Software likes to treat borders as a settings pane. Robotaxis keep finding out that liability is not. So we get a trench war fought in permit hearings, state letters, and federal investigations. The point is not a final verdict on autonomous vehicles. It is who gets to see the data, and when.

The D.C. Test: Who Sets The Onramp

Washington, D.C. is a tidy case study. A proposed bill would allow autonomous vehicles to operate in the District, and it has already sorted the industry into uneasy camps, according to TechCrunch Mobility. Uber opposes the bill, arguing it would displace for-hire human drivers and give Waymo a de facto monopoly. Uber has lobbied for a hybrid setup that would require robotaxis to operate on a ride-hailing network alongside human drivers. Insiders quoted by TechCrunch say that approach has little chance of becoming law, but the intent is clear enough.

Waymo largely supports the bill, and it has already met a pair of thresholds the proposal contemplates. The company has at least 180 days of testing and 250,000 miles in the District with human safety operators. If the bill passed as written, TechCrunch reports Waymo would enter the market with a head start of at least six months. Tesla, through a senior policy adviser, objected to the 180 day and 250,000 mile mandatory testing requirement, the size of the application and permit fees, and a per mile tax. Several companies argued that testing miles accumulated in other jurisdictions should count toward the mileage threshold. The D.C. Council hearing drew Lyft, Tesla, Uber, and Waymo, plus disability rights advocates, safety groups, unions, and think tanks.

You can read that lineup two ways. One is market strategy. The other is governance by evidence. Days, miles, and money are the proxies on the table. They are blunt instruments, and blunt instruments invite carve outs. The better instrument is data that lets the city, competitors, and the public examine what went wrong and what improved from one software release to the next.

State Power, City Problems

If D.C. is about the onramp, California shows what happens once you are in the stream. San Francisco mayor Daniel Lurie urged state regulators to toughen autonomous vehicle rules after Waymo robotaxis became immobile in heavy July 4 traffic, ran out of power, and blocked key streets, according to TechCrunch Mobility. The detail that matters is procedural. The mayor asked state regulators to act. Cities feel the operational pain, but state agencies often control the permits.

That gap is where transparency either grows or dies. Cities want assurances tied to lived reality, like how fleets handle edge cases during extraordinary events. States want uniform rules that apply across jurisdictions. The bridge is not a press conference. It is logs, incident reports, and release notes that let state staff check claims and let city officials point to requirements that map to gridlock or clearance times.

The battleground is not permission, it is proof.

Federal Scrutiny Reads The Black Box

The federal lane looks different, but the theme repeats. The National Transportation Safety Board said the driver in a June Tesla crash pressed the accelerator to 100 percent, overriding the company’s Full Self Driving Supervised software. That is a reminder that federal investigators reduce brand narratives to timestamped inputs and outputs. It is also a reminder that autonomy is not a monolith. Automated systems interact with human behavior, and the record of that interaction is the only way to separate speculation from sequence of events.

Make The Evidence Portable

The throughline from a D.C. mileage threshold, to a California holiday traffic jam, to a federal crash readout is simple. Everyone wants to know what happened, and who is responsible for preventing it next time. Right now the answers depend on who holds the data and what they are required to share.

Data transparency is not a slogan. It is a set of practical requirements. Disengagements need consistent definitions across jurisdictions. Incident reports need standardized fields that cover software build identifiers, sensor status, and handoff timings. Release notes need versioned change logs that document safety relevant changes. Access rules need to specify what gets reported publicly and what regulators can audit privately. None of this decides if robotaxis are good or bad. It decides whether the next debate is about metrics or marketing.

Consider how the D.C. fight is framed by miles and fees. Those are simple to audit, but they do not tell the Council how a fleet behaved during a power outage, a protest, or a football game. San Francisco’s letter to state regulators is about those moments. The only scalable way to answer both is to require companies to publish comparable operational data and to let regulators verify the raw feed when needed. Companies already generate these logs for engineering. The question is whether policy converts them from competitive folklore into shared facts.

The industry should want this as much as skeptics do. A transparent incident record turns single company controversy into common learning, which can shorten the time between a failure mode and a fix. It also lowers the temperature on every permit fight. Instead of arguing over who gets a head start, participants can argue over how fast observed risks are shrinking from one release to the next.

Software can be global. Liability remains local. The only thing that scales across that divide is evidence. Cities will keep writing permit rules. States will keep preempting or standardizing. Federal investigators will keep reading black boxes. The pivotal choice is whether every disengagement and incident becomes shared proof, or another round of PR.