Today, I submitted a comment on the Autonomous Vehicle Deployment Authorization Amendment Act of 2026 (Bill 26-684).
Chairperson Allen and members of the committee, thank you for the opportunity to testify. My name is Lucas Pombo, and I am a research fellow at the Foundation for American Innovation, a think tank focused on promoting innovation, strengthening governance, and advancing national security.
I strongly support the rollout of autonomous vehicles in the District, which the bill provides for, but I am concerned about a variety of problems with the proposed legislation which may undermine its purpose. This written testimony will cover my four concerns:
- Need for Permit Shot Clocks: this bill has no safeguards against the District Department of Transportation (DDOT) dragging its feet for years, as it did previously, and the council should add a time limit before AVs are presumptively legalized.
- Anti-Deadheading Rules: attempting to minimize deadheading (travel without a paying passenger) may require increased parking and raise wait times. The council should reconsider this requirement or clarify when deadheading is permissible.
- Ward Wait Time Equalization: this bill also requires all residents to be exposed to increased vehicular activity for a service not everyone can use. The council should clarify how AV operators are expected to equalize wait times across wards, to avoid unintentionally making wait times longer for everyone or burdening residents with AVs they cannot afford to use.
- Workforce Retraining Program Spending: while public transit is a worthy source of funding, workforce retraining programs are not, especially for rideshare drivers. Consequently, the portion of the AV Deployment Fund revenue currently directed to workforce retraining programs should either be reallocated to WMATA, or returned to AV riders through lower fares.
Need for Permit Shot Clocks
The path to this bill has been long and full of delays. The 2020 Autonomous Vehicles Testing Program Amendment Act (AVTPAA) directed DDOT to both submit a report on how to safely deploy autonomous vehicles (AVs) within two years (that is, by October 2022), and to create a testing framework and permitting process for autonomous vehicle testing in the District. Neither directive was meaningfully followed.
The report that this council requested was delivered in April 2026, and contained an overview of how AV legislation was implemented in other municipalities. However, it did not contain any specific recommendations on how to implement AVs in DC. Meanwhile, the permitting process that was required by law was proposed in January 2022, but no final regulations actually emerged from that process., This bill is required in large part because DDOT did not create a permitting process the previous time it was so directed.
The proposed bill already does contain some consequences for bureaucratic delays. In section 3a-1(c)(5), the bill stipulates that if DDOT fails to promptly issue a renewal, the previous operating permit is extended until DDOT responds. The appeals process has a similar but more binding consequence: if the Office of Administrative Hearings fails to hold a hearing within 90 days of a DDOT determination, then the appeal succeeds, and the enforcement action is automatically reversed.
Those sections of the bill mirror permitting efficiency best practices at the federal and state levels. At the state level, the Permit Streamlining Act in California provides a strong example. It requires any body issuing a development permit to deem a permit application complete or name specific deficiencies within 30 days. Further, a permit is presumed approved after 60 or 180 days, depending on the permit's complexity if not explicitly approved or denied.
California's law was passed in the context of local government processes dragging on unreasonably, similar to DDOT's response to the 2020 AVTPAA. In 1975, Dow Chemical proposed a plant in Collinsville, about halfway from San Francisco to Sacramento. Twenty-three months into its permitting process, Dow had received only 4 of the 65 permit approvals it needed to build its plant, and partly blamed local governments for causing indefinite uncertainty. The California Permit Streamlining Act, with its capped processing times, was the response to that saga. It is my personal opinion that DDOT's reaction to the 2020 AVTPAA, and the uncertainty that response has caused, is at least as outrageous as California's Dow Chemical saga, and warrants similar action from this council.
The Federal Communication Commission (FCC)’s regulations draw from a similar record of local bureaucracies dragging their feet on critical permits. Like AVs, 5G is a leading-edge technology that, to be accommodated, required regulatory changes. In the case of 5G, connections got faster but more brittle and so required dramatically more antenna locations than 4G or LTE technologies. In turn, that meant permits became dramatically more important for the rollout of 5G than any previous cellular technology. At the same time, local permit timelines were getting worse. Lightower, a networking infrastructure provider in the Northeast and Midwest, reported to the FCC that average processing times for networking structures (LTE and 5G antennas) had risen from 300 days in 2016 to 570 days in 2017. AT&T reported that in the worst case, an unidentified California locality had taken more than 800 days to process a permit application.
Given that DDOT has failed to create a permitting process for nearly four years—longer than any delay in the California or FCC incidents—I recommend that this council treat the creation of an AV permitting process the way it treats permit renewals and appeals. If DDOT does not create a permitting process for testing within 60 days and commercial operation within 180 days, as the bill requires,, then AV testing and commercial operation after those thresholds should be presumed legal, subject to all other statutory and regulatory requirements for vehicle operators in the District. Failing to create meaningful consequences for DDOT's disregard for statutory requirements has already delayed the rollout of AVs in the District for far too long.
Anti-Deadheading Requirements
The legislation also requires that AV operators submit a plan which will "minimize the operation of an autonomous vehicle without a passenger" and which will "fairly rebalance fleet vehicles to reduce disparate availability or wait times." As the legislation suggests, AV networks are centrally controlled and can be coordinated. That means AV networks do not suffer collective action or distribution problems inherent to taxi or rideshare networks. However, those advantages can only be accessed through deadheading. The legislation's requirement that AV operators minimize deadheading and redistribute vehicles at the same time creates a contradiction, and subjects AV operators to heightened legal uncertainty.
As an illustrative example of why even access to AVs requires rebalancing through deadheading, consider the period immediately after rush hour, around 9:15 a.m. At that time, a large number of rush-hour trips will (hopefully) have been recently completed, but travel demand back to outer wards will be very low. Consequently, in the time before an AV operator returns all those vehicles back to base, vehicle availability in other wards will be much lower than in downtown.
An AV operator could employ several strategies to mitigate disparate wait times, but none of them seem particularly desirable. It could, for example, artificially lengthen wait times downtown to match the longest-wait-time ward. It could also deadhead vehicles in the opposite direction, filling residential streets with vehicles to match the availability of vehicles where they naturally accumulate downtown. Another alternative would be to simply reject most trips to downtown during morning rush hour, to avoid the disparate impacts of vehicles pooling there afterwards.
Some hybrid combination of these strategies exists which mitigates the intensity of any specific one. No similar equal-distribution requirements exist in other cities to point to as examples, but logic dictates that if DC law requires AV operators to combat the natural accumulation of demand, then they have to intervene somehow.
The problem with the proposed bill is that it directs DDOT to both require and prohibit those interventions. While the proposed bill seeks to reduce the regulatory uncertainty that AV operators face, these contradictory directives may instead worsen it.
At a deeper level, though, it seems as if the legislation is so indecisive because the legislative intent was not decisive either. The bill seems to favor creating access to AVs, and short wait times for AV rides, because those things benefit District residents. It also seems to perceive an increase in traveled vehicle miles as a cost to DC residents, in both congestion and potential safety terms.
Either position is reasonable, but they create a direct tradeoff that prohibits having both positions at the same time. Either access to AVs is a bigger benefit than the congestion cost of hypothetical deadhead miles, or the cost in congestion and safety is a bigger cost than the access benefits. Members of this council should exercise their representative authority to pick a side, and reduce the subsequent regulatory uncertainty. My perspective as a witness and resident is that the access benefits are larger, especially because a deadheading AV is just as useful to passengers as a non-autonomous taxi deadheading between trips. But more important than which side the council takes is that it takes a position at all.
Vehicle Balancing Requirements
While the vehicle balancing requirements create regulatory uncertainty, they also fail to consider that different wards may genuinely want different levels of AV prevalence. If AVs have benefits in terms of improved transportation, and costs in terms of additional miles traveled on neighborhood streets, then the optimal amount of AV prevalence depends on how useful the AVs are to residents.
In the more distant future, when AV operators improve their operations and costs are lower, AVs may be useful to all people. But the proposed bill fails to consider that the AVs likely to be deployed by the bill are not universally useful to all people. AVs as they currently exist are expensive—data from San Francisco shows that through November 2025, AV rides are more expensive than traditional, non-automated rideshare trips. That means that AV demand will likely be distributed similarly to traditional rideshare demand, which is not an even distribution.
Recent data from Chicago suggests that demand for rideshare trips is overwhelmingly clustered in the top quintile of incomes, and consequently confined to more affluent areas of the city. Demand in Washington DC is likely shaped similarly. That demand distribution conforms with how one might expect it to be shaped intuitively: rideshare trips are expensive, and expensive goods are disproportionately consumed by richer consumers. That's not to say low-income demand for rideshare or AVs is zero, but merely that it would be financially unsustainable for low-income residents to use rideshare at the same rates as their high-income counterparts.
In light of that fact, equal vehicle availability across all wards is likely to be exceptionally difficult to achieve in practice. For vehicles to be available, they have to be on the roads. In high-income neighborhoods, having many AVs on the roads makes sense, since the residents would be using them. But how would an AV operator replicate that availability in lower-income neighborhoods? Fill their roads with deadheading AVs they can't use? Barring that, waiting in parking spaces across low-income neighborhoods? Make the wait times longer everywhere? For vehicles to be available with a short wait time, they have to be nearby. The council should articulate an acceptable means of positioning vehicles near users in all neighborhoods. Especially when combined with the requirement to have equal wait times between wards, failure to do so risks reducing AV access for everyone.
Workforce Retraining Program Spending
My fourth concern is that the programs receiving half of the proposed VMT tax are likely to be wasteful and ineffective. The tax revenues in the proposed bill are split, with half going to fund WMATA and half funding workforce retraining programs after paying for AV deployment costs. Workforce retraining programs have a dubious track record. WMATA, which receives the other half of the fund’s revenue, moves hundreds of thousands of people per day and is far more worthy of District funding.
There is extensive research on how workforce retraining programs impact unemployment. The evidence is decidedly mixed. An analysis of Department of Labor (DOL) programs which separated workforce retraining effects from other DOL programs found that the retraining programs in particular had a negative but statistically insignificant impact on earnings. Meanwhile, a meta-analysis of 40 workforce retraining programs abroad found that workforce retraining programs raised employment by only 2.6 percentage points on average.
In addition to the generally spotty record of workforce retraining programs, there are reasons to believe that they might be a particularly poor fit for rideshare drivers specifically, for two reasons. First, workforce retraining programs typically seek to place participants into full-time jobs, which are not an appropriate work substitute for most rideshare drivers. Second, the bill specifies that the workforce retraining programs will be targeted at "ride-share drivers who have been or are at-risk of job displacement." Because of how these platforms find and lose drivers, it will be difficult to draw a line around eligibility that does not include all District residents.
The ephemeral, supplemental nature of rideshare employment makes workforce retraining programs a poor substitute for rideshare driving. Lyft's 2024 analysis of driver characteristics found that 88 percent of drivers drove fewer than 20 hours per week on the platform, 58 percent of drivers in 2024 took a break longer than two weeks, and fully 91 percent of Lyft drivers were either employed or studying in addition to driving for the app. Meanwhile, DOL evaluates retraining programs by whether their participants find full-time employment. Targeting workforce retraining programs at workers who are already employed and not looking for full-time jobs is a waste of tax dollars, especially when the alternative is funding WMATA.
Targeting retraining programs at displaced drivers is either extremely challenging or logistically impossible, due to the way this displacement occurs. If AVs replace some rideshare demand, as a near-perfect substitute, then fewer trips are available for drivers, which will increase wait time between trips. In turn, the marginal change in revenue will push away some drivers who were previously unsure whether rideshare driving was worth their time. New potential drivers looking for supplemental income will be incrementally more likely to choose other options. These decisions are not well-documented in a way that would allow a program to target the people who make them.
Seeking to help displaced drivers is commendable, but rideshare driving is not analogous to more typical full-time employment. Rideshare drivers use these platforms as a short-term bridge between jobs, as a way to supplement a full-time income, or to support themselves while studying. Appropriate replacements for rideshare jobs must be able to do the same, and workforce retraining programs do not accomplish that.
However, the Chairman explains in his letter to the Secretary to the Council that the AV tax is not entirely about revenue. The reason the $0.15 per mile VMT tax exists is that it “encourages AV operators to minimize passenger-less operations and circuitous routing.” Consequently, the uselessness and inapplicability of workforce retraining programs do not justify lowering the tax given its stated rationale. However, the same letter indicates that this bill has a goal of improving public transit. Consequently, I encourage the council to redirect the whole of the AV Deployment Fund surplus to WMATA, rather than the half that it currently does.
Conclusion
I strongly support the rollout of autonomous vehicles in the District, and I hope the council adopts my recommendations. Thank you for the opportunity to testify, and I am looking forward to seeing AVs on DC streets. I would be happy to follow up on any questions or concerns by email.
