Javier had been at Harrison Motors for seven months. He was a Cost Reduction Engineer II, which meant his job was to look at things and try to make them go away. A vehicle contained about thirty thousand parts, and most of them had been there, in that exact or near-exact form, for decades. Some of them didn’t need to be: a clip could be plastic instead of steel, a weld could be two instead of three, a gasket could be thinner.
The part first came to his attention through a teardown of a competitor’s sedan. Every year, Harrison bought one of every new vehicle on the market, took it apart, and laid the pieces out on long, foam-lined tables, side by side with their Harrison equivalents, so the engineers could compare. On the competitor’s firewall, bolted behind the master cylinder, was an S-shaped bracket of stamped steel, about the size of a playing card, bent along two axes. On Harrison’s firewall, bolted behind the master cylinder, was an S-shaped bracket of stamped steel, about the size of a playing card, bent along two axes. Just a bracket, bolted to the firewall.
“What’s that?” Javier asked a tech.
He glanced over. “Firewall S-bracket.”
“What does it do?”
“It’s a firewall S-bracket,” he repeated, impatiently.
In the drawings, the part appeared on every Harrison platform going back to 1957. Its part number was F-4731-B-02. It had been reissued during the 1984 metric conversion and assigned a new part number; the old number, whatever it had been, was not in the system. The drawing of F-4731-B-02 itself was readily accessible. In the notes field, in the cramped shorthand that had once been standard, someone had written “A/B HG. ISOL. ref. SPM-1957 pg138.” It was, so far as he could find, the only document in which the part had a name.
He requested SPM-1957 from the document archive. Someone over there called his desk phone directly about three minutes after he’d submitted the ticket. The archive only went back to 1969, they explained; anything before that was in a storage facility that was, for all intents and purposes, not accessible, and in any case the index indicated that the 1957 Standard Practice Manual had been part of a “water event” in 1991, along with a great many other things nobody had since missed.
He asked the senior powertrain engineer. “Oh, that thing. Yeah, that’s always been there.” He asked the senior body engineer. He said, “Yeah, that’s a... that’s for something or other. Oh, know what, you should ask ol’ Sobieski.”
Sobieski was 83, retired, lived in Sarasota, and answered his phone (“howdy-howdy”) on the second ring.
“Behind the master cylinder? That thing? Oh, sure. Phil — boy, I haven’t thought about, gosh, it’s been, ha ha — Phil Malden woulda drew that, or anyway his team did. Now, Phil, oh gosh, had to be, he died in ‘84, maybe ‘85.”
“Do you know what it’s for?”
There was a pause, during which Javier could hear a game show in the background.
“You know,” he said, “I don’t know that I ever did.”
Of course, there was a protocol for changes like this. The summary deletion of a part had to be validated across nineteen distinct areas, from crash dynamics to salt-spray corrosion to metallurgical compatibility to cold-start behavior and beyond. If the bracket really did nothing, it wouldn’t affect any of them.
It didn’t affect any of them. After six months of testing, the data was so unremarkable that it looked falsified, and Javier had to add a memo to the effect that no, really, the bracket just was not doing anything.
The savings, he estimated, were $1.87 per vehicle. At 2.1 million vehicles produced per year, that was $3.9 million. All the detective work had been worth it — Javier had made his bones for the year; he’d be far enough ahead of his performance goals that he could technically play Minesweeper until the end of Q4 and still get “excellent” on his performance review and hit the bonus cap.
The review committee balked. “What if it does something we haven’t tested?”
“We’ve tested everything we can think of,” Javier said.
“Well, exactly. If we lost track of what the damn thing is for, maybe we lost track of the validations that prove we need it.”
He understood the objection, thought about it for a few days, and then he wrote a second memo, wherein he put forth the following argument: if we decline to remove a part because we do not know what it does, we have assigned infinite value to the unknown. If we cannot bring ourselves to remove this, we have essentially decided that nothing installed before living memory can ever be removed again. We have decided, in other words, that every generation of engineers going forward will be the custodians of somebody else’s half-remembered decisions, rather than makers of their own.
The memo went to the VP of engineering, who forwarded it to the VP of manufacturing, who forwarded it to the CEO, who wrote back: “Remove.”
The first production vehicles without the bracket rolled off the Warrensburg line later that year, entirely without incident. There were no lawsuits, no recalls, nothing weird showing up in fleet telemetry. Customer satisfaction scores and warranty repair stats were typical. The part had so outlived its purpose, whatever that may have been, that its disappearance made no mark at all.
Javier got his bonus, and then another bonus and a steak dinner for two once the CEO’s admin folks had traced the memo back to him. A couple years later, he moved to the generator-and-industrial division, and then to an electric aircraft startup in Grenoble. He did not think about the firewall S-bracket very often, but occasionally, when he read about some old ritual or phrase whose origin had been lost, he thought of Phil Malden, in the ground since the Reagan years, and the little stamped bracket he had drawn for reasons that had gone into the ground with him.
In 2041, a junior engineer at Harrison-Kinetix named Abhay Pillai was assigned to draw up a few bits of the reference geometry for the new unified platform that would underlie the company’s next generation of vehicles. Harrison-Kinetix, like every major manufacturer, maintained an enormous internal library of approved components that could be dropped into new designs without requiring fresh validation. Under “Firewall, Auxiliary Hardware,” Abhay found a part called RTN-4731-BR-232, listed as “ABHG Isolator.”
The entry dated to 2034, from the post-merger consolidation. Harrison Motors and Kinetix Industries had combined their engineering assets under a 400-page integration protocol that specified, among many other things — and this is a gross oversimplification of a seven-page directive — that where only one predecessor company carried a part in a given functional role, the part was retained in production. Harrison had retired its firewall S-bracket over a decade before the merger. Kinetix had never retired it. In part, this was because Kinetix, as a newer company, had built its first platforms by licensing reference geometry from Harrison, and the bracket had been in that reference geometry, and no one at Kinetix had ever since had cause to look at it, let alone been stubborn enough to ask what it was for. The Kinetix part had been preserved, renumbered to RTN-4731-BR-232 under the unified nomenclature, and that was that.
None of this history was in the note field Abhay saw. The note field cited a supplier drawing, which cited an SAE reference, which cited a 1998 cross-manufacturer study — a pedigree that looked, at a glance, entirely respectable. None of the citations, taken together, quite added up to a full explanation of what the part was for, but it was fully specified, had been on thousands upon thousands of vehicles without incident, had never been even remotely implicated in a single recall or defect, cost almost nothing, and Abhay had forty or fifty other parts to place that afternoon.
He dragged it onto the firewall, behind where the master cylinder used to be.
The platform went into production, and before long, RTN-4731-BR-232 was on every new vehicle across four continents — not only, as it turned out, Harrison-Kinetix’s, since part of the library was shared through an industry working group established in the early 2030s. Bolted to firewalls, behind where the master cylinder used to be, doing nothing that anyone currently alive had ever been able to detect, measure, theorize, model, or simulate.
Somewhere, there was a folder with Phil Malden’s initials in the corner. He had drawn the bracket one afternoon in 1957, for a reason that had made complete sense at the time — so much sense, in fact, that he hadn’t bothered to write it down, because it was one of those things that everyone who needed to know already knew — and then the reason had died with him, and the bracket had not.