Wednesday, March 21, 2007

Here's to engineers

There are many great names you have never heard nor will ever hear if you don't major in certain engineering fields. You will never hear of the formulas and methods they developed. If Schockley hadn't been a racist loony, his name and his transistor company and his Nobel prize would be a trivia question, unfamiliar outside of places like Intel, Motorola and engineering departments with older faculties.

You can not go to work, read this sentence or cook your breakfast without leaning all your weight on the technical sinews of our world. You can hardly afford to be conscious of the ubiquitous utilities devised by people who you are even less conscious of. NYTimes reports that John W Backus has died. The name is fairly obscure outside of computer science curricula. He invented one of the earliest computer languages, Fortran, and led the team that developed it. He was an indifferent student in high school, finally a drop out, who still managed to get into the math program at Columbia U by blowing the top off of all aptitude scoring when he tried to enlist during WWII. The army sent him to school. The job interview that led to his work in computer languages at IBM would be a minor legend, at least among us geeks, if anyone knew it. [That is written up in the linked obit.]

He did more than design one of the most widely adapted language compilers in the history of computing; he invented, with Peter Naur, a language in which to describe computer languages in such a way that compilers can be automatically generated for the language so described. It is called "Backus Naur Form", BNF for short and sometimes, incorrectly, "Backus Normal Form". Normal forms are also mathematical notations for syntax rules that have various constraints or stipulations that guarantee any conforming language can be parsed with a particular class of algorithms. This whole line of meta-tool inventing has been immensely but invisibly fruitful. No use of computers you are now dependent upon does not rely to some degree on outcomes of this work.

I think it is fitting and a bit ironic to use the most visible layers of the computing world, the indispensable Google, to demonstrate how invisible one of its founding thinkers has become. I searched the web for references to the tool used in the context of several current programming languages. Clearly, BNF is ubiquitous to this day in the labs and back office cubicles where your computing infrastructure is forged. You may have heard of some of these languages even if you don't code. The count of matching pages are indicative but the older languages precede the WWW era and may be somewhat under reported.

By contrast, If you search for the name of the man who invented BNF, and eliminate false matches to men with identical names who taught acoustics or ran investment banks, you get fewer hits. 55,900 English pages for "john backus" -woodwind -music -acoustical -acoustics -"managing -partner" -"technology -investor" -entrepreneur. And that probably over-reports due to other false matches. If you demand a precise full name, Backus' presence shrinks to what amounts to nonentity beside the presence of his accomplishments.


--------======= And now for your quiz =======--------

  1. Leave a comment naming the person who gave the name "bug" to the concept of a programming error or computer fault.
  2. For extra credit, leave a comment naming the brilliant hardware designer who, among many contributions to computer engineering, accelerated the Computer-Aided-Design that makes possible our multi-million transistor computer chips...but who was near suicide and forced out of a stellar career at IBM because of a sex change operation.




UPDATE: when I published this morning, the google of "john w backus" got only a few thousand matches. I see Mr. Backus has been slashdotted. Most of the hits are dated from a crawl done late yesterday and what looks like a majority of them are just quotes of NYTimes. I don't think that fully accounts for the explosion to 1.15 million hits. That would NOT equate to "insignificance"!. Every RSS feed in the world has echoed it too. Am researching.
UPDATE of UPDATE:yesterday morning it was 82,000 hits. Yesterday evening it was 1.15 million, this morning it is 2.1 million hits. What we have hear is a census that is capturing the way a topic explodes on the internet. You will just have to take my word for it: your 15 minutes of internet fame may only follow your death. We executioners are observers of these effects.
UPDATE of UPDATE of UPDATE: Wheee! the "nonentity" link now finds 3.48 million. Is that a phenomnon you could explain to your grandparents? I feel like Winnie the Pooh ominously discovery more and yet more heffalump tracks in the snow with each turn round the bushes. The crawl-and-index engines at google must be pulling enough juice to dim lights all up and down the west coast.

If you are geeky enough to want to read Backus' original papers, he donated them to the Library of Congress two years ago and they were just indexed last month.

2 comments:

etbnc said...

Backus-Naur Form was one of the last things I learned in my senior year of computer engineering. (Or at least, one of the last things I remember!) That was the class in which a little text-search-and-replace project turned into a crash course in compiler design.

If I recall correctly, the rulesets of /etc/sendmail.cf were (are?) pretty much condensed BNF production rules. I remember becoming way too familiar with them ... once upon a time.

I don't wanna spoil the quiz for other readers. The first answer came to my mind readily, like a moth to a brightly glowing vacuum tube. I recognized the story of the second, but I had to look up the person's name.

Quiz question for you, GS: By any chance, have you met those people?

GreenSmile said...

Thanks for your restraint ETBNC...I recall you are one who would know.

I missed a chance to meet the admiral when ACM brought her to MIT back in the 70's. I did get to see Mead's laboratory at CalTech but at that time his collaborator was working at Xerox.

[I'd swear I put this comment in 2 hours ago but it did not register. might be one of those moths got stuck in the intertubes.]