The Giddings & Lewis Episode
In the early 1970s, at the Giddings & Lewis machine tool plant in Fond du Lac, Wisconsin, a group of machinists began using a technology called manual data input (MDI) to write and edit numerical control programs directly at their workstations, without routing the programming through a separate engineering department. The machinists were not credentialed programmers. They were production workers who understood the materials, the machines, and the work. Their programming was often faster, more accurate, and more responsive to actual shop conditions than the programs produced by the remote engineering department. Noble's documentation of management's response to this development — restricting access to MDI consoles, requiring engineering-department review of machinist-written programs, and in some plants removing MDI capability entirely — became one of his most telling case studies of how alternative designs are actively suppressed when they distribute power in ways management finds threatening.
In the AI Story
Manual data input was a straightforward technical capability: a console mounted at the machine that allowed the operator to enter and modify programs using a simplified language. The technology was cheap, available, and had been included in many NC machines as a debugging feature — a way for programmers visiting the shop floor to correct small issues without returning to their offices. What happened at Giddings & Lewis and similar plants was that machinists recognized they could use this feature to program whole jobs, bypassing the programming department entirely.
The results were revealing. Machinists who wrote their own programs could adjust them on the fly as material conditions varied, test modifications immediately, and iterate toward optimal parameters in hours rather than the weeks that remote programming required. The programs were often technically superior because they incorporated the machinist's embodied knowledge of how specific machines actually performed, knowledge that the engineering-department programmer did not possess. Production scheduling improved. Quality improved. The organizational hierarchy that numerical control had created began to look optional.
Management's response was not celebratory. It was defensive and systematic. Access to MDI consoles was restricted to authorized programmers. Machinist-written programs were required to be reviewed and approved by the engineering department, which typically took longer than the original writing. In several plants, the MDI capability was removed entirely — often at additional cost, because the feature had been standard on the purchased machines. The restoration of the centralized programming model was pursued deliberately, against the technical and economic advantages that decentralized programming had demonstrated.
The episode matters because it demonstrates that the separation of conception from execution in numerical control was not a technical necessity that the machinists had wrongly tried to reverse. It was a political arrangement that the machinists had revealed to be optional, and management had reacted by reimposing the arrangement against the evidence that it was less productive. The reimposition was rational from management's perspective: the productivity loss was less significant than the loss of control that widespread machinist programming would have represented.
Origin
Noble documented the Giddings & Lewis case through shop-floor interviews, company records, and union grievance files in the late 1970s, publishing the findings in Forces of Production. Similar episodes occurred at other plants — Sundstrand, Cincinnati Milacron, and several aerospace subcontractors — and the consistency of the pattern across companies with different managements was one of Noble's strongest pieces of evidence that the suppression was structural rather than idiosyncratic.
Key Ideas
Machinists programmed better. When given the tools, the people who understood the materials and machines produced programs that were technically superior to those produced remotely.
Management suppressed the practice. The response was systematic and deliberate: restrict access, require review, remove the capability, restore the centralized model.
Political, not technical, rationale. The suppression was rational from management's perspective because it preserved control even at the cost of productivity.
Demonstrates the pattern is reversible. The separation of conception from execution is not a technical necessity; it is an institutional arrangement that workers can in principle reverse — and that management can in principle re-impose.
Appears in the Orange Pill Cycle
Further reading
- David Noble, Forces of Production (Knopf, 1984), Part III
- Harley Shaiken, Work Transformed (Holt, 1986)
- Mike Cooley, Architect or Bee? (Hogarth, 1980)