'Soft Law' Is Eating the World

Thursday, October 11, 2018
Jennifer Huddleston
Adam Thierer

In his widely-cited 2011 Wall Street Journal essay, “Software is eating the world,” venture capitalist Marc Andreessen famously noted that software and digital communications platforms were increasingly disrupting nearly all economic sectors and precincts of modern life. He was right. Now “soft law” is eating the world of technological governance and regulation more generally. The evolution of automotive technology, particularly the increases in automation, is proof positive of both points. We are witnessing a radical metamorphosis in the way that cars and the laws that govern them operate—and in both cases for the better.  

Andreessen’s point about software devouring everything was based on the realization that the underlying drivers of the digital revolution—massive increases in processing power, exploding storage capacity, steady miniaturization of computing, ubiquitous communications and networking capabilities, and the digitization of all data—were rapidly spreading to other sectors of the economy. No longer was information technology or software its own isolated silo, but industries from farming to communications were being revolutionized by it. This affected not only new, emerging areas but also long-standing and mature sectors like the automotive industry, which is now being further revolutionized by intelligent driving and autonomous vehicle (AV) technologies.

Driverless Tech: From Sci-Fi to Reality

In the short span of 15 years, AV technology has gone from science fiction to government-sponsored research projects and now to regular headline news. Major technology companies, such as Waymo and Apple, and legacy automobile manufacturers, such as General Motors and Toyota, have all increasingly invested substantial sums in this potentially socially-transformative technology—whether through R&D, roadway testing, information-sharing associations, or formal partnerships. Elon Musk recently hinted that Tesla vehicles could soon receive a new software upgrade that enables a “fully driverless option” via their Autopilot system. Uber, Waymo, and others have already started testing their driverless cars on public roads.

These companies have recognized how imperative it is to join the AV race, despite the law’s failure to keep pace with the technological development. That’s led some entrepreneurs, such as Musk and Comma.ai’s George Holz, to push the regulatory boundaries as regulators and policymakers grapple with the difficulty of regulating 21st century technologies with a set of 20th century institutions. This problem is what scholars refer to as the “pacing problem”—referring to the fact that technological change is unfolding faster than public policy’s ability to keep up with it. This gap has grown so wide, in fact, that it is even posing a vexing challenge for incumbent automakers who may lack the experience, resources, and staff necessary to address regulatory compliance.

For example, Audi recently announced that its new A8 will be the first commercially-available vehicle capable of operating at Level 3 autonomy . Unfortunately for American consumers, that functionality won’t be coming to these shores any time soon. Despite the lack of rules that affirmatively prohibit the operation of AVs on American roadways, lingering uncertainty about the domestic regulatory environment is enough to dissuade the German automaker from exporting its cutting-edge technology in the United States. While a few individual states have begun to embrace the autonomous future, the patchwork of regulatory uncertainty on a national level is enough to discourage some companies from believing the technology will be welcome here.

How Soft Law Solves Hard Problems

What are lawmakers and regulators doing to bring some order to this uncertainty and chaos? There have been attempts by Congress to enact new laws like the AV START and SELF DRIVE Acts, which would create a national framework for driverless car governance. But these efforts have largely stalled, leaving a governance void. “Soft law” is filling that void and providing the “rules for the road.”

What is “soft law”? As we explain in a forthcoming law review article, soft law represents a set of informal norms, multistakeholder arrangements, and non-binding guidance standards that provide an adaptable alternative to more traditional regulations or legislation. As Gary Marchant and Braden Allenby note, soft law can be described as “instruments or arrangements that create substantive expectations that are not directly enforceable, unlike ‘hard law’ requirements such as treaties and statutes.” These “instruments or arrangements” have played a significant role in promoting the rapid emergence of numerous life-changing technologies over the past quarter-century, including the rise of the Internet. Now, soft law is once more plugging the “pacing problem” gap—this time, for the governance of AVs.

Last week, the US Department of Transportation (DOT) released its third iteration of guidelines for autonomous vehicles. “Preparing for the Future of Transportation: Automated Vehicles 3.0” (AV 3.0) builds upon the DOT’s earlier “2.0” guidance document. It reiterates many of the points outlined in the previous version, maintaining a staunch commitment to the voluntary, non-binding nature of the provisions while expanding consideration to include commercial trucking and other surface vehicles. Most notably, AV 3.0 avows that the DOT will no longer assume the driver of a vehicle is human, and outlines a strategy for addressing the numerous regulations (especially the Federal Motor Vehicle Safety Standards) that need to be updated in order to alleviate the uncertainty currently discouraging companies like Audi from deploying their AVs on American roads. DOT has come a long way from its first iteration of the AV guidance released under the Obama administration, which provided a far more prescriptive and “hard law”-oriented approach to governing this emerging technology.

Why Soft Law Dominates Emerging Tech Policy

The fact that the agency is “versioning” its guidance as if it is a software release (i.e., “2.0,” “3.0”) provides even further evidence that Andreessen was on to something when he predicted that software was eating the world, and that even included the world of laws and regulations. Some might find it shocking that an undated slideshow presentation is essentially all that now guides federal AV policy, but that is an indication of just how pervasive soft law approaches are becoming in the modern regulatory environment. It also underscores the extent to which many policymakers and agency officials—much to their credit—have recognized the institutional limits of their ability to effectively govern new technologies. In essence, technologies like AVs are forcing policymakers and agency officials to increasingly rely on more adaptive, informal, and flexible rules to govern this new era of punctuated change.

In our new law review article and subsequent essays, we document how soft law is being tapped in countless other sectors, such as artificial intelligence, the Internet of Things, advanced medical technologies, and many more. These approaches are not merely limited to the federal level, but are also being utilized by state regulators in fields such as autonomous vehicles and FinTech and even by cities collaboratively working on regulatory schemes for disruptive technology such as electric scooters. The “soft law-ization” of technological governance is probably inevitable for two primary reasons. The first explanation goes back to Andreessen’s point about software eating the world, and the increasing gap between the rate of innovation and policymakers’ ability to achieve legal and regulatory parity without strangling innovation in the cradle (i.e., the “pacing problem”).  

The second reason soft law is now ascendant is that traditional legislative and regulatory hard law processes are somewhat broken. The combination of regulatory accumulation, over-bureaucratization, and the influence of special interests has led to increasingly dysfunctional governing mechanisms. This is a particularly acute problem in the field of technological governance, compounded by a lack of technical expertise in Congress. Similarly, while the courts may still serve as a check on administrative actions in some cases, judicial deference makes it increasingly less likely that they will overrule most agency actions, barring some particularly egregious circumstances.  

The Worst of All Systems, except All the Others before It

For better or worse, soft law is becoming the dominant modus operandi for modern technological governance. Conservative critics may take issue with its informal, open-ended nature, thinking it will be ripe for abuse. Liberal critics might be angry that stricter rules are not being put on the books.

Both sides have a point, but both sides should be cognizant of the realities on the ground. With the “pacing problem” and dysfunctional “demosclerosis” of government becoming the new norm, something needs to fill the governance gap. That something will increasingly be soft law. On the upside, soft law offers a chance to finally craft policy in a more flexible, adaptive fashion. It can adjust to fast-changing markets in a more reasonable and efficient way. While it won't be perfect, it may provide the best solutions we can hope for.

For driverless cars, the proverbial “rules of the road” may not end up being formal rules so much as an amalgam of informal guidelines, self-regulatory standards, ongoing oversight by regulators, and targeted enforcement of some older laws that still make sense (like recall notices for defective products). That mix of policies will be married up with ex post enforcement of torts and other consumer protection laws in the courts.

This is the future of technological governance. Few will love it, but all will need to adjust and figure out how to make it work so that innovation keeps flowing while important social goals are still addressed. In an age where regulators confront a variety of new and unique governance challenges, balancing these aims is more crucial than ever. What comes next is anyone’s guess, but one thing is clear: software has already eaten the world; now soft law is eating the regulatory state.

Photo credit: Jae C. Hong/AP/Shutterstock

Driverless Car Edition

Removing Roadblocks to Intelligent Vehicles and Driverless Cars

June 1, 2015

This paper addresses some of the early policy concerns about "connected cars" and driverless vehicles and promotes "bottom-up" solutions to ensure that innovation continues to flourish in this space. The authors argue that the generally unabated advancement of intelligent-vehicle technology will produce significant economic and social benefits. Various technical and policy barriers to more widespread adoption remain, however, and misguided regulation could delay or curtail the adoption of this important technology. This paper outlines ways of overcoming those hurdles. The authors also argue that policymakers should keep in mind that individuals have gradually adapted to similar disruptions in the past and, therefore, patience and humility are needed when considering policy for intelligent-vehicle systems.

Operation and Certification of Small Unmanned Aircraft Systems

April 24, 2015

The Federal Aviation Administration (FAA) is proposing the adoption of specific rules to integrate unmanned aircraft systems (UASs) into the National Airspace System. The proposed rules would, among other things, prohibit small unmanned aircraft from conducting external load operations (i.e., deliveries); prohibit the operation of small unmanned aircraft over people not involved in the operation; prohibit the use of see-and-avoid responsibilities through technological means, such as onboard cameras; and prohibit operation outside of the hours of official sunrise and sunset.

In a public interest comment published by the Mercatus Center at George Mason University, technology policy scholars Eli Dourado, Ryan Hagemann, and Adam Thierer explain why the FAA’s proposed rules fail to consider all of the benefits of UASs and are an exercise in overly precautionary thinking. Rather than worry about hypothetical harms with relatively low risk, government policy should encourage what is known as “permissionless innovation.” While other countries around the world are already benefiting from unmanned aircraft technology, the FAA’s proposed rules will not allow such innovation to flourish in this country, to the detriment of consumers and the American economy.

Key Findings 

The FAA does not adequately consider all of the benefits of small UAS technology in accordance with Executive Order 12866:

  • The FAA states that the flight characteristics of aircraft carrying external loads pose additional risks. However, the FAA supplies no discussion of the benefits of allowing small unmanned aircraft to conduct external-load operations.
  • The FAA has not considered the benefits of allowing UASs to operate beyond line-of-sight, only the risks. Technologies are being developed and deployed that address the FAA’s concerns.
  • The FAA does not consider the benefits of allowing fully or partially autonomous UASs to operate on the basis of a single operator for multiple aircraft. Allowing such a practice would drastically lower the cost of operating a large fleet of unmanned aircraft.
  • The FAA does not consider the benefits of allowing UAS operations over persons not involved in the operation. Many creative and valuable uses of UASs will likely develop in urban areas, where greater density enables higher benefits from drone-based transportation of goods.

Additionally, the FAA fails to provide clear guidance on UAS activities that have an academic, noncommercial, or humanitarian nature.


The FAA’s proposal not to require UAS operators to obtain a commercial pilot certificate and its determination that small UASs should not be subject to airworthiness certification are both sound policy choices. But there are further improvements that must be made to make the FAA’s proposal workable:

  • Conduct proper benefit-cost analysis. The FAA should consider the benefits of external load operations, operation beyond line-of-sight, multiple craft operation, operation over noninvolved persons, and nighttime operation. The benefits of these options exceed the costs that the FAA mentions.
  • Pursue legal remedies in lieu of prohibitions. The FAA should consider how longstanding legal remedies in tort law, such as strict liability, negligence, design defects law, failure to warn, and breach of warranty can address concerns that the FAA is seeking to alleviate through prohibitions.
  • Adopt rules more quickly. The FAA has a statutory obligation to permanently integrate civil UASs into the airspace by September 30, 2015, pursuant to § 332. The FAA is likely not moving fast enough to meet this obligation.
  • Consider the effect of rules on the economy. The FAA must carefully consider the potential effect of UAS restrictions on the US economy. If it does not, innovation and technological advancement in the commercial UAS space will find a home elsewhere.

Don't Slam the Brakes on Smart Cars With Too Much Regulation

Monday, December 1, 2014
Adam Thierer

The world is on course to change very rapidly in a short amount of time. From new business models changing how individuals interact with one another to the onset of deliveries-by-drone, technology is altering the way we live our daily lives — and some of the most significant changes have yet to arrive. Among all the contributors to the emerging digital age, no force has been as substantial as the idea of “permissionless innovation.”

As the Washington Post has highlighted, the essence of permissionless innovation is the idea that, in the absence of proof that a particular technology or idea will cause real, immediate and clear harm to society, “innovation should be allowed to flourish” unimpeded by preemptive regulations. The hurdles that innovators and entrepreneurs face in bringing new technologies to market should not include precautionary regulations that stifle technological development, especially as it relates to autonomous vehicles.

Continue reading

Intelligent Vehicles Could Save Lives

Wednesday, November 26, 2014
Adam Thierer

The American Automobile Association estimates that 43.6 million Americans will take extended trips during this Thanksgiving holiday season. Of those, approximately 90 percent – or 39 million people – plan on traveling by car. And tragically, according to the National Safety Council, some 418 Americans may lose their lives on the roads this Thanksgiving, in addition to over 44,000 injuries from car crashes.

This is why the debate over new “intelligent vehicle” and “driverless car” technologies is so important: Anything we can do to decrease driver error will help save many lives and reduce the cost of accidents. The technologies help automate many driving functions and could be in even more vehicles soon, assuming public policies don’t throw up unnecessary roadblocks to their development.

Last holiday season, the number of traffic-related fatalities fell below the quarter-century average, but with gas prices holding well below last year’s average of $3.27 during the same time, more Americans will travel by car. With more people driving, longer distances being traveled and the increased potential for “Black Wednesday” alcohol abuse, the prospect for disaster looms large during this holiday.

Many of these accidents, and fatalities, could be a thing of the past in coming years as intelligent vehicle technology begins to mature, alleviating many of these roadway perils. With regulators acting as gatekeepers to what’s allowed on the road, however, these technologies may be condemned to bureaucratic limbo; that is, unless regulatory agencies embrace the concept of “permissionless innovation.”

The essence of permissionless innovation is the idea that, in the absence of proof that a particular technology or idea will cause real, immediate and clear harm to society, innovation should be unimpeded by preemptive regulations. The hurdles that innovators and entrepreneurs face in bringing new technologies to market should not include precautionary regulations that stifle technological development, especially as it relates to autonomous vehicles. Such prescriptions are unlikely to yield the effects regulators anticipate. They are, however, more likely to spawn unintended consequences that could, among other things, hamper innovation, decrease economic growth and result in ongoing costs to society.

Among these costs are the more than 33,000 lives lost on American roads annually, due to driver error. How many lives could be saved if some of those drivers one day used autonomous vehicles?

In addition to fatalities, the National Highway Traffic Safety Administration points out that over 2.3 million people were injured in automobile crashes in 2012 alone – over 169,000 of which were children under the age of 14. Motor vehicle crashes like these are, in fact, the leading cause of death for children between the ages of 11 and 14. Every new development that pushes us closer to integrating this new technology into the mainstream means fewer fatalities.

With these safety issues in mind, policymakers ought to embrace the concept of permissionless innovation and avoid trying to predict every possible harmful scenario that might result from the introduction of this new technology. Flexible and creative solutions that could most ideally solve problems as they arise can only develop in a policy framework that promotes regulatory patience. There will be challenges, but progress can only occur when we accept a certain amount of risk and allow experimentation with new technologies. As was the case with the telephone, radio, the Internet and almost every other modern innovation, human adaptation and social acclamation will likely provide the best solutions to how we incorporate these technologies into daily life.

We should not delay this trial-and-error process based on the mistaken perception that we can anticipate every worst-case scenario that might result from integrating driverless vehicles into our transportation system. Inaction has its costs, too. Every day, approximately 92 people die from roadway accidents caused by human error.

Unleashing the powerful potential of permissionless innovation for autonomous vehicle technology could start to alleviate some of the known, and clearly detrimental, harms to drivers and passengers this holiday season. That would be something for which we can all be truly thankful.

Federal Aviation Administration Interpretation of the Special Rule for Model Aircraft

September 23, 2014

As part of the FAA Modernization and Reform Act of 2012 (FMRA),1 Congress ordered the Federal Aviation Administration (FAA) to integrate unmanned aircraft systems (UASs)—sometimes referred to as drones—into the National Airspace System by September 2015. As part of that effort, the FAA is currently accepting comments on its “Interpretation of the Special Rule for Model Aircraft” (Section 336 of the FMRA) and the FAA’s enforcement authority over model aircraft as affirmed by the statute.2

The Technology Policy Program of the Mercatus Center at George Mason University is dedicated to advancing knowledge of the impact of regulation on society. As part of its mission, the program conducts careful and independent analyses employing contemporary economic scholarship to assess rulemaking proposals from the perspective of the public interest. Therefore, this comment on the FAA’s “Interpretation of the Special Rule for Model Aircraft” does not represent the views of any particular affected party or special interest group but is designed to assist the administration as it carries out Congress’s mandate to safely integrate UASs into the National Airspace System.

In this brief comment, we discuss the need for the agency to conduct a thorough review of the benefits and costs associated with this rule. We argue this is essential because airspace is poised to become a major platform for innovation if the agency strikes the right balance between safety and innovation. To achieve that goal, we stress the need for flexibility and humility in interpreting older standards, such as “line of sight” restrictions, as well as increasingly archaic “noncommercial” vs. “commercial” distinctions or “hobbyists” vs. “professional” designations.

We also highlight the growing tension between the agency’s current regulatory approach and the First Amendment rights of the public to engage in peaceful, information-gathering activities using these technologies. Finally, we close by noting the important role that voluntary self-regulation and codes of conduct already play in governing proper use of these technologies. We also argue that other “bottom-up” remedies are available and should be used before the agency imposes additional restrictions on this dynamic, rapidly evolving space.


Before addressing the substance of the rule at issue here, we wish to remind the agency that it is required to conduct a formal benefit-cost analysis (BCA) of any “significant regulatory action” it undertakes.3 As defined by Executive Order 12866, a “significant regulatory action” includes rules or guidelines that “have an annual effect on the economy of $100 million or more or adversely affect in a material way the economy, a sector of the economy, productivity, competition, jobs, the environment, public health or safety, or state, local or tribal government or communities.”4 A “significant regulatory action” can also include regulatory actions that “raise novel legal or policy issues.”5

Although the exact economic ramifications of integrating commercial drones into the national airspace is uncertain, there have been numerous assessments of the potential impact this technology would have on the US economy. The Association for Unmanned Vehicle Systems International (AUVSI), for example, estimates that between 2015 and 2025, the integration of unmanned aerial systems into the national airspace “is expected to contribute $82.1 billion to the nation’s economy by agriculture, public safety, and other activities.” These benefits run the gamut from over 100,000 new high-paying ($40,000 or more per year) jobs created in aerospace manufacturing to almost 850,000 job years worked over that period.6

The FAA has not yet undertaken a comprehensive benefit-cost analysis of the rule in question. Such an analysis would help determine how new regulations in this space could affect the market for this evolving class of technologies as well as how these technologies might impact the broader economy or other important values. This is particularly true as the distinction between model aircraft and other types of unmanned aircraft systems is rapidly blurring.

The agency may believe that the matter under review is merely an interpretative exercise requiring no formal BCA review at this time. Nonetheless, as noted below, the agency’s “Interpretation” could indeed result in the sort of significant economic impacts and novel legal issues contemplated by Executive Order 12866. Consequently, the agency should not avoid formal BCA but instead take this opportunity to consider the ramifications of its actions in this matter.7

Continue reading

Removing Roadblocks to Intelligent Vehicles and Driverless Cars

September 17, 2014

This paper has been accepted for publication in a forthcoming edition of the Wake Forest Journal of Law & Policy.

Intelligent cars are possible today after decades of research and development in vehicle automation and computer processing. Advanced intelligent vehicle technology can bring significant economic and social benefits. Unfortunately, policymakers often impose “precautionary principle” policies on developing technology, stunting growth and discouraging innovation. Though it is well intentioned, trial-without-error policymaking results in fewer choices, lower-quality goods and services, and diminished economic growth. Regulators should not demand that developers prove that intelligent vehicle technology will not cause any harm.

In a new study for the Mercatus Center at George Mason University, scholars Adam Thierer and Ryan Hagemann argue that “permissionless innovation”—the primary driver of entrepreneurialism and economic growth in many sectors of the economy—should be the default principle for policymakers addressing the rise of intelligent vehicles. Any perceived or actual problems with new technologies can be corrected later through better-informed policymaking. 


Policymakers should focus on clearing existing roadblocks to the development of intelligent vehicles, and exercise restraint regarding hypothetical concerns about their use. “Permissionless innovation” is the idea that experimentation with new technologies and business models should generally be permitted by default. Permissionless innovation brought the Internet, an open and lightly regulated platform that allows entrepreneurs to adopt new business models and offer new services without first seeking approval from regulators.

Increased use of intelligent vehicle technology will bring social and economic challenges, but governments should maintain a flexible system that deals with real problems rather than hypothetical ones. Additionally, the problems that may arise due to intelligent vehicles should be understood in context. For example, in 2012, 33,561 people were killed and 2,362,000 injured in traffic crashes, largely as a result of human error. Reducing the number of accidents by allowing intelligent vehicle technology to flourish would be a success. If regulators hinder accident-reducing innovative technology, avoidable injuries and deaths will continue unnecessarily.


  • Driver errors resulting in accidents cost $300 billion annually in the United States. While intelligent vehicles will not be 100 percent accurate 100 percent of the time, they will likely achieve a level of control and awareness that no human could possess, thus reducing the economic impact of accidents. Additionally, insurance premiums could fall or even disappear entirely.
  • Intelligent vehicles will also reduce congestion and lower fuel consumption. In 2011 congestion caused drivers to spend an extra 5.5 billion hours on the road and purchase 2.9 billion gallons of fuel at a cost of $121 billion. Intelligent vehicles will help reduce human-initiated driving errors, allowing vehicles to travel at higher speeds and closer together, reducing congestion costs.
  • Increased use of intelligent vehicles may cause some sectors of the economy to change or disappear entirely, such as professional driving of taxis, buses, and trucks. However, policymakers should not choose winners and losers in the market; that benefits entrenched industries, not consumers. If regulators in the early 20th century had curtailed the development of the automobile for the sake of carriage drivers and woodworkers, whose livelihoods depended on horse-drawn carriages, the world might never have seen Henry Ford’s Model T.


  • A cultural shift toward roadways with entirely driverless vehicles will not happen overnight: comprehensive change takes time. As consumer demand for intelligent vehicles increases, market penetration will increase in proportion. While some may resist changing traditional cultural norms about hitting the “open road” as a driver, eventually most people will accept the benefit and value of accommodating a cultural shift.
  • Security concerns may slow the adoption of intelligent vehicles, but concerns over remote car-hacking are likely overblown. Manufacturers have powerful reputational incentives to continuously improve the security of their systems and adopt best practices within the industry, much as the information-technology sector has taken steps to secure its networks. Legislation and regulation is unnecessary and could hinder growth in the industry.
  • Intelligent vehicles also raise privacy concerns. Manufacturers and application developers would be wise to develop best practices for data retention, consumer consent to collection and sharing, and safeguarding collected data. But imposing burdensome privacy regulations during the development of this technology would create complex and costly tradeoffs, possibly resulting in higher costs for consumers, a decrease in content and services, and less competition in the market. 


The government’s approach here should be  guided by humility and patience, allowing intelligent vehicle technology to develop while refraining from overbearing regulation. Lawmakers should sunset any laws that inhibit innovation and experimentation. Policymakers should also examine infrastructure and network operations, as well as licensing issues. In the private sector, businesses should work together and with policymakers to overcome hurdles to the widespread adoption of intelligent vehicle technology, and stakeholders should develop clear and fully transparent guidelines and best practices to allay safety, security, and privacy concerns. 

Additionally, government data collection should be constrained to the fullest extent possible. Consensual data collection should be allowed between consumers and producers of goods and services, which will translate to practical benefits, cheaper systems, and a more robust marketplace.


Policymakers should embrace permissionless innovation when dealing with intelligent vehicle technologies. They should not live in fear of hypothetical worst-case scenarios related to security, safety, and privacy. While disruptive at times, these new technologies will bring incredible economic and social benefits to society. In the near future, it will be very difficult to use a car to hurt yourself or others. The sooner that day arrives, the better.