Monday, November 8, 2021

A Technological Appendix to July 1951, II: Cracks, Clerks, Picks


John E. "Red" Parkinson, Chosin veteran, of the Weapons Company, 3rd Battalion, 7th Marine Regiment, 1st Marine Division, was to have been honoured in death by a "Patriot Guard Rider retinue at his 11 April 2018 funeral, according to the website

Doesn't sound like the footloose troubadour of the Summer of Puppy Love, but that's your Silent Generation for you, and, anyway, people change. 

A long time ago, in the first days of this blog, I offered a half-assed connection between various British ventures in computing and sound processing during WWII, and Abbey Road. With July of 1951, we've come to the point where the connection is pretty much inevitable. We're done with faffy British stuff and have arrived at the beating heart of the rock and roll music of the kids today. Kids like Sergeant Parkinson. 
Les Paul and  Mary Ford recorded their version of The World is Waiting for the Sunrise in July of 1951 and premiered it on their own show, which seems to have been a five-minute feature sponsored by Listerine. The audience for their 19 August 1951 appearance on the Ed Sullivan Show clearly knew what they were waiting for
and the performance features Paul's picking prominently while giving Mary  Ford (Iris Colleen Summers) some multitracking. The producers knew what made this act special. Les Paul, in particular, is a rock and roll legend, even if, at the time of the Sullivan appearance they were touring as a country act. Apart from his guitar styles, Paul is known for early experiments with electric guitars and multitrack recording, but for the musically inclined (your humble blogger most definitely not included), the name will probably be associated with the Gibson Les Paul, a very successful attempt by that major guitar maker to pick up the electric guitar market pioneered by Fender (or possibly a hapless Gretsch.) Paul himself had experimented with electric guitars earlier in his career, but he arrived at The World is Waiting in collaboration with Bing Crosby, who was enthusiastically pushing taped music after persuading ABC to air his Philco Radio Time show, starting in August of 1947, recorded on an Ampex reel-to-reel tape recorder. According to Les Paul, Crosby presented him with an Ampex 200A tape recorder in 1949, after inspiring Paul to open his own studio. It was there that Paul worked up his "Sound to Sound" recording technology, in which an additional playback head allowed him to play along with a previously recorded track and mix it into a new one (which was recorded over the original track.)  Multitrack machines would follow in the late 1950s, the first Ampex 300 being installed at Les Paul's studio in 1955, with "Sel-Sync" technology to reduce hum and noise in the process.

I haven't given the other big new hit of the summer, Rocket 88, quite the same attention, because, well, it's Ike Turner, but Youtube has a video, with a Betty Page cosplayer, or possibly the lady herself, just to let you know that, well, it's the Fifties.  Musicologists are in some disagreement about just what made this "twelve-bar blues" song so different from other rhythm and  blues singles of the day, and in particular, Jimmy Liggin's "Cadillac Boogie." Was it Turner's arrangement, or Willie Kizart's electronically distorted "fuzz guitar" performance, one of the first of its kind? The blog will plump for the latter! "Fuzz guitar" is a variety of distortion, an electronic music technique in which amplification gain is pushed out of the range in which the amplifier can give adequate fidelity, producing a distorted sound sometimes described as "fuzzy," but also "growling," or "gritty."
(Because I am not putting Ike Turner up here.)

So, yes, it does go up to eleven. Now shut up.

On the other hand, we have Joshua Whatmough, who makes Newsweek on the strength of his application of information theory to language. I was going to say that I couldn't pull up anything useful about Whatmough, only to find the 28 April 1964 Harvard Crimson obituary at the top of my search results today and a pretty informative bibliography shortly afterwards. A former chairman of Harvard's Department of Linguistics, The Crimson pays tribute to his indefatigable advocacy of "statistical method in linguistic analysis" that "now promises to bring order into the field." It appears that there are grounds for a breach of promise suit, and the aforementioned bibliography includes a positively brutal note that I cut and paste in part:

His later work on language and general linguistics will soon be forgotten, though his enthusiasm for statistical methods, currently out of fashion, may well prove prescient one day. His teaching at Harvard was vatic and inspirational rather than informative, and he puzzled his classical students more than he taught them. He had early discovered that most such students, though required to take courses in comparative grammar, are incapable of learning linguistics. He therefore gave up the attempt, talking instead on any number of topics, some linguistic, others not. He would introduce the Indo-European lost laryngeal consonants to students who could barely comprehend changes of sounds with which they were familiar: he was no doubt intrigued by the symmetry and quasi-mathematical neatness he saw in the larygneal theory. For the interested student he was most generous and helpful, and aided such students in their careers. He was very good company in social settings.
 . . . And it goes on. Whatmought's enthusiasm for applying information theory to language is entirely passed over. Given the connection with codebreaking (and automatic translation) that constitutes the received version of the origins of computational linguistics, the "cracks" of my title, there maybe some classified subtexts here. (Whatmough's "Monuments Men" inspired biography certainly thinks so.) Or possibly not. This being also the month that Kim Philby was released from custody because MI5 couldn't find a way to link him to the defections of Burgess and MacLean, espionage is most definitely in the news this month. The human angle is the only one in the press, but Philby is notorious for giving the Soviet intelligence services their clearest and most complete picture of the Venona project, the excruciatingly technical analysis of compromised Soviet one time pad-encrypted diplomatic cables that was the bridge that American codebreaking crossed from its WWII efforts against the Nazis into the Cold War. 

 Prosaically, "information theory" sounds impressive but is really at this point mostly a theory of gain. Analog amplifiers consist of input currents modulating a power source. That is, you plug the plus side to the minus side, which gives you a constant current, and run it through a switch that it is controlled by a current that is a "signal."

This may not be the clearest illustration of what I am talking about, but most Youtube videos on old time cathode ray oscilloscopes are enthusiastically long, and for my purposes it suffices to imagine that this switch in the steady current goes ON when the curve goes up, and OFF when it goes down, and the result is that you have modulated the steady current into an output signal that looks like the input signal, only with an arbitrary amplification or even diminution of power --your "gain." As a practical matter, the amount of gain you can have is limited by electrical effects that introduce distortion, and information theory is a powerful tool for predicting just how much distortion you can have, and just how effectively it can be "filtered" or removed from the output signal.  It is a statistical analysis, that can be applied to "filtering" a variety of signals, from music to dirty commies talking on the radio to automated controllers. 

The central piece in the information theory puzzle is the "switch" of my analogy, and that is what the germanium wafer transistor that Bell has announced this month, is for. Well, it is also a pretty good thingie for running the input signal through, and for carrying the power supply that will provide the output signal, but, mainly, we're talking about the "switch," or "logic gate." 

None of this, it will be noted, is digital. It is all analog signal processing. Even the transistor is being sold as a tool for analog signal processing. They are applicable to digital methods. The Ferranti Mark 1, first publicly demonstrated this month, and inspiring The Economist to dream of a day when such a machine might be used to figure out paycheque income tax withholding and freeing up all those clerks to do something more useful, uses vacuum tubes for the same work, but also requires "memory," which, it turns out, was also done on vacuum tubes and even magnetic drums by this time. (Before I reread the Wikipedia article I had the idea that it used delay line technology.) Of course, as The Economist correctly points out, the real issue is data input rather than processing, and no-one's got the answer for that. 

The thing about music is that it is all about the input and output. Les Paul is famous for processing musical signals, but he is much more famous for the way that he inputted notes.  

Check out that Listerine ad! Amazing that we made any progress at all what with all the domestic murdering going on. Lucky that we have a way of making money from these here "computers" where you don't have to worry about finding the clerks to do all the inputting!  


  1. Hope you and yours are all staying safe with all the road outages around right now!

  2. Fortunately, apart from my high-flying pediatrician brother, none of my extended family was caught on the roads by this catastrophe.

    Everybody in the province and beyond is going to be affected by the closing of Highway 5 (Coquihalla) and the railroads, though. That most definitely includes us, although as I am currently working in a fresh produce related role, I am still dealing with the aftermath of the California rain storm of last month, which destroyed many crops in the fields.