the war on zombie jesus and his legions of pastel painted pagan bunnies...,
| Main |
With absolutely zero regard fo...,
[I won't remember where I put this if I set it aside, so I'll just post it and be embarassed into finishing this thought later. Nobody reads this fucking thing anyway so I don't have to worry about that either. Everything's coming up Milhouse.]
David Noble, Forces of Production, p240:
A similar pattern was followed in most plants surveyed by the Center for Policy Alternatives. At the Torrin Company in Torrington, Connecticut, a manufacturer of spring-winding machinery, the management
believed that "N/C [ed: numerical control] people are potentially less skilled" and this means "less pay." Here
as elsewhere, much of the N/C programming was relatively simple, and supervisors were asked why the operators were not doing
their own programming. The supervisors argued that the operators would have to know how to set feeds and speeds,
that is, be industrial engineers. When it was pointed out that the same people knew how to set feeds and speeds
for conventional equipment, routinely making adjustments on the process sheets provided by teh methods engineers in order to "make out,"
the supervisors agreed. But they claimed that the operators could not understand the programming language.
The interviewers noted that the operators in the shop could often be seen "reading" the tape, to anticipate
programming errors, and this indicated that they could at least learn to program (having had to figure it out
the hard way, and backwards). The supervisors finally conceded: "We don't want people on the floor making
programs," and added, "these guys can beat you in a thousand ways. N/C eliminates some of the ways they can
At the TRW in Cleveland, the management "brought in N/C as a good way to lower skills." Originally,
the "philosophy was to deskill." Operators were not allowed to "mess with" the programs, on the grounds that
"too many cooks spoil the stew." N/C was used to set time standards for operators, "With N/C, you can time a
job via the computer postprocessor," one porduction manager pointed out, "and know the exact machine cycle time,
as programmed." This "tape time" determines the rate, "You can break pacing with N/C," he added. "Once the
tape is in, and the override is locked," another manager remarked, "the operator is helpless - all you need is a
robot." A reduction in director labor of 20-25 percent at non-union Cincinnati Milacron was made possible
with N/C, where managers were using the technology to transform the management art "into a science." At Hamilton Standard
in Hartford, Connecticut, the direct shop floor work force was cut in half between 1969
and 1979 (while output increased each year). N/C displaced workers, one manager insissted, "and anybody
who tells you different is a liar." "A guy has to be an idiot if, when he sees as machine that works by itself,
he does not see his job threatened." Here too, management wanted "to eliminate all decisions from the floor." One shop
supervisor indicated that he did not "want the operator to make the decision to override or to mess around with
the programming." "If programming is done by operators you have to pay them higher rates." "I don't want
the guy making the decision himself, he has to get the forman's okay for all, even minor changes." At
the same time, the shop supervisor expected instant wisdom on the part of operators in the
event of an emergency. "I need guys out there who can think," he insisted, acknowledging the
expense of the equipment. But this wisdom was never formally acknowledged, nor compensated for. Officially, the
operators needed no skill - and, in practice, niether did the foreman. "You don't have to know
how to run the machine to manage it," one foreman intimated. "To be honest with you, I can't even start the damn
things, but it doesn't matter. I know who to call when things go wrong."
I've been reading through this book for some time and still can't really comprehend why I enjoy it so much when I pick it up. If I had more energy I could explain it better, but let's give it a very rough draft in the waning hours of my evening: It's dry as bone and little passages like this are about as colorful as it gets, and this one in particular was wearing punch gloves of absurdity that KOed me on the train. Makes ya laugh and cry.
The technologies were developed around the same time shortly after the war by numerous gadgeteers: a variety of forgotten machinists, engineers, and small businessmen that typically had had a good deal of shopfloor experience with the tools they wished to automate. These were in turn either shut out or stolen blind by state and corporate interests: Record/Playback systems or machinist programmed tools like the Specialmatic were excluded from defense industry supports. Lucrative Air Force contracts, with garunteed profits based off - as they still appear to be - a simple percentage of cost were the only means by which machine automation ever occurred at all, or, in the negative sense of suppressing alternatives, the primary reason why it took so long to be adopted.
Air Force Material Command imposed demands on contractors to overcapitalize in a highly advanced system unnecessary for all but a minor fraction of machine work required - the rationalization being a matter of command and control: in the wake of a hypothetical, catastrophic attack the AF would require production to be shifted to their other contractors, and so demanded that any shop that wanted defense work would have to be able to immediately take over the damaged work, including the few parts that required the full capabilities of the Air Force's - to abbreviate a great deal - computer run N/C APT-standardized punch card system. 
Because more economical designs recieved no such public subsidization in their development and faced a hostile, fearful industry who's primary interest in automation was wresting control of production from the shop floor (even in non-union shops skilled machinists were considered to hold an unacceptable amount of influence over 'business' - the complaints ranging from "pacing" to limiting engineers' designs to the realities of production), they were never developed fully - in the United States at least: by the 70s economical R/P and N/C systems had been developed abroad along the same lines as those early American inventors had envisaged - some even openly acknowledging their earlier designs - and the US became an importer of automated machine tools, its own industry standard being untenable in the marketplace:
Meanwhile, foreign machine tool manufacturers concentrated on producing equipment for the commercial market. Fujitsu Fanuc, a leading Japanese machine tool builder, in 1973 alone produced more N/C machines designed for the commercial market than all US machine tool firms combined. Likewise in West Germany, machine tool builders concentrated upon the commercial market. Accodring to Paul Stockmann of Pittler - a central figure in German N/C development - German manufacturers were locked out of US military contracts and the APT Program and found, besides, that "no one was interested here in a highly sophisticated program which required access to a big computer." Instead, manufacturers focused upon less expensive and less demanding programming methods, and designed their cheaper machines accordingly. Not surprisingly, with domestic machine tool builders tied up with military and aerospace industry orders and specifications, foreign manufacturers were able to gain a significant foothold in the US commercial market. Between 1960 and 1975, US imports of machine tools increased 30 percent. By 1978, the US had become a net importer of machine tools; Japanese machines accounted for one-third of these imports and West German machines accounted for one-fifth.
The Air Force's demands managed to create the most inefficient (for the vast bulk of machining requirements), costly standards who's only demonstrable "benefit" that I've figured out so far, in the long run, was its utility in waging an inexplicable class war against highly skilled machinists and their co-workers, which in turn ran in the interests of a new group of non-management, highly skilled workers... programmers.
- Recall that this discussion spans, primarily, the three decades after world war two, and that computing power was incredibly expensive. The details of the expense of the Air Force system deserves some further exposition:
The APT programming standard, in effect, standardized the programming of machine movements. Simpler programming languages had been developed - e.g. NUMIFORM - that allowed geometrical instructions to be input into a processor that produced the tape to run the machine, and later conceptually similar tools such as John Parsons developed in the early 60s, in which the piece's geometry was 'drawn' into a computer that displayed an image of the piece and then produced the program tape. Some of the readers of this might compare it to programming in an Assembly language versus programming in C.
The punch card standard, compared to magnetic tape; continuous path control vs. point control; etc. In every instance the Air Force chose the most complex, expensive alternative: tax payer money, of course, wasn't a consideration.
- [footnote looking for a leg] There's an episode that comes to mind, p183-187, based on interviews with Steve Heid (a Ford engineer) and Ralph Kuhn -
a tool and die supervisor for Ford Motor Company.
Kuhn in literally a matter of a few days converted Ford's loss-leading FORSUR programmed N/C system into a functional R/P system for 500 bucks plus past, non-functional, investment. It deserves to be recounted more fully, and I'd prefer not to gloss over the specifics, so:
Kuhn was allowed to set up a low-key five-hundred dollar project to test out the motional method on typical parts, such as a window regulator and an inner car door. He adapted the DEA inspection probe with urethane balls of varying sizes
to simulate different-sized ball cutters used in the actual milling. By using the balls to trace the casts, Kuhn was able to
genereate a cutter path directly, without having to derive it laboriously from surface information. He also added basic machine commands
to the generated cutterpath information and thus produced an N/C tape capable of producing the parts. He found the programming to be extremely accurate, relatively cheap and quick, and totally accessible to tool and die machinists. It took Kuhn only ten minutes in February 1972 to prepare the tape for the window regulartor and begin machining, whereas FORSUR programming required the preparation of a detailed planning manuscript, then coding, keypunching, computation, and plotting and testing of the completed tape before it could be used to run a machine tool - all of which could take days, counting delays. Yet, for all its obvious advantages, the new method was never adopted by Ford.
Opposition to Kuhn's idea came from several quarters and for different reasons. The Ford management, first of all, was no more disposed to having machinists prepare the programs for profile milling than they were to having them do it for straight cut milling. The motional method was opposed precisely because it lent itself to shop floor programming by workers rather than engineers.
[...very interesting commentary from Kuhn, actually, but here we are saving time by snipping it: we work in the morning...]
Finally, "the N/C Staff Group had spent untold millions" creating the FORSUR format for generating shapes and niether they
nor the managers and executives who approved of this massive expenditure were dispose to admit they had perhaps made a colossal mistake.
Thus, reflect what Kuhn referred to as the "corporate mentality" or the "cya syndrome" (cover your ass), they constructed a
"massive cover-up." Kuhn's immediate superior, who had allowed him to experiment with the inspection machine, was transferred,
Kuhn himself was ordered to abandon the project, and those few knew about it knew also to keep quiet if they wanted to hold on to
their jobs. Thus, there was no public disclosure, no papers published in the proceedings of the Society of Automotive Engineers or the
Society of Manufacturing Engineers, no demonstrations for equipment manufacturers or suppliers. Kuhn tried to convince DEA, the
manufacturer of the inspection equipment, to promote the "tracer-record" application but without success. (Some years later,
DEA did finally begin to promote the use of its machines for programming, but ran into many of the same obstacles as had confronted
Kuhn.) Kuhn thus remained a self-described "voice in the wilderness." "For twenty years these guys have been blowing it," he observed
in 1982, alluding in frustration to the demise of the US auto industry. "For twenty years we've been on the wrong track."
:: posted by buermann @ 2006-04-19 01:50:54 CST |
I own that David Noble book. But it's too long and boring. Of course, I find this post to be too long and boring. I'll read it later :)
posted by sansfrontieres
@ 2006-04-19 18:37:00 | link
A real page turner, I think. Shortly after this bit there's a story about a company in Lincoln Nebraska that actually hired a mentally retarded guy to run their N/C machine: during an interview the shop manager explains why, "It's the kind of tedious work that some non-handicapped individuals might have difficulty coping with."
Examples like this demonstrate why and how American workers need more retraining to recover jobs lost to technology, capital flight, and trade :P
posted by buermann
@ 2006-04-20 10:53:26 | link