Three years before that Christmas, an unusual fault had appeared during testing at a plant in Indiana. A controller reacted to a changing load more slowly than the design specification allowed. It recovered before any equipment was damaged, but the intermittent delay concerned Nolan because it disappeared during routine follow-up tests. He spent two shifts repeating the conditions, then submitted a detailed report recommending more demanding verification procedures before the system was approved for broader industrial use.
The report passed through quality control, engineering, and the office of chief operating officer Preston Blackwell III. Nolan received an acknowledgment number and an assurance that management would review the cost of additional testing. Weeks passed without approval. He wrote again, this time including comparative data and a proposal for a limited trial that would reduce expense. Another acknowledgment arrived, followed by silence.
A year later, after a controller at another location behaved similarly under a rare load combination, Nolan submitted a revised request. He was careful not to overstate the evidence. The available data did not prove that the units would fail in ordinary use, and he had no interest in frightening anyone with a claim he could not support. What the data proved was narrower and more important: the routine test did not always demonstrate compliance with the response limit. If the company wished to certify that limit, it needed a more complete test.
That report disappeared into the same administrative corridor. His immediate supervisor, who had once encouraged questions, advised him to stop making trouble for a problem that might never produce a customer complaint. Nolan kept testing. He stopped expecting the reports to win him friends.
By the time the Great Lakes Industrial Consortium contract entered its final acceptance phase, WestBridge had transformed into a larger and more aggressive business. Hawthorne Tower rose above the Chicago River like a monument to the company's confidence, with glass elevators, quiet executive floors, and meeting rooms named for famous inventors. Analysts celebrated the proposed four-hundred-and-twenty-million-dollar agreement as a defining achievement, a chance to install control platforms across a network of industrial facilities and provide service for years. Preston was the face of the deal. His presentations made complicated engineering work appear as smooth and inevitable as a line climbing upward on a graph.
For Nolan, the agreement meant a new set of machines to test and a new list of specifications he had promised himself to respect. His team received the final verification schedule in early December. The testing window was narrow, and production crews had already arranged holiday staffing to meet it. Nolan did not object to the urgency. He knew that delays could hurt workers throughout the company, including people who had nothing to do with executive decisions. He objected only when the schedule began to be treated as proof that the results must be satisfactory.
On December 16, during an extended load test, a feedback signal arrived later than the contractual threshold permitted. The delay lasted only a fraction of a second, and the system continued operating. Nolan marked the result and asked for a repeat run. On the second attempt it passed. On the third, under a slightly different load, the delay returned, like an unwelcome acquaintance appearing just when everyone had decided the room was empty.
He brought the data to Marla Bennett, head of quality control, who studied the graphs through reading glasses balanced low on her nose. At fifty-four, Marla had spent enough years inside manufacturing companies to distinguish a genuine risk from an engineer's favorite theoretical argument. She traced the curve with one finger, compared it against the requirement, and frowned. "That isn't a clean pass," she said.