Prove It for Twenty Years

Definition

Bell Labs' rule for undersea cable engineering: an innovation at the system level (a transatlantic phone cable had never been built) should be assembled from the least innovative possible components. Mervin Kelly calculated the 1956 transatlantic cable, TAT-1, had to run without a single failure for twenty years to justify its roughly $42 million cost, since a mid-ocean repair could take months. So the cable's underwater signal repeaters used vacuum tubes designed in the late 1930s — some of which had already been running continuously in a Labs test facility for sixteen years — rather than the brand-new transistor, because "there was no telling how long it would last underwater."

In the Book

Chapter Eleven follows Kelly's TAT-1 planning: decades of prior data from shorter test cables (notably Key West to Havana), meticulous stress-testing of components (telephone poles buried in swamps for twenty-five years to record decay rates), and a design philosophy summarized in an internal Labs history — "compromise was essential but innovation and relatively untried methods were too risky." Kelly, the executive who had spent the prior decade pushing the lab toward its riskiest, most speculative research (solid-state physics, the transistor itself), drew a hard line at deployment: the transistor was excluded from TAT-1's repeaters entirely. The cable worked flawlessly for twenty-two years after opening in 1956.

Why It Matters

It's a rule for where the frontier of innovation belongs and where it doesn't: push the newest ideas hard in research, where failure is cheap and recoverable, but keep them out of the parts of a system where failure is catastrophic or unrecoverable, until they've accumulated the track record the failure cost demands. It's a useful check against the instinct to reach for the most advanced available component just because it exists.