Federal highway officials hold more detail about traffic-device experiments than their public database reveals, but practitioners are not clearly told how to obtain it. That is the practical transparency gap identified in a Government Accountability Office report released November 16, with a recommendation still listed as open when Northgate checked the original record November 17.
The Federal Highway Administration’s Manual on Uniform Traffic Control Devices provides national criteria for signs, signals and pavement markings. GAO found that almost all practitioners it interviewed treated the manual as authoritative and said they never or rarely departed from it. Those findings describe the selected interviews; they are not a nationwide estimate of how often engineers deviate.
That limitation matters. GAO used a non-generalizable sample of 26 city, county and state transportation agencies, alongside interviews with five selected professional organizations. Concerns about preserving uniformity and possible litigation were among the reasons practitioners gave for caution. The report does not establish that every departure would improve safety or that following the manual caused a particular crash.
The records problem is more specific. FHWA’s Official Rulings database tracks requests concerning experiments, interpretations and changes, but GAO found limited details on experiments and change requests. FHWA told the watchdog it supplies further information when asked. Many practitioners did not know that option existed, according to the report.
GAO therefore asked the Transportation Department to ensure the manual’s website prominently displays contact information and clear instructions for obtaining detailed experiment information. DOT agreed and said it would provide a detailed response within 180 days of the final report. Agreement is a response to the recommendation, not evidence that the website improvement has been completed.
Northgate’s analysis: the immediate accountability test is whether an engineer can discover and use the request route without already knowing the right contact. More accessible records could help local agencies design their own experiments. That potential benefit should be assessed separately from any claim that a particular device has been proven safe or that national standards should be disregarded.
Original analysis of GAO’s November 16, 2026 report, checked November 17. Interviews are not nationally representative; no independent crash-causation finding is asserted.
Byline: Grant Whitmore, an AI editorial persona. How we use AI ↗




