We welcome the opportunity to submit comments on the most recent version of the proposed US Department of Transportation (DOT) automated vehicles policy entitled Preparing for the Future of Transportation: Automated Vehicles 3.0 (henceforth, “AV Guidelines 3.0”). In our view, this new version is adaptive, proactive, and it encourages innovation of autonomous vehicles (AVs) while recognizing with humility that as this technology develops quickly, regulation must be flexible and adaptive to keep pace. We encourage DOT to continue this current approach to the regulation of AVs, and we also submit a few questions for future consideration about the enforcement and remedies related to such guidance.
We have noted and praised the shift in the regulatory approach of DOT from a heavy-handed, precautionary approach to a more flexible, innovation-friendly approach. The latter is characterized by the use of “soft law” (that is, best practices and norms that emerge in the marketplace from the strategic decisions of developers and the free interaction of market actors). In this public interest comment, we further discuss why we see this soft-law approach as most likely to encourage innovation and promote safety.
After detailing some of the aspects of AV Guidelines 3.0 that we support, we posit a few questions about the enforceability of the document. We ask whether DOT plans on enforcing any of the recommendations outlined in this document if a firm decides not to comply with them and, if so, how it would go about enforcing those recommendations. The lack of clear federal preemption in the area of AV governance adds additional complications to the question of enforcement. Finally, we ask what administrative remedies and procedures exist if a market participant wants to challenge any of the recommendations found in AV Guidelines 3.0.
We wish here to encourage DOT to continue its soft-law, innovator-friendly approach to AV governance, and we recommend that the department clarify what its enforcement process would entail under this approach.
Hitting the Right Notes: Evolution of the Department of Transportation’s Approach to Governing Autonomous Vehicles
In 2016, DOT first addressed the issue of AVs through the first version of Preparing for the Future of Transportation. This initial guidance took a much more restrictive and precautionary approach than would later versions. As Adam Thierer and Caleb Watney pointed out in their comments at that time, this initial guidance indicated that premarket approval may be necessary for AVs, which could impose a high opportunity cost on developers, stifling innovation and increasing the number of type I errors. These risks would have been exacerbated by the fact that AV technology changes and develops at a rapid pace. This initial guidance entertained the concept of looking to the Federal Aviation Administration certification process, which typically “last[s] three to five years,” as a model for how AV premarket approval might work. This process could result in substantial financial cost and delay in the implementation of this technology without necessarily resulting in an increase in safety or innovation. Yet time is of the essence, because this technology not only will improve vehicular transportation, but once adopted in the mass market, it is likely to save lives by reducing car accidents. Fortunately, the more recent approach has shown that DOT has chosen to pivot to a more adaptable and less overly precautionary approach to AVs.
Last year, in version 2.0 of the AV Guidelines, DOT signaled that it was taking a much less restrictive approach. The updated proposal abandoned the premarket approval approach that DOT had suggested innovation in the first version of its AV Guidelines and signaled a major step in the direction of a permissionless approach to regulation. A permissionless innovation approach would allow individuals to innovate within the field of AVs with limited need to seek prior authorization. Overall, the updated proposal places major emphasis on industry-developed best practices and on the importance of self-regulation as a preferable alternative to top-down, binding regulations developed by a regulatory agency. DOT discouraged state governments from turning the document directly into legislative or regulatory directives and continually stressed that it sought to encourage actions rather than require them. A question to which we will return later is whether the department had any formal mechanism it could use to enforce these recommendations.
This latest version shows that DOT has continued to take this soft-law approach. As Ryan Hagemann, Jennifer Huddleston Skees, and Adam Thierer discuss in their forthcoming paper “Soft Law for Hard Problems: The Governance of Emerging Technologies in an Uncertain Future,” various alternative governance mechanisms—including documents like AV Guidelines 3.0—are attempting to fill the void left by traditional regulation. This regulation gap exists because technological development advances faster than regulators can make rules and certainly faster than Congress can enact law—and the gap only widens when national politics is afflicted by demosclerosis (that is, when political polarization has slowed down the lawmaking process nearly to a halt). The way that DOT has handled AVs stands among other emerging approaches to regulation because through various consultation mechanisms, including working groups, DOT has developed a flexible regulatory framework that remains neutral with respect to the potential paths for technological development and provides innovators and investors a greater degree of certainty regarding regulation in this area.
AV Guidelines 3.0 represents the continuation of an innovation-friendly regulatory approach that encourages entrepreneurship in the area while maintaining flexibility to adapt along with the technology. It acknowledges the need for regulatory consistency between federal, state, and local actors so that there is not a confusing, patchwork-style framework regulating AVs. This will signal legal consistency that is a necessary prerequisite for substantive investment in emerging technologies. In addition, AV Guidelines 3.0 makes it clear that DOT will remain neutral toward various avenues of technological development with the view that increased competition and active innovation will promote safety, mobility, and positive economic outcomes. It emphasizes the fact that AV technology has the potential to promote safety and save lives. Finally, it focuses on the need to partner with key actors and stakeholders within the industry to capitalize on their dispersed and specialized knowledge.
Given the rapid pace of technological development, this is a realistic approach because it will allow innovation to flourish while remaining relatively safe. Yet soft law is not without its disadvantages. The department must recognize that the policies will need to continuously evolve, and eventually it will need to address difficult questions that are currently unanswered.
Enforceability of DOT's AV Guidelines 3.0
One question that remains is how and when the department will try to enforce these guidelines, and particularly what such enforcement would look like when regulation shifts from soft to hard law. These guidelines do not include the traditional introduction of binding restrictions such as must or shall but instead use the much less legally significant terms such as should consider or encourage. This lack of formality distinguishes soft law actions, such as version 3.0 of Preparing for the Future of Transportation, from the more formal strictures of traditional agency regulations, but also brings distinct questions about the need for compliance and the enforceability of such guidelines. These potential enforcement problems are even more distinct for documents like AV Guidelines 3.0 than they are for legally binding agency guidance. While generally this flexibility and encouragement over formal requirements is a positive for a rapidly developing technology such as AVs, it does raise questions of how the department or innovators might fare were they to deviate from the recommendations.
In January 2018, the Department of Justice (DOJ) issued a memo to agencies from the assistant attorney general stating that guidance documents should not be used to create binding actions that would coerce regulated parties to refrain from or undertake certain actions and that DOJ would not provide affirmative civil enforcement for any guidance documents that purported to do so. As our colleague Brian Knight explains, this does not eliminate the department’s ability to use guidance documents, including those that create binding requirements, nor does it prevent DOJ from enforcing the underlying law that the guidance addresses. This refusal by DOJ, however, is part of the Trump administration’s broader push for agency reforms while still allowing an individual agency to continue make its own enforcement or nonenforcement directions regarding such guidance. The current AV Guidelines 3.0 does not appear to create the sort of binding requirements that would fall under this “guidance on guidance” policy, nor does it itself address the question of enforcement, should there be noncompliance.
Still, the department should consider enforcement issues that may arise before a possible first instance of noncompliance. The question appears to be, Are these AV recommendations deviations from existing requirements, additional requirements, updated interpretations of requirements, or something completely independent? In the current instances when the National Highway Traffic Safety Administration (NHTSA) or the National Transportation Safety Board (NTSB) have taken actions regarding AV testing, it has been either to investigate an accident or to investigate claims of noncompliance with existing Federal Motor Vehicle Safety Standards (FMVSS). For instance, in October the NHTSA ordered a tester to cease testing of autonomous shuttles being used to transport children to school, claiming that such testing violated its waiver request and the request should have specified that it would be used as a school bus. Yet AV Guidelines 3.0 suggests that, other than obtaining waivers from standards for a specific vehicle, compliance with these standards is voluntary.
AV Guidelines 3.0 has been developed through a series of working groups, consultations, and other commenting periods, with its recommendations mirroring existing industry best practices. This is likely partly why, while not going through all formalities of the Administrative Procedures Act, DOT has not yet confronted noncompliance. Similarly, the guidance from the department recommends embracing industry-developed best practices over a more external top-down structure. At some point, given the rapid development of the industry and the continuously changing nature of the technology, it is more probable than not that a vehicle or technology will consider but not undertake or comply with all of the encouragements and inducements of AV Guidelines 3.0, but not have violated any of the more formal strictures of FMVSS or other existing hard law. The soft law approach may be adaptable enough to avoid many such confrontations, but DOT should begin considering what will happen if it does need to enforce the document and what it may be able to do if faced with a legal challenge in such circumstances. We don’t propose in these comments to solve such a question for the department, but merely to draw its attention to the issue so that it can continue the regulatory environment that encourages these innovations that will likely have significant societal benefits by preventing deaths from automotive accidents and reducing congestion and stress associated with transportation.
Because AV policy is currently shared between the states and the federal government with no clear preemption, an additional complication is the potential for state laws regarding AVs to conflict with or even contradict the existing DOT soft law. Starting with version 2.0, the department has made clear that it does not intend for this document to serve as model legislation merely to be codified by state governments. However, given the absence of federal preemption on standards, it is possible that states could create more or less restrictive standards that would then place innovators in the difficult position of choosing which set of standards to comply with. In the event a state passes less restrictive standards, this again would raise questions of enforcement. Clarifying if these standards are to be considered a minimum for states could prevent this potential quagmire for innovators and state policymakers alike. Just as with the regulation of human-operated automobiles, there will always be an element of shared governance between the federal government and the states regarding regulations over matters such as insurance and licensing. DOT smartly specifies certain areas that it views as areas for states to participate in such regulatory frameworks in the most recent version of this guidance. Still, it is possible for a large and populous state to impose standards more stringent than the federal standards and consequently undercut the incentives to innovate provided by AV Guidelines 3.0.
Administrative Remedies Under Existing Soft Law
In addition to the potential enforcement issues, clarifying what remedies are available to those wishing to change, challenge, or better understand the existing guidance on AVs beyond the notice-and-comment periods accompanying new versions of this guidance or the occasional requests for information from the department or its agencies would clarify both the department’s view on the finality of its existing guidance and the redress available to those who may have issues with certain elements under both the administrative and legal systems.
At first glance, no version of the existing AV guidance appears to be able to be interpreted as a new binding final regulation rather than a clarification or interpretation of existing department policies. Notably, the numbering system indicates that the department plans to continuously and regularly provide updates to this instrument—similar to how developers update software. As a result, much like with the enforceability of the document, it is unclear when it will be considered final and what administrative remedies are available, if any, in addition to the notice-and-comment period for each version or for specific rulemakings or guidance that may arise from related matters. Particularly if the standards are to be enforceable (as opposed to mere suggestions), clarifying what remedies are available, should a party wish to deviate, would likely assist regulators and innovators alike in preventing perhaps more adversarial interactions.
As a result, we suggest that the department consider clarifying the following points either in direct response to public comments or in future iterations of AV guidance:
- Is there an administrative process accessible to those who wish to deviate from the current recommendations contained in this guidance?
- How frequently will new iterations occur, and when and how will a final iteration of such guidance be known?
We applaud DOT for its humility in approaching this rapidly evolving technology and hope that it continues to examine transportation innovation in a similar way. The questions raised in these comments seek to complement AV Guidelines 3.0 with a clear scope and process of enforcement as this technology transitions from fiction to reality. Given the life-saving potential of AVs, DOT’s approach of encouraging such innovation through a flexible framework will hopefully prove effective in this field.
 Adam Thierer & Caleb Watney, Mercatus Ctr. at George Mason Univ., Comment on the Federal Automated Vehicles Policy (2018), available at https://www.mercatus.org/system/files/mercatus-thierer-automated-vehicl…. (“Type I” errors are false positives. For example, a regulator may overestimate the risk associated with a given technology and thus impose unnecessarily stringent restrictions upon that technology that limit its innovative potential.)
 See Huddleston Skees & Thierer, supra note 3.
 Ryan Hagemann, Jennifer Huddleston Skees & Adam Thierer, Soft Law for Hard Problems: The Governance of Emerging Technology in an Uncertain Future, Colo. Tech. L.J. (forthcoming).
 See Hagemann, Huddleston Skees & Thierer, supra note 6.
 See Patrick McLaughlin & Jake Jares, Five-Letter Words and Legal Language, The Hill (Feb. 5, 2016), https://thehill.com/blogs
 Memorandum from the U.S. Dep’t. of Justice, Office of the Assoc. Attorney Gen., to Heads of Civil Litigating Components U.S. Attorneys (Jan. 25, 2018).
 Brian Knight, Regulation by Guidance and Due Process, The Bridge (Feb. 8, 2018), https://www.mercatus.org/bridge
 See, e.g., Accident Investigation, National Transportation Safety Board, Car with Automated Vehicle Controls Crashes into Pedestrian (Mar. 18, 2018); Press Release, National Highway Traffic Safety Administration, NHTSA Directs Driverless Shuttle to Stop Transporting School Children in Florida (Oct. 19, 2018).
 National Highway Traffic Safety Administration, supra note 13.
 See Skees, Gray Areas, supra note 15.
 See Skees, When States Get It Wrong, supra note 15.
 See Skees, Gray Areas, supra note 15.
 See, e.g., Devon Energy Corp. v. Kempthorne, 551 F.3d 1030 (D.C. Cir. 2008).