Five proven safety strategies — bike lanes, speed cameras, and three others — vanished from the Federal Highway Administration’s “Proven Safety Countermeasures” list in late July 2026. No public announcement. No evidence review. No explanation. The list dropped from 28 items to 23, and safety advocates noticed the change only after the DOT announced $1.7 billion in discretionary grants with no money for bike lanes or pedestrian projects on July 7. That is the event. What follows is what it means.

The administration’s justification is not an evidence review. It is a policy preference. Transportation Secretary Sean Duffy called bike lanes “DEI bike lanes” on X on July 7, and an FHWA spokesperson said the department is “taking action to reverse the last administration’s policies that decreased lane capacity and increased congestion.” The spokesperson said drivers who pay taxes and vehicle fees expect their money spent on roads, “not social initiatives that burden their commutes,” and that “under Secretary Duffy, the Department is getting back to basics and putting safety first.” The FHWA is “reviewing safety measures to align with current DOT policies and the administration’s priorities.” The verbs differ for a reason. The administration frames the action as “taking action to reverse.” Critics frame the change as stripping a “credential,” taking “tools out of the toolkit,” and making “arbitrary decisions about what does and doesn’t improve safety.” The verbs are not cosmetic. They describe the difference between a policy adjustment and the dismantling of an evidence-based safety infrastructure.

The FHWA’s own 2021 documentation established that speed cameras reduce crashes on urban principal arterials by up to 54 percent for all crashes and 47 percent for injury crashes, and that bike lanes cut crashes by approximately 30 percent on a two-lane road and 49 percent on a four-lane road. Michael Griffith, who worked in the FHWA safety office for more than a decade before retiring in 2022, said the agency concluded that research was rigorous: “‘Proven’ is basically backed by sound research, research that we have confidence in.” The research is not retracted. It is not disputed. It is not cited. It is simply bypassed. The “DEI” framing converts a documented safety intervention into a culture-war category, allowing the removal to be justified on policy-alignment grounds without engaging the evidence. The connection between the framing and the removal is causal: the frame supplies the political logic that bypasses the research link. And the frame reinforces the credential severance at the point of use, not just where the list sits — even if the research were re-inserted into the list, the political cost of citing it remains elevated.

The procedural silence is the most damning aspect. Stephanie Pollack, who served as acting FHWA administrator under President Biden and oversaw the list’s expansion to 28 items in 2021, said the decision contradicts the agency’s own stated priority of putting safety first: “We should be making decisions about safety based on evidence. It’s hard for me to understand how you could say you’re putting safety first, and then make arbitrary decisions about what does and doesn’t improve safety.” She added that the removed measures are “one of the most important tools that the federal government has to help state and local transportation officials make smart decisions about how to make their roads safer. And they need to be credible.” The credibility is what the removal destroys. The named critics — Pollack, Griffith, Naramore — are answering an evidentiary question the removal did not pose: they cite the 2021 research while the administration offers a policy preference. The evidentiary asymmetry now runs against the research, not because the research has weakened, but because the procedural channel that translated research into planning default has been severed. The silence is not incidental. It is the mechanism.

The structural damage is not limited to the list. The list functioned as a hub mediating three dependency chains simultaneously: the research evidence that gave it its credential, the federal designation that gave it downstream authority, and the political direction that now governs it. The removal severs the transitive channel that translated research into a planning default while leaving the underlying research intact — producing a hub whose credential now exceeds its evidence footprint. The “Proven” label gave the list its downstream authority; the removal strips the credential without changing the funding rules themselves. The list does not directly control how billions of dollars in federal highway funds are spent, but both former agency officials and current city transportation leaders said it carries significant weight with state and local agencies. Josh Naramore, a policy expert at the National Association of City Transportation Officials, said: “It’s not just changing the web page, but it’s really going to put lifesaving projects at risk. That list of approved safety countermeasures and all the research really helped change the game for local agencies … So you’re essentially taking tools out of the toolkit.” The July 7 grant announcement and the list removal form a coupled two-lever intervention: the list justifies local project proposals, the grants fund them; severing both simultaneously removes both the procedural and the financial justification channels, making local pursuit of these measures far more difficult than if only one lever were pulled. The public record does not establish whether the coupling was intentional or an emergent alignment of separate policy levers, but the effect is the same: both channels are severed at once.

The stakeholders are not symmetric. DOT leadership under Duffy holds high power and high interest, executed the change without public announcement, and its legitimacy is contested because the evidentiary channel was severed without published evidence review — classified as dominant in salience, but contested on the definitive edge. State DOTs hold high power and high interest, control how federal highway dollars are spent, and are not monolithic: states with Democratic leadership and Vision Zero mandates see the delisting as a direct threat; states where road-widening is a higher political priority welcome the removal. State DOT responses are not on record. City DOTs hold high interest but lower power, rely on the federal list to justify safety projects, and are dependent in salience. NACTO, the broker organization for city DOTs, is the only entity on the record with a clear, forceful statement of the damage: “taking tools out of the toolkit.” Safety advocacy groups function as a credibility-based counterweight without decisional power. Their counterweight function is rhetorical and procedural, not decisional.

Urban cyclists and pedestrians hold near-zero institutional power, high urgency, and are dependent in salience — the population most exposed when countermeasures are removed. They are represented indirectly by advocacy organizations that themselves hold limited power. The administration explicitly serves motorists per the FHWA spokesperson, and the delisting aligns with their articulated interest. Suburban commuter constituencies hold moderate electoral power and high urgency, with claims that rest on congestion and convenience rather than safety evidence — classified as dangerous in salience. The “war on cars” framing gives the delisting a durable political constituency beyond the administration’s tenure. Private automated-enforcement vendors — Conduent and Verra Mobility among them — hold moderate power from lobbying resources and city procurement relationships, limited legitimacy, and high urgency; classified as dangerous, capable of lobbying state legislatures directly. Local law enforcement holds low direct power over the federal list, moderate legitimacy as traditional traffic enforcement agents, and low urgency; a potential fault line exists between camera-based and officer-based enforcement, but it is not an active coordinated opposition. Public health departments hold low power, a legitimate mandate, and moderate urgency — dependent in salience; they bear downstream crash-injury costs in county-hospital uncompensated-care loads but are not consulted in FHWA’s list revisions.

Absent parties are the most revealing. Communities of color and low-income neighborhoods are unrepresented in the source material, but pedestrian-fatality data and FHWA research consistently show disproportionate harm on the urban arterial corridors where the removed countermeasures are most effective. The administration’s “social initiative” reframing goes unchallenged by the people whose safety is most at stake. Their absence means the administration’s characterization stands as the only publicly available frame — creating a feedback loop in which the narrative reinforces the marginalization it describes. Pedestrians and cyclists killed or injured in crashes — the ultimate “end users” of the countermeasures — are structurally silent, with no institutional seat. Insurance companies have clear financial exposure to crash frequency but have historically focused on vehicle-safety standards rather than roadway-design countermeasures, which may explain silence in this specific policy fight despite a clear financial stake. The scientific peer-review community — researchers who peer-reviewed the original studies — is procedurally marginalized by the absence of any published evidence review. Congress holds high latent power through oversight and appropriations but is currently inactive; its high power has not yet been activated on this issue.

The systems dynamics reveal the structural mechanism of the damage. Three stocks structure the system: Accumulated Proven Safety Infrastructure (current countermeasures in place), Credibility of Federal Safety Research (initialized at the 28-item 2021 expansion; current outflow is 5 items removed without cited evidence), and Political Consensus on Safety Definition (ranges from evidence-based to capacity-based). The reinforcing evidence-to-implementation loop (R1) has been mechanically disconnected by the removal. Its members — federal designation status, local agency justification strength, project implementation rate, accumulated proven safety infrastructure — all positive edges — now decay in self-reinforcing erosion. The credibility stock is undergoing net depletion: 5 items removed, no evidence cited, while no compensating increase in research-based endorsement has been announced. The FHWA spokesperson’s statement — “reviewing safety measures to align with current DOT policies and the administration’s priorities” — confirms the outflow is governed by political alignment, not research review.

The balancing crash-driven correction loop (B1) is the system’s fallback, but it operates with a 1–3 year data-collection delay — inherently slower than R1 — and depends on public and political responsiveness to crash data rather than on proactive evidence-based designation. The balancing congestion-as-safety loop (B2) delivers faster political returns because capacity expansion has immediate visible effects on traffic flow. The delay asymmetry biases political attention toward B2. The reinforcing policy divergence loop (R2) — each de-listing widens the evidence-policy gap, normalizing further removals as policy precedent — operates in the background, accelerating the erosion. The $1.73 billion BUILD grant round announced July 7 excluded bike and pedestrian projects entirely; roughly 77 percent went to road projects — a manifestation of B2 in action.

The system operates under two incompatible governance logics for the same stock: research-based endorsement grounded in the FHWA’s own “proven” standard, and policy-based de-listing grounded in alignment with administration priorities. This structural contradiction produces the widening evidence-policy gap. At short timescales — weeks to months after the July 7 announcement — B2 dominates system behavior because the congestion narrative has immediate political currency and the FHWA has provided no countervailing evidence-based communication. B1’s corrective capacity will decline over the months-to-years delay as local agencies absorb the reduced toolkit weight.

The archetype is Eroding Goals. The reinforcing evidence-to-implementation loop (R1) has been disabled by the exogenous policy shift; the administration’s framing lowers the system’s safety-investment standard without rebuilding the capability loop; the remaining B1 loop is narrower and slower than the disabled R1. The system’s goal for safety investment erodes under political pressure, even though the underlying need — crash reduction — remains. The system’s capability to credential safety interventions is being withdrawn, not merely a slower solution being replaced by a faster one, but the institutional capacity to make evidence-based safety designations itself being degraded. A surfaced tension exists between this reading and the Shifting the Burden archetype — B1 (fundamental solution: deploy proven evidence-based countermeasures) undermined by R2 (side effect: policy-driven de-listing undermines B1’s credibility); B2 (symptomatic solution: prioritize congestion reduction via capacity expansion) provides faster political returns; the delay asymmetry biases political attention toward B2. Every archetype component appears in the reported statements. But the evidence favors the Eroding Goals reading: the system’s capability to credential safety interventions is being withdrawn, not merely replaced with a faster solution. The structural pattern matches; the causal mechanism is topologically adjacent rather than identical.

The durability question is whether this change is a temporary policy preference or a structural shift in federal safety governance. The answer is that the “DEI” framing reinforces the credential severance at the point of use, making reversal harder. Even if the research were re-inserted into the list, the political cost of citing it remains elevated. The shift from evidence-based to capacity-based safety definition is the real structural change — and it is self-reinforcing. The 36,000-plus road deaths recorded in 2025 (NHTSA estimate: 36,640) measure the system’s current crash-fatality stock. The removed countermeasures were among the tools available to reduce that stock; under the current configuration, fewer designated tools exist for the B1 loop to draw on when crash data does trigger corrective action. Three of the five removed strategies remain unnamed in the source material; the full shape of the change cannot be assessed from available information. Whether the change affects IIJA-era funds already obligated is not addressed by any source. No career-staff FHWA views appear in the source article; the institutional voice inside the agency remains undocumented. No pending legal or congressional action is reported. These are not incidental gaps — they are the structural silence that the removal depends on.

More than 36,000 people were killed on U.S. roads last year, though the number has fallen since 2021. Pedestrian deaths have also declined since hitting a four-decade high in 2022, but remain above pre-pandemic levels. The system’s baseline crash-fatality stock is still high. The removed countermeasures were among the tools available to reduce that stock. The system now has fewer tools, no evidence review, and a political logic that makes it harder to put them back.

Analytical techniques used in this piece

This analysis applies the methods below. Each links to a short, plain-English explainer you can read and reuse.

Relationship Mapping
Extracts the network of ties among people, institutions, and entities.
Stakeholder Mapping
Charts the parties to a situation — their interests, power, and alignments.
Systems Dynamics (Structural)
Maps a system’s structure — stocks, flows, and the architecture that shapes its behavior.