The presentation is in almost identical order as the PopSci article, and all of the facts in this mashable.com essay are a subset of those in PopSci, including the quote from "Dr. Grace Murray Hopper". A modern author who knew the topic is more likely, I think, to use "Rear Admiral Grace M. Hopper". The PopSci article also covers many other details - you should read it if this article interests you.
That quote, by the way, comes from page 1 of her 1955 book "Automatic Coding for Digital Computers", available at http://www.textfiles.com/bitsavers/pdf/univac/HopperAutoCodi... . A search for that quote using DuckDuckGo and Google finds matches only from the book, the PopSci article, and this new essay. This is consistent with my belief that PopSci is the primary and sole information source for this new piece.
FWIW, I learned about the PopSci article because of a reference in the seminal book 'Punched Cards'. One of the chapters describes textual analysis, and cites several articles about the Univac work, at http://babel.hathitrust.org/cgi/pt?id=uc1.b3958636;view=1up;... .
We've changed the URL from http://mashable.com/2015/09/27/univac-bible/, which I think dalke has shown has plagiarized this excellent 1955 article. It's pretty sickening to think of what else these websites must be ripping off without attribution.
I've marked this subthread off-topic because hopefully the thread can be about the extraordinary details of the original article. But props to dalke for an extraordinary bit of research.
Might be plagiarized, but the pictures from the mashable article add to the story. People should still check it out after reading the original but they should really source this issue :/
Particularly the last photo, a man drawing out computer workflow with a bible open in front of him. Quite a juxtaposition to think about!
This makes me think of King James Programming: "Posts generated by a Markov chain trained on the King James Bible, Structure and Interpretation of Computer Programs, and Why's Poignant Guide to Ruby" http://kingjamesprogramming.tumblr.com/
What gets me is there's no need to not attribute the Popular Science article. I think it's perfectly reasonable to present treasures from the past. If this has ended with "this is a condensed version" or "this is a summary", then no one would have thought the worse. And the addition of the images (which do have proper citations), does add more to the original Pop Sci article.
I am researching the history of my field, cheminformatics. It grew out of documentation management, more specifically, chemical documentation. While that topic goes back to the 1800s, I limit my research to mechanical methods for chemical documentation search and management.
This started with punched cards in the 1940s. James Perry, then at MIT, was a nexus in the field. It's difficult to find people of the era that he wasn't connected to.
One of these is Calvin Mooers, who is a special focus of mine. Mooers developed some of the fundamental concepts that formed the basis for the computerization of chemical information. (To be more precise, he came up with ideas of how to solve problems. These ideas weren't solutions, but were in the ballpark, and ended up influencing the people who actually solved and implemented the problems.) Mooers was a Master's student at MIT, where he met Perry, and was influenced by the sorts of problems that Perry was interested in.
Mooers' Master's thesis was on a method for superimposed coding which he termed Zatocoding. A superimposed code is one where the keys are assigned to 1 or more bits, and where the bits for different keys may overlap. A Bloom filter is a type of superimpose code. In Zatocoding, there is one pass over the keys to get population statistics, then a bit of information theory to determine the number of bits to use for each key and its bit assignments. A common key ends up with 1 bit, while a rare key might have 7 bits.
Mooers is one of the contributors to "Punched Cards". I likely read the book while trying to track down his publications. I may have read it because Perry was the editor. Or I might have read it because I was trying to understand more about edge-notched punched cards.
As for 'seminal', I based that on the number of sources I've read which cite it. For something more outwardly visible, there are 100 citations at https://scholar.google.com/scholar?cites=464718015108686727&... . And of course the book went through two editions.
Interesting. One of those references is to "Automated concordances and word indexes: The fifties". That article says 'Father Busa inaugurated the history of the automatic production of concordances and word indexes' and 'In 1951, using the punched-card equipment available to him, Father Busa produced and published the first automated word index and concordance to four hymns written by Aquinas for the Feast of Corpus Christi, ...'
It's the same Busa who had the chapter in "Punched Cards" about working with the Dead Sea Scrolls; the one that got me to track down the Popular Science sciation. While that chapter doesn't establish the seminal nature, it does give an idea that the book was presenting cutting edge work with punched cards.
Thanks for the detailed reply! I mentioned the PopSci article to an acquaintance of mine who is a New Testament scholar and he brought up Busa, although could not recall the name offhand. I took a closer look at your Punched Cards reference and it was quickly clear Busa was the person he had in mind.
I'm somewhat surprised this sort of interesting early work on coding and natural language processing using mechanical and electronic devices with very limited online storage (not to mention speed) is so seemingly obscure.
I think it's obscure because it solves a problem we don't really have. We take computers for granted, and mostly group solutions into "with a computer" and "manually" (with pencil and paper), because that's what we are used to.
We aren't used to thinking about mechanical solutions, and clever indexing schemes for how to organize piles of paper. Before I started on this research, I had never heard of edge-notched cards, much less appreciated coding schemes that minimize the number of manual needle insertions needed to search a deck.
I've been trying to understand why superimposed coding systems did not reach the modern world. The principles behind it really helped me understand how some modern techniques work, but those modern techniques are ad hoc. Already by 1974 superimposed coding was considered an 'old technique.' I quote from "The Large Data Base File Structure Dilemma", http://dx.doi.org/10.1021/ci60001a005 :
> The second solution to the dilemma, suggested in Figure 7, is sequential search. This may appear to be a throwback, particularly within the framework of on-line systems, but it should be noted that the speed of the CPU combined with the extremely high data transfer rate from secondary storage of third and fourth generation computers has reached a point where one might profitably reexamine the potential of sequential processing.
I interpret this and similar articles to mean that the older techniques weren't a good fit for the then-current computer architectures. After two generations of disuse, institutional memory fades. And reading the old literature doesn't help if you read it through the eyes of people with 1GHz processors - many of the driving problems are trivially brute-forceable.
I read "Punched Cards" in part to get a sense of what it was like to think in those old terms. Now I want to acquire edge-notched cards and physically experience it. But no one sells those cards any more, and haven't for decades. There's a lovely article about this 'dead medium' at http://kk.org/thetechnium/one-dead-media/ .
Luckily, we also have new solutions. This modern crafting era has machines that can mark and cut paper product with ease. I think it would be possible to make each card "from scratch" on the computer, rather than the old school style of using large machinery to cut and print a template, followed by manually content addition.
I think it's obscure because it solves a problem we don't really have. We take computers for granted, and mostly group solutions into "with a computer" and "manually" (with pencil and paper), because that's what we are used to.
Yes I think in a way this drives home the gigantic conceptual advance of 'stored program computer'. Even if your day job is 'isomorphic javascript', if you had no choice you could probably crack open TAOCP and, with a bit of work, understand what it has to tell you, even if written in MIX. A tiny temporal step back puts you in an incomprehensible world of clattering opto-mechanical machinery and bits of cardboard.
Are you publishing this research somewhere for all of us who will neither find nor read Punched Cards, vol 17?
> Are you publishing this research somewhere for all of us who will neither find nor read Punched Cards, vol 17?
That's a really hard thing to answer. First off, there is a huge amount of literature on the topic. I just recently came across "Punched-Card Systems and the Early Information Explosion, 1880–1945", which is the generation before what I'm really interested in. So what will I be able to contribute, and why might what what I do stand out from the rest of the wealth of information already there?
Even in my small field, there is decades of publication history by some very clever people - and some in German. I've got an overview now, and am fleshing out the details.
That publication history includes many tellings or summarizing of the history. Of which quite a few are wrong in the details. For example, I yesterday read some papers from the early 1990s which said that the "connection table" (a graph theoretical model of a molecule) comes from the 1970s. When I can point to papers from the early 1960s which use the term, and people in the 1950s who were implementing the concept on a computer - the concept often comes before the name.
Which means most of the history I want to publish will be deathly boring to someone like you who isn't in the field. I'm hoping for something a bit TAOCP-like, as nothing like that exists in my field.
I think the best part of the article for me was the times given, how long to prepare the data, process it, and such. The manual labor hours are simply astounding let alone the computing hours once the data was ready
If you keep scrolling down, you'll see some interesting simple mechanical hacks in "Hints from the Model Garage" on page "Sidan 181" (that page number is at the top right corner of the browser window).
My favorite hack there is a sartorial one: "Your best hat won't get sat on [in the car] if you stow it out of the way in a holder." The text and a diagram show how to make a hat holder with a coat hanger and metal screws. This of course was in the days when men and women wore hats. (I've read that John F. Kennedy single-handedly killed the hat industry: He didn't like to wear them, and when he became president he set a new fashion trend.)
14 comments
[ 2.7 ms ] story [ 44.6 ms ] threadThe presentation is in almost identical order as the PopSci article, and all of the facts in this mashable.com essay are a subset of those in PopSci, including the quote from "Dr. Grace Murray Hopper". A modern author who knew the topic is more likely, I think, to use "Rear Admiral Grace M. Hopper". The PopSci article also covers many other details - you should read it if this article interests you.
That quote, by the way, comes from page 1 of her 1955 book "Automatic Coding for Digital Computers", available at http://www.textfiles.com/bitsavers/pdf/univac/HopperAutoCodi... . A search for that quote using DuckDuckGo and Google finds matches only from the book, the PopSci article, and this new essay. This is consistent with my belief that PopSci is the primary and sole information source for this new piece.
FWIW, I learned about the PopSci article because of a reference in the seminal book 'Punched Cards'. One of the chapters describes textual analysis, and cites several articles about the Univac work, at http://babel.hathitrust.org/cgi/pt?id=uc1.b3958636;view=1up;... .
We've changed the URL from http://mashable.com/2015/09/27/univac-bible/, which I think dalke has shown has plagiarized this excellent 1955 article. It's pretty sickening to think of what else these websites must be ripping off without attribution.
I've marked this subthread off-topic because hopefully the thread can be about the extraordinary details of the original article. But props to dalke for an extraordinary bit of research.
Particularly the last photo, a man drawing out computer workflow with a bible open in front of him. Quite a juxtaposition to think about!
"The lord is my shepherd,
I shall not want;
he makes me lie down in green pastures.
He leads me...."
This makes me think of King James Programming: "Posts generated by a Markov chain trained on the King James Bible, Structure and Interpretation of Computer Programs, and Why's Poignant Guide to Ruby" http://kingjamesprogramming.tumblr.com/
This started with punched cards in the 1940s. James Perry, then at MIT, was a nexus in the field. It's difficult to find people of the era that he wasn't connected to.
One of these is Calvin Mooers, who is a special focus of mine. Mooers developed some of the fundamental concepts that formed the basis for the computerization of chemical information. (To be more precise, he came up with ideas of how to solve problems. These ideas weren't solutions, but were in the ballpark, and ended up influencing the people who actually solved and implemented the problems.) Mooers was a Master's student at MIT, where he met Perry, and was influenced by the sorts of problems that Perry was interested in.
Mooers' Master's thesis was on a method for superimposed coding which he termed Zatocoding. A superimposed code is one where the keys are assigned to 1 or more bits, and where the bits for different keys may overlap. A Bloom filter is a type of superimpose code. In Zatocoding, there is one pass over the keys to get population statistics, then a bit of information theory to determine the number of bits to use for each key and its bit assignments. A common key ends up with 1 bit, while a rare key might have 7 bits.
Mooers is one of the contributors to "Punched Cards". I likely read the book while trying to track down his publications. I may have read it because Perry was the editor. Or I might have read it because I was trying to understand more about edge-notched punched cards.
As for 'seminal', I based that on the number of sources I've read which cite it. For something more outwardly visible, there are 100 citations at https://scholar.google.com/scholar?cites=464718015108686727&... . And of course the book went through two editions.
Interesting. One of those references is to "Automated concordances and word indexes: The fifties". That article says 'Father Busa inaugurated the history of the automatic production of concordances and word indexes' and 'In 1951, using the punched-card equipment available to him, Father Busa produced and published the first automated word index and concordance to four hymns written by Aquinas for the Feast of Corpus Christi, ...'
It's the same Busa who had the chapter in "Punched Cards" about working with the Dead Sea Scrolls; the one that got me to track down the Popular Science sciation. While that chapter doesn't establish the seminal nature, it does give an idea that the book was presenting cutting edge work with punched cards.
I'm somewhat surprised this sort of interesting early work on coding and natural language processing using mechanical and electronic devices with very limited online storage (not to mention speed) is so seemingly obscure.
I think it's obscure because it solves a problem we don't really have. We take computers for granted, and mostly group solutions into "with a computer" and "manually" (with pencil and paper), because that's what we are used to.
We aren't used to thinking about mechanical solutions, and clever indexing schemes for how to organize piles of paper. Before I started on this research, I had never heard of edge-notched cards, much less appreciated coding schemes that minimize the number of manual needle insertions needed to search a deck.
I've been trying to understand why superimposed coding systems did not reach the modern world. The principles behind it really helped me understand how some modern techniques work, but those modern techniques are ad hoc. Already by 1974 superimposed coding was considered an 'old technique.' I quote from "The Large Data Base File Structure Dilemma", http://dx.doi.org/10.1021/ci60001a005 :
> The second solution to the dilemma, suggested in Figure 7, is sequential search. This may appear to be a throwback, particularly within the framework of on-line systems, but it should be noted that the speed of the CPU combined with the extremely high data transfer rate from secondary storage of third and fourth generation computers has reached a point where one might profitably reexamine the potential of sequential processing.
I interpret this and similar articles to mean that the older techniques weren't a good fit for the then-current computer architectures. After two generations of disuse, institutional memory fades. And reading the old literature doesn't help if you read it through the eyes of people with 1GHz processors - many of the driving problems are trivially brute-forceable.
I read "Punched Cards" in part to get a sense of what it was like to think in those old terms. Now I want to acquire edge-notched cards and physically experience it. But no one sells those cards any more, and haven't for decades. There's a lovely article about this 'dead medium' at http://kk.org/thetechnium/one-dead-media/ .
Luckily, we also have new solutions. This modern crafting era has machines that can mark and cut paper product with ease. I think it would be possible to make each card "from scratch" on the computer, rather than the old school style of using large machinery to cut and print a template, followed by manually content addition.
Yes I think in a way this drives home the gigantic conceptual advance of 'stored program computer'. Even if your day job is 'isomorphic javascript', if you had no choice you could probably crack open TAOCP and, with a bit of work, understand what it has to tell you, even if written in MIX. A tiny temporal step back puts you in an incomprehensible world of clattering opto-mechanical machinery and bits of cardboard.
Are you publishing this research somewhere for all of us who will neither find nor read Punched Cards, vol 17?
That's a really hard thing to answer. First off, there is a huge amount of literature on the topic. I just recently came across "Punched-Card Systems and the Early Information Explosion, 1880–1945", which is the generation before what I'm really interested in. So what will I be able to contribute, and why might what what I do stand out from the rest of the wealth of information already there?
Even in my small field, there is decades of publication history by some very clever people - and some in German. I've got an overview now, and am fleshing out the details.
That publication history includes many tellings or summarizing of the history. Of which quite a few are wrong in the details. For example, I yesterday read some papers from the early 1990s which said that the "connection table" (a graph theoretical model of a molecule) comes from the 1970s. When I can point to papers from the early 1960s which use the term, and people in the 1950s who were implementing the concept on a computer - the concept often comes before the name.
Which means most of the history I want to publish will be deathly boring to someone like you who isn't in the field. I'm hoping for something a bit TAOCP-like, as nothing like that exists in my field.
My favorite hack there is a sartorial one: "Your best hat won't get sat on [in the car] if you stow it out of the way in a holder." The text and a diagram show how to make a hat holder with a coat hanger and metal screws. This of course was in the days when men and women wore hats. (I've read that John F. Kennedy single-handedly killed the hat industry: He didn't like to wear them, and when he became president he set a new fashion trend.)