Tuesday, March 22, 2011

Fall of France, 9: Manpower, Part 6, 2: Computers: Wet Work

Norbert Wiener was probably a pretty difficult guy to work with. Apparently, he'd already cut ties with his fellow "cybernetic" enthusiasts by the time his book, Cybernetics: Or Communications And Control in Animal and Machine appeared in a well-orchestrated burst of publicity in 1948. Be that as it may, the book was a publishing sensation, a heads up about the startling world that's coming. I was never terribly clear on what "cybernetics" might be, but that may be because I'm the child of the waning days of the buzz, and my first idea of what it was about was that it allowed a miniaturised scientist to talk to ants, and, thereby, fight crime.
Other things, too, but we won't go there. Credit to ComicBookMovie.Com

Which wasn't nearly as wrong as it could be. It comes, of course, directly out of the antiaircraft experience, where decoding machines, radars, and radios fed national telephone networks with information that was synthesised by control centres and sent down to assemblies of vacuum tubes, magslips, "oilgears," amplidynes, and swashplate engines that pointed 3.7" AA guns and searchlights at German bombers. Surely there needed to be some kind of heuristic that swept this up into an easily comprehensible thing.

Bruno Latour is either still working on that, or he isn't. As it happens, John von Neumann ended up winning the heuristic war with another (and probably equally ill-considered) "machines are like brains, because brains are like machines" manoeuvre that may or may not have something to do with the thing that I'm typing on right now. Those are big questions, though, and while the Big Thinkers cogitate, it seems to me that there's room for Interesting Facts to perhaps deepen the discussion. That's why I'm going to talk about hydraulics today.

One way of looking at this is that Wiener's book comes out of his efforts in the supercession of the Mark 37 with the new Mark 56 Naval Fire Control Director. Yet it is precisely the supposed superiority of the Mark 37 that activated so much nerd rage in the late 90s when it was rediscovered. I've discussed this elsewhere, so  suffice it to say that we are in a weird position where the hydraulics community has risen from the grave and wrenched history --but certainly not engineering!-- back from electronics.

Wow. But there was a time when oil hydraulics was a Big Deal. Hydraulic power proper has the typical engineering-heroic pedigree, and a particularly interesting (to me) one. One of  the founder-heroes, Joseph Bramah, gets slighted by Smiles' Life of the Engineers, probably for being an Anglican. The other is William Armstrong, a well-known name in engineering to be sure; but lack of interest in the things that I'm interested in manages to drop the fact that he was also the heir of Bamburgh Castle! At this point I'm beginning to think I have enough material for a "hidden history" fantasy here....But as F. H. Towler used to say, their inventions were an unmitigated disaster to the field because they insisted on using accumulators, when the (distant) future belonged to direct, pump-actuated machines. You have probably never heard of F. H. Towler, but he appears to have designed important components of 10 turnkey shell-forging factories exported from the United Kingdom during World War II, including 7 to the United States. (Have I ever intimated my pleasure at discovering factoids tending to disprove the Barnett Thesis?) (Proc. Inst. Mech. Eng. 154 (1946): 178-89; ASME Trans. 71 (1949): 501–14.; Hornby, Factories and Plant305; citing W. Littlejohn Philip, “The Hydraulic Operation of Lathes for the Production of Shells,” Proc. Inst. Mech. Eng. 151 (1944), which actually left me none the wiser.)

Towler's interest, since he's all for the elimination of accumulators, is in the new high-speed hydraulic pumps. And by "new," I mean the Hele-Shaw variable piston-length radial engine, described here by Stuart Bennett. I hate to dump on Bennett's work, since he's the only control engineer to even attempt to give us a monograph history of this complex field, but he often misses developments in the Services that might not have been so well publicised in the technical press. ( A general complaint.) I believe that the Hele-Shaw pump was already in use in new battleship turrets by this time, and also was in the guts of the big forging presses at this point. Believe is a pretty weak phrasing, but I really need to reread the articles on technical developments in the pre-WWI Brasseys at some point. If my memory is serving me, this is an interesting although not unexpected convergence between Admiralty and naval industry practice; the same convergence that I believe prejudices military technical developments in general down industry-friendly paths.  

Okay, but I'm talking about computers, not hydraulic pumps, right? Now, it could be argued, via a discussion of control and stability issues, that they're practically the same thing. But then I'd have to talk about control theory, which I barely begin to understand, and launch into the pure blue sky of consciousness theory. (What do we mean when we say that a "carburettor thinks for itself," as in the old Dowty ads. Is it any different from saying that an artificial intelligence thinks? Or that a human thinks?)

Moving right along..Here's a real oil-hydraulic brain in the mainstream of "cybernetics:" the Kerrison Predictor.  Briefly, in the later prewar years, in both the UK and the USA, work began on automatic remote power control of AA weapons emerged as a priority. In the UK in 1938, adaptation of the ARL Oil-hydraulic system for use on the army’s 40mm Bofors AA gun began. (We can go to Douch for this, but also to S. Bennett, History of Control Engineering, 1930--1955 (London(?): Peter Peregrinus for the Institution of Electrical Engineers, 1993), 131).Tthe upshot was Major Kerrison’s predictor, described in Brown, 1955, 171 as the 500lb Kerrison (Vickers) device. Quoting (apparently) NARS Warren Weaver, ‘Foreword to report on A.A. Director T-10,’ undated (about 1944) ‘Division & (3)’, Office files of Warren Weaver, Records of Division 7 on the Kerrison and Sperry predictors, “Angular and linear variables were characteristically represented by positions of rotating shafts or by lengths of metal bars; resolution and synthesis of vectors and certain multiplications were handled by differential gears; ballistics and other functions were incorporated in two and three dimensional cams.”
David Mindell has more to say. By the beginning of WWII the principal American (and widely used foreign AA fire control director ws the Sperry M-7. It now “incorporated an altitude predictor for gliding targets, could accept electrical inputs from radio rangefinders, and implemented full popwer control of the guns. This computer, culminating 15 years of work at Sperry, was a highly developed machine, optimized for reliability and ease of operation and maintenance. Its design capitalised on the strengths of the Sperry Gyroscope: data transmission, intimate involvement of technical officers in the armed services; human mediation in the computation cycle; and manufacturing of precision mechanisms. It was an elegant, if intricate device, weighing 860lb and including roughly 10,000 parts.
Still, producing the M-76 ws not easy, and the difficulty limited its usefulness. The much-touted ballistic cams best illustrate the manufacturing difficulties of Sperry directors. These strangely shaped parts originated in the numerical firing tables provided by the army’s Aberdeen Proving Ground. From the data in these tables a machinist would fabricate the cams directly, without going through the intermediate stage of blueprints. First, a rough cam would be cast, and then the machinist would drill hundreds of small holes, working from numbers on the artillery firing table. He would then file the cam and polish it, both smoothing the cam mechanically and smoothing the data mathematically. These operations required a great deal of time and skill, and a ballistic cam manufacture proved a major bottleneck in Sperry’s production of directors.” This is Mindell's jumping-off point for an argument with Noble’s Forces of Production and, more accessible, Vonnegut in  Player Piano. Far from deskilling the workforce, the introduction of industrial automation and numerical control created a new elite of skilled worker. Primitive programmers, doing code with files and and jeweller's rouge. (I imagine.) The Kerrison's main virtue was simplicity: only one ballistic cam. (David Mindell,  945–6,  224.) The Kerrison was adopted by the army in early 1941. It used a servo to drive the gun from the director, and now concerns arose over the hydraulic pump. “Firestone Tire and Rubber had a model in production, but it barely worked. Brown and the Servo Lab were working with a servo designed by the Oilgear Company of Milwaukee, which might replace the troubled British design. At this point MIT's Servo Group, and Australian-born lead scientist Gordon Brown got involved. Of this I will only say that historians remain too-often dependent on self-chosen sources. Brown emerges from all this, like Wiener in his own accounts, as a towering genius. That the upshort was, in the very latest stages of the war, a reliable "oilgear" of a kind that had been manufactured in Britain since the beginning of the war, is more telling of the understandable difficulties in mobilising industry for war than it is of the talents or lack thereof of any particular staff. 

What comes out of all this is two points. First, that the massive AA force that existed in Britain in 1940 did not come out of nowhere. It has wide ramifications back to industry and out of it. I've brought in shell manufacturers and armour forgers and American production difficulties in the interest of showing an intersection between the services and industry, of skilled labour moving from one sector to the other, and designs --and prejudices about what successful designs should look like-- in both directions. If France fell in 1940, might it be in part that this was not the kind of war that British industry wanted to fight? 

No comments:

Post a Comment