electronic,
              it recognized patterns of instructions set up in advance on wiring
              panels. That's practically the definition on an analog computer. Why
              is it bad that ENIAC was an analog computer? Electronic does not equal
              digital. There was no punched card or papertape fed into the ENIAC.
              Please explain why I am wrong about the ENIAC being an analog computer.

              The ability to multiply and divide is not relevant, computers only add /
              one's compliment math, etc. in order to multiply and divide.

              But seriously. Why the need to defend the established dogma? Why can't
              Mauchly have been directly influenced - at a specific, technical level
              - by Atanasoff. I have a letter from Mauchly to Atanasoff asking
              Atanasoff for permission to use some of his ideas in his new computer
              project. Why can't the ABC be a step, and ENIAC be another step, in
              the continuum of computing advancements?

              Bill
            • system@great-escape.tmesis.com
              ... Yes, it was a digital computer; it wasn t a Von Neumann architecture. I believe you are both tainting your impression of these two predecessors simply
              Message 6 of 23 , Nov 8, 2010
              • 0 Attachment
                B Degnan <billdeg@...> writes:

                >ENIAC had wiring panels (not papertape), and although it was electronic,
                >it recognized patterns of instructions set up in advance on wiring
                >panels. That's practically the definition on an analog computer. Why
                >is it bad that ENIAC was an analog computer? Electronic does not equal
                >digital. There was no punched card or papertape fed into the ENIAC.
                >Please explain why I am wrong about the ENIAC being an analog computer.

                Yes, it was a digital computer; it wasn't a Von Neumann architecture. I
                believe you are both tainting your impression of these two predecessors
                simply because neither are the familiar stored-program architectures you
                have come to know today.
              • Bill Degnan
                Bill Degnan ... Atanasoff ... If you re saying that a computer deos not have to have automatic sequencing and still be digital then ok, but then at best ENIAC
                Message 7 of 23 , Nov 8, 2010
                • 0 Attachment
                  Bill Degnan

                  -------- Original Message --------
                  > From: system@...
                  > Sent: Monday, November 08, 2010 8:38 AM
                  > To: midatlanticretro@yahoogroups.com
                  > Subject: Re: [midatlanticretro] Wired article about John Vincent
                  Atanasoff
                  >
                  > B Degnan <billdeg@...> writes:
                  >
                  > >ENIAC had wiring panels (not papertape), and although it was electronic,

                  > >it recognized patterns of instructions set up in advance on wiring
                  > >panels. That's practically the definition on an analog computer. Why

                  > >is it bad that ENIAC was an analog computer? Electronic does not equal

                  > >digital. There was no punched card or papertape fed into the ENIAC.
                  > >Please explain why I am wrong about the ENIAC being an analog computer.
                  >
                  > Yes, it was a digital computer; it wasn't a Von Neumann architecture. I
                  > believe you are both tainting your impression of these two predecessors
                  > simply because neither are the familiar stored-program architectures you
                  > have come to know today.
                  >

                  If you're saying that a computer deos not have to have automatic sequencing
                  and still be digital then ok, but then at best ENIAC was an analog digital
                  computer because the inputs were analog and only the output was digital.
                  That's how I see it. You're in effect simulating digital, not true
                  digital.


                  bd
                • Evan Koblentz
                  ... Huh? An analog computer measures; a digital computer counts. ENIAC most certainly did the latter. It s irrelevant that it used ad-hoc programming circuits
                  Message 8 of 23 , Nov 8, 2010
                  • 0 Attachment
                    >>> although it was electronic, it recognized patterns of instructions set up in advance on wiring panels. That's practically the definition on an analog computer.

                    Huh? An analog computer measures; a digital computer counts. ENIAC most certainly did the latter. It's irrelevant that it used ad-hoc programming circuits in the form of wiring panels. Nothing is being measured there.

                    >>> Electronic does not equal
                    digital.

                    And wiring panels do not equal analog, as I just explained.

                    >>> There was no punched card

                    Wrong. It could use punch cards for both input and output.

                    >>> Why the need to defend the established dogma?

                    Bill, have you and I ever met before? :)

                    In NO way is my goal to "defend" established anything. As a journalist and historian, I take neutrality to extremes. But as I said in my prior message, this isn't about ENIAC! All that's happening to make me pissy is that you guys are singing the praises of a very bad book, and trying to solve a problem that was already solved eons ago, by people who had the whole story.

                    >>> I have a letter from Mauchly to Atanasoff asking Atanasoff for permission to use some of his ideas in his new computer project.

                    And speaking of not having the whole story ..... yes Mauchly asked, but you're missing important context. 1., he was just being respectful to a fellow academic. 2., he used some abstract IDEAS, but a markedly different implementation. 3., the more he studied the ABC, the more he realized it was an incomplete semi-automated calculator that couldn't even add (1+1) * 1 without human intervention.

                    >>> Why can't the ABC be a step, and ENIAC be another step, in the continuum of computing advancements?

                    Nobody says it's not.
                  • Bill Degnan
                    I am not knowledgable of all of the various books and articles on the subject, so I really can t speak about the stories, patents, etc. I understand your
                    Message 9 of 23 , Nov 8, 2010
                    • 0 Attachment
                      I am not knowledgable of all of the various books and articles on the
                      subject, so I really can't speak about the stories, patents, etc. I
                      understand your points, thanks for sharing your information.

                      My focus on this thread has been my *belief* that if one can't input a
                      program is in a digital format then you're not really digital. I
                      understand that later ENIAC became capable of accepting punchcard input
                      later.

                      Maybe you know more about how the ENIAC was programming initially, when
                      first invented. How did they enter in the programming code using patch
                      panels. I always thought that the ENIAC was more like a Heathkit EC-1, but
                      I don't know much about how the programming "read in" the patch panel
                      input, and what other controls were used.



                      Bill Degnan

                      -------- Original Message --------
                      > From: "Evan Koblentz" <evan@...>
                      > Sent: Monday, November 08, 2010 10:03 AM
                      > To: "MARCH Yahoo Midatlanticretro" <midatlanticretro@yahoogroups.com>
                      > Subject: Re: [midatlanticretro] Wired article about John Vincent
                      Atanasoff
                      >
                      > >>> although it was electronic, it recognized patterns of instructions
                      set up in advance on wiring panels. That's practically the definition on
                      an analog computer.
                      >
                      > Huh? An analog computer measures; a digital computer counts. ENIAC most
                      certainly did the latter. It's irrelevant that it used ad-hoc programming
                      circuits in the form of wiring panels. Nothing is being measured there.
                      >
                      > >>> Electronic does not equal
                      > digital.
                      >
                      > And wiring panels do not equal analog, as I just explained.
                      >
                      > >>> There was no punched card
                      >
                      > Wrong. It could use punch cards for both input and output.
                      >
                      > >>> Why the need to defend the established dogma?
                      >
                      > Bill, have you and I ever met before? :)
                      >
                      > In NO way is my goal to "defend" established anything. As a journalist
                      and historian, I take neutrality to extremes. But as I said in my prior
                      message, this isn't about ENIAC! All that's happening to make me pissy is
                      that you guys are singing the praises of a very bad book, and trying to
                      solve a problem that was already solved eons ago, by people who had the
                      whole story.
                      >
                      > >>> I have a letter from Mauchly to Atanasoff asking Atanasoff for
                      permission to use some of his ideas in his new computer project.
                      >
                      > And speaking of not having the whole story ..... yes Mauchly asked, but
                      you're missing important context. 1., he was just being respectful to a
                      fellow academic. 2., he used some abstract IDEAS, but a markedly different
                      implementation. 3., the more he studied the ABC, the more he realized it
                      was an incomplete semi-automated calculator that couldn't even add (1+1) *
                      1 without human intervention.
                      >
                      > >>> Why can't the ABC be a step, and ENIAC be another step, in the
                      continuum of computing advancements?
                      >
                      > Nobody says it's not.
                      >
                      > ------------------------------------
                      >
                      > Yahoo! Groups Links
                      >
                      >
                      >
                    • Dan Roganti
                      ... Evan knows I like to kid around a lot here, sometimes email is a poor way of conveying this. I just like to yank his chain. I can hear him saying - just
                      Message 10 of 23 , Nov 8, 2010
                      • 0 Attachment
                        On Mon, Nov 8, 2010 at 10:28 AM, Bill Degnan <billdeg@...> wrote:

                        I am not knowledgable of all of the various books and articles on the
                        subject, so I really can't speak about the stories, patents, etc.  I
                        understand your points, thanks for sharing your information.


                        Evan knows I like to kid around a lot here, sometimes email is a poor way of conveying this. I just like to yank his chain.

                        I can hear him saying - just like the farmer, "you dang woodchucks, quit chucking my wood ! "

                        I just take electronics seriously, and I also I like the bantering, it's healthy.

                        Sometimes not enough Smileys will help, but here goes,
                        :) 8) ;) 8-) :) 8) ;) 8-) :) 8) ;) 8-) :) 8) ;) 8-) :) 8) ;) 8-) :) 8) ;) 8-)

                        But the ABC is still a computer !

                        EOL
                        =Dan


                      • Evan Koblentz
                        ... In case anyone s not clear (since Dan inadvertently replied here to a different email) -- he s talking about his earlier comment that one needs a firm
                        Message 11 of 23 , Nov 8, 2010
                        • 0 Attachment
                          > Evan knows I like to kid around a lot here, sometimes email is a poor
                          > way of conveying this. I just like to yank his chain.

                          In case anyone's not clear (since Dan inadvertently replied here to a
                          different email) -- he's talking about his earlier comment that one
                          needs a firm grasp of how computers work -- to which I took some offline
                          offense, even though I had a feeling he was only joking.

                          There, now I feel better. :)

                          Rest assured, MARCHins, your sometimes-flaky leader DOES know how
                          computers work!
                        • Mike Loewen
                          ... There are only 10 types of people in the world: those who understand binary, and those who don t. Mike Loewen mloewen@cpumagic.scol.pa.us Old Technology
                          Message 12 of 23 , Nov 8, 2010
                          • 0 Attachment
                            On Mon, 8 Nov 2010, Evan Koblentz wrote:

                            > Rest assured, MARCHins, your sometimes-flaky leader DOES know how
                            > computers work!

                            There are only 10 types of people in the world: those who understand
                            binary, and those who don't.


                            Mike Loewen mloewen@...
                            Old Technology http://sturgeon.css.psu.edu/~mloewen/Oldtech/
                          • Dan Roganti
                            ... I think a good way to visual this is the patch cables are the bits in your code to program this computer. These patch cables connected digital circuit
                            Message 13 of 23 , Nov 8, 2010
                            • 0 Attachment
                              On Mon, Nov 8, 2010 at 9:44 AM, Bill Degnan <billdeg@...> wrote:



                              If you're saying that a computer deos not have to have automatic sequencing
                              and still be digital then ok, but then at best ENIAC was an analog digital
                              computer because the inputs were analog and only the output was digital.
                              That's how I see it.  You're in effect simulating digital, not true
                              digital.


                              I think a good way to visual this is the patch cables are the bits in your code to program this computer.  These patch cables connected digital circuit modules and not analog circuits. Although it resembles the traditional method of connecting analog computers there wasn't any analog input being used there only digital signals.

                              =Dan

                            Your message has been successfully submitted and would be delivered to recipients shortly.