As said by Merch:
>Uh, actually, try October 1939... with the ABC. That's the Atanasoff -
>Berry Computer, which is now recognized as the first functional electronic
>computer. I posted 2 or 3 URL's for more info on the web, if there's a
>searchable archive of this list, check there for my post. If not, I'll look
>it up in mine if I still have it.
>
>BTW, you're timeline is still a bit off... IIRC John von Neumann invented
>the stored-program concept, didn't he? It was Markus & someone who designed
>the first UNIVAC mostly with concepts hijacked from the ABC... methinks it
>was in 1972 or so when the U.S. Patent office yanked the patent away from
>them and gave it back to Atanasoff & Berry.
I depends on how you define "computer." Oh no, not another
definition war... In what follows I'll try not to use the word
computer in isolation.
Around the late 1930's there were three groups that were developing
machines that would have some claim to the status of pioneering
computing devices. At Iowa State, there was the ABC as
mentioned. At Harvard, Howard Aiken was developing the Mark I
(actuall built by IBM under the name Automatic Sequence Controlled
Calculator) and in Germany, Konrad Zuse was building his Z1 in
his parent's living room. All three were architected by people
who were motivated to solve difficult mathematical problems,
and all three had some interesting technological aspects.
The ABC was electronic. That is, it used active electrical
devices which in those days meant tubes (valves). It also had
an interesting dynamic memory system that foreshadowed the
delay line memories and later dynamic semiconductor memory.
As to purpose, the ABC was very much a single purpose machines;
it was to solve simultaneuous linear equations. It was also
not particularly automatic in operation. The human operator
was responsible for taking temporary results in the form of
punched cards and putting them back in as new data to be
operated upon. To what degree the device worked is not at
all clear. The card punching mechanism was based on a spark
burning holes in the card and was reported by Atanasoff and
Berry as not ever working reliably enough to solve useful
sized problems. There has also been some question raised as
to whether or not the circuits would have ever worked reliably.
The Harvard Mark I was (according to Aiken) based on Babbage's
designs for the Analytical Engine. Though if you read Babbage
carefully and look at the Mark I design, there were some things
that either Aiken didn't fully understand or that Aiken decided
were too hard to implement, so the Mark I was less capable than
the Analytical Engine would have been (at least some of Babbage's
designs would have been). It was an electromechanical device
driven by punched tape. (Actually uncut punched card stock.)
This machine was much more general purpose and did work reliably
and was used by the Navy. Grace Hopper cut her computing
teeth on that machine.
The Z1 was also designed as a general purpose calculator
that was driven by punched tape. In Zuse's case, the tape
was discarded movie film. His machine was fully mechanical
though and binary in contrast to the decimal Mark I. It also
used a floating point representation as opposed to the Mark I's
fixed point.
A fourth project began late in this same time, the ENIAC. It's
genesis also came from difficult mathematical problems and it
was funded by the Army. John Mauchley was interested in weather
prediction and began discussing a machine to do it with J. Presper
Eckert. Mauchley was a Physicist and Eckert an Electrical
Engineer. Eckert was quite familiar with the use of electronic
circuits as counters for radar and for counting nuclear events.
During the development of the project, Mauchley
met Atanasoff at a conference and began comparing notes.
Atanasoff invited Mauchley to come to ISU for a visit which
Mauchley later did. It was this visit that later provided one
nail in the patent cofin. The ENIAC was also a general purpose
calculator. It was electronic and decimal. Since the electronics
were much faster than the mechanical systems, punched tape or
cards would not have kept up as a control mechanism, so it
was patchboard programmed. It did have the capability to
read data from punched card. (Later a clever engineer did
devise a way for it to be programmed by it's bank of coefficient
switches for a crude stored program capability. However, this
was still not as general as what was to come.) The machine
was very much like a combination of an electronic version of
the Harvard Mark I and a digital version of the differential
analyzers that Bush and others had built.
During this project, in 1944, Eckert wrote a paper about magnetic
disk devices and as an aside mentioned the possibility that
the sequencing instructions could be kept on the disk along
with the data, providing the earliest description of stored
programming. Before ENAIC was completed, work began on it
successor, the EDVAC. This machine was to be a stored program
machine. After the design work had been started but before it
was completed, von Neumann joined the project and along the way
wrote the now famous First Draft paper. Because Eckert's earlier
paper had been lost/forgotten, this became the blueprint for
all stored program computer work for many years. Von Neumann
indicated that he put his name alone on it just because it was
a rough draft, but was planning to put the other names (Eckert,
Mauchley and others) on the final version. Several on the project
felt that he was trying to steal credit for himself though.
In 1946, Penn (where this was taking place) hosted a meeting
about building computers and the cork was out of the bottle.
Projects sprang up at Princeton (led by von Neumann), at
Manchester (led by Wilkes IIRC) and at Cambridge in addition
to the work at Penn. While all this was going on, Eckert and
Mauchley decided to file for patents on the ENIAC. Penn decided
to change its patent policy at the same time and asked Eckert
and Mauchley to sign over their rights to the University. They
declined (over and over again) and left to start the Eckert-
Mauchley Computer Company, which through many changes of hands
is now part of Unisys. Over the next few years, the Manchester
folks first got a small (memory wise) version of their Mark I
up and running. Soon thereafter, the folks at Cambridge got
the EDSAC up and running and then Eckert and Mauchley got the
BINAC up and running in the states. Well, it was apparantly
running in the lab. There's been some disagreement on whether
or not it ever ran reliably after it was shipped to the customer
site. These were the first stored-program computers to be
built.
It took many years for the patents on the ENIAC to be granted
during which time, the computer industry had really taken off.
Once they were granted, Honewell and Sperry (then the owner
of UNIVAC the successor to EMCC) sued each other. The ABC was
used as part of the case. The judge did state in his opinion
that ENIAC was substantially derrived from the ABC, but the
patents were also overturned on the basis of having been filed
too late. Inventors have 1 year after the invention is made
public. It seems that there was a demonstration of the machine
prior to 1 year before the filing. I (and others) disagree
that ENIAC was substantally derrived from ABC. The only similarity
seems to be the fact that they were electronic and the idea
of electronic counters (which is most of what ENIAC was) had
been in place and well known by Eckert before the Mauchley
trip to ISU. I don't mean this to dimish the importance
of the ABC as the first electronic special-purpose calculator,
nor does the ABC dimish the importance of the ENIAC as the
first general-purpose electronic calculator. Historically,
the ENIAC had a much greater influence, spawning the whole
of computing as we know it today. Atanasoff went on to other
things and had no dealings with computing except working on
the patent trial, and the ABC was dismantled and mostly forgotten.
Brian L. Stuart
Math/CS Dept, Rhodes College, Memphis, TN
stuartb_at_acm.org
http://www.mathcs.rhodes.edu/~stuart/
Received on Tue Jul 01 1997 - 10:48:54 BST