The register lists 99 federal defect investigations as currently open. Sixty-two of them were opened this decade and are real, active cases, the recalls that may come next. But 28 were opened in the 1990s, and the oldest of them has now been open for more than thirty years. This article is about those 28, because they answer a question the register kept forcing me to ask: what does an open investigation mean when nobody has touched it since the Clinton administration? The answer turns out to say less about car safety and more about what happens to paperwork when everyone who cared about it retires.
The shape of the pile
Line up all 99 open dockets by the decade they were opened and the pile has a hole in the middle: 28 from the 1990s, five from the 2000s, four from the 2010s, 62 from the 2020s. Modern investigations open and close; ancient ones just sit. And the ancient ones share a second property that is the whole story. The federal agency's dockets come in flavors: preliminary evaluations and engineering analyses, the real defect investigations with teeth, and then the administrative genres, recall queries that audit whether a completed recall worked, service queries that ask a manufacturer about a technical bulletin, and defect petitions filed by outsiders. Every one of the 28 zombies is from the administrative genres. Not a single ancient docket is an actual engineering analysis; those all closed, every one, as the register's durations study found. The agency closes its investigations. What it forgets are its questions.
The oldest question in the file
The most ancient open docket in the American car safety system was opened on January 10, 1996, and it concerns a brush guard. SQ96009 asks Isuzu about a campaign involving grille guards on mid-90s trucks, on the reasonable ground that a large metal object detaching from a moving vehicle would have safety consequences. That is the entire matter. It has now been pending for more than thirty years, which means this question about an accessory for a truck that is no longer made, from a brand that no longer sells passenger vehicles in America, has outlasted the truck, the brand, and most of the trucks' working lives. Its neighbor SQ96012, opened that June, concerns the rod that holds up an Oldsmobile trunk lid. Oldsmobile itself closed in 2004. The docket did not.
The docket that went to court and never noticed
One zombie is famous. DP96011 is the Consumers Union petition asking the agency to investigate the 1995-96 Isuzu Trooper for rollover risk, the opening move of one of the loudest product-safety fights of the decade. Consumer Reports had rated the Trooper Not Acceptable; the petition landed at the agency; and the docket's frozen summary still records the state of play in early 1997, an attorney at Arnold & Porter, new material submitted in December, a review deadline moved to April 21, 1997. What happened next is thoroughly documented, just not here. The agency's petition analysis, dated June 1997 in the court record, criticized the testing methodology and the petition was denied. Isuzu then sued Consumers Union for defamation in federal court, and the case ran for years. The dispute was resolved in the agency's files, resolved in a courtroom, and litigated into the case reporters, and the docket that started it all still reads as open, awaiting a deadline that expired while Titanic was in theaters. Its sibling DP96014, a petition about airbag sensors that its filers amended to cover airbag aggressivity, preserves the other great safety argument of 1996 in the same amber.
Mission accomplished; file still open
My favorite genre among the zombies is the docket that worked. RQ99020 asked Toyota about steering wheel detachment on a 1995 model. Its entire summary, verbatim from the federal file, reads: OBATINED RECALL 99V-307. Typo and all. The query did its job, the recall happened, the paperwork announced its own success, and then nobody wrote the close date. It has been open, victorious, for twenty-seven years. SQ99018 is the same story with a border crossing: it opened because Nissan was recalling Sentras in Canada for coil springs that fracture from road salt corrosion while doing nothing in the United States, exactly the cross-border asymmetry the register's Canada article is about, and the pressure worked, a US recall followed in 2000. The docket that extracted it never closed. And a matched pair of queries about corroding GM engine cradle bolts, RQ97017 and RQ99002, sit open with a completion audit frozen mid-sentence: 59.3 percent of vehicles repaired as of the sixth quarter. The seventh quarter never arrived.
The questions no one can answer anymore
Several zombies are unanswerable in a more final way: the company they were addressed to is gone. SQ99024 asked Excelsior-Henderson, the Minnesota motorcycle startup, about a fuel line that could disconnect and stall the bike; the company had agreed to conduct a recall, says the summary, and then collapsed into bankruptcy within months of the docket opening, building its last motorcycle soon after. The question remains formally pending, addressed to a company that has not existed for a quarter century. The two Victory motorcycle dockets outlived the Victory brand, which its parent shut down in 2017. Others are addressed to the small-manufacturer fringe of school bus builders and RV makers, Carpenter, Damon, Holiday Rambler, where companies dissolve, merge, and vanish, and their dockets, having no one left to answer them, become permanent.
The register's own fossils
Because this site's whole method is trusting the federal file, honesty requires showing where the file itself has fossilized. RQ98010 is filed as an investigation into braided natural-gas fuel hoses on Blue Bird buses; the summary attached to it describes, in loving detail, a taillight recall on the Mercury Villager and Nissan Quest. That mismatch, presumably a data-entry slip in 1998, has now been preserved in the federal record for twenty-eight years, uncorrected because correcting it would require someone to open the file, and nobody opens these files. Several others carry the note There is no summary currently available, which on a docket from 1997 has achieved a kind of permanence the phrase never intended. The register reports these dockets as open because open is what the record says, and the record, as this site keeps finding, is an instrument with its own optics.
What open means
Here is the honest reading. The 28 zombies endanger no one and protect no one; they are the sediment left when the agency's paperwork practices changed around the turn of the millennium, visible in the data as a cliff, 28 undead dockets from the 1990s, then almost none, then the modern era, which closes its files. An open investigation from 2026 is a live question about your car. An open investigation from 1996 is a fossil of one, and the difference matters if you ever look up a vehicle and find an ancient docket attached: check the date before you check the defect. The current, breathing investigations, the ones that may become recalls, are cataloged in the open-docket tour, and they are on average four years old, not thirty. The zombies will presumably remain open forever, since closing them would require the one thing they have never received: attention. The register will keep counting them, because the register counts what the file says, and will keep noting, in articles like this one, what the count actually means.