Computing Pioneer Dies
bhilpert at shaw.ca
Mon Nov 13 15:10:54 CST 2017
On 2017-Nov-10, at 12:37 PM, Evan Koblentz wrote:
>>> I should point out there is a technical error in the Guardian. The Baby was the first Electronically Stored Program in what today we would call RAM. ENIAC had been configured in stored program mode earlier in the year and had run a program stored in the function switches, e.g. ROM a couple of months before baby. Despite the fact that when running stored programs ENIAC's parallel processing features were not available, it was exclusively in this mode from 1948 onwards. Note both machines are theoretically "Turing Complete" but having only 32 words of 32 bits the Baby was not of any practical use for a further 18 months whilst major surgery was carried out to add extra store and instructions to the machine leading the emergence of the Manchester MK1.
>> Funny, I didn't see a "technical error" in the article.
>> The best that can be said for your position is that you (and the ENIAC/Mauchlyite crowd) have a particular opinion and definition regarding 'stored-program computer'.
> Dave is correct.
> Perhaps he should have said "over simplification" rather than technical error.
> But what he wrote is well-documented. Tom Haigh and team of researchers explained the origins and varying definitions of "stored program" in their paper which you can freely read at http://eniacinaction.com/the-articles/1-reconsidering-the-stored-program-concept/. Tom, I'll point out, is British.
And a team of researchers, eh? Wow.
The Haigh, et al article (3 parts) does not disprove the Manchester Baby as the first stored-program machine, nor claim to do so.
And it neither shows nor claims the 48 ENIAC to be the first stored-program machine.
Well, it doesn't explicitly make the claims, but it is rather curious that people interpret it as doing so.
Quote from the article [pg5]:
Should ENIAC therefore be considered the first operational stored-program computer?
Well, it all depends on what we mean by “stored program.”
Which is what I said earlier.
The article proclaims a set of criteria they call the "modern code paradigm" and
claims the 48 ENIAC as the first machine to have fulfilled their criteria.
As we explained earlier, we do not view the modern code paradigm as a new name
for the “stored program concept”.
Using the article to make a claim vis-a-vis "stored program" for the 48 ENIAC as disproving the Baby's widely regarded status
is conflating the "stored-program concept" with some other meaning of "stored program", or creating your own definition
of "stored-program concept", and misrepresenting the (stated) claims of the article.
> Thanks to their research, there is no longer any gray area.
> ENIAC stored a program in what's now known as ROM, and very soon after the Baby stored a program in what's now known as RAM.
> The timeline and facts are very clear.
The Harvard Mark I stored programs, as a sequence of instructions recorded in a storage medium,
and automatically executed them while reading from that medium, several years ahead of the 48 ENIAC,
not to mention the conceptualisation of such for the Analytical Engine.
That's not what you meant about storing a program? You mean you have other defining qualifiers?
Harken yonder! I see grey clouds on the horizon.
On 2017-Nov-10, at 6:16 PM, Noel Chiappa via cctalk wrote:
>> From: Brent Hilpert
>> The best that can be said for your position is that you (and the
>> ENIAC/Mauchlyite crowd) have a particular opinion and definition
>> regarding 'stored-program computer'.
> I'm harly a member of the "ENIAC/Mauchlyite crowd" (in fact, I used to not
> have a good impression of them at all), but I thought Haigh et al made a
> pretty good case.
The article (or the latter parts) are an interesting examination of the 48 ENIAC and it's programming.
As a claim of historical consequence for the 48 ENIAC however, it's just too much contrivance.
Here's an alternative abstract:
We look at the etymology of the phrase "stored program" and decide it's all so confusing,
so we dismiss all the previous historical assessments [pg8-11].
We note as others have that declaring firsts in history is problematic as it inevitably comes
down to providing a string of qualifying descriptors [pg9].
To avoid these problems, we define a set of conceptual elements from the
Draft Report on the EDVAC, to form a new set of criteria of assessment.
We call this set of criteria the "modern code paradigm" [pg12].
We find that the 48 ENIAC fulfills our set of criteria.
We now declare the 48 ENIAC to have been the first machine to fulfill the modern code paradigm [pg15].
(See page-long string of qualifying descriptors defining the modern code paradigm [pg12-13]).
What about that little issue of writeable program storage?
That little characteristic that is rather seminal to computing ever since 1946, and (previous) meanings
of "stored-program" all included, but an element that the 48 ENIAC did not have ?
That was dismissed because it wasn't used in the ways the Draft Report described it (to be) used.
But it's also dismissed because the ways it did come to be used were NOT described in the Draft Report. [pg10-11]
Meanwhile, other criteria were included even though they weren't done the way described in the Draft Report.
And the 48 ENIAC wins ! Huh. What a coincidence.
The authors could have just written an article about the 48 ENIAC with a comparison of it's programming characteristics
to modern programming, but they just had to also attempt to dismiss the "stored-program concept" of historical relevance,
while nonetheless leaving everyone they proselytise to conflating the "modern code paradigm" with "stored program".
Yup, they sure did clear out all the confusion, not added to it.
It's just another attempt to make an historical claim for the ENIAC, this time trying to steal the light away from the Manchester Baby.
Sorry, no sale.
I could go on but I have to pick up my newly tailored suit and a gallon of facts.
More information about the cctalk