Full text for verification
How to Regulate Disruptive Innovation - From Facts to Data
Canonical record: https://ssrn.com/abstract=2808044
This is an alternate SSRN posting of the same work published at 2808132. Its 52 claims are recorded under that identifier.
Source extraction SHA-256: 5e753ae0b4fa50fb445ead47f47948376ff079251f1e83445da21991a09ad416
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
> * Associate Professor, University of St. Thomas School of Law (Minneapolis), United States of America. The authors would like to thank the participants at the 2015 Annual Meeting of the Midwestern Law and Economics Association (MLEA), the participants at the 2016 First Conference on Empirical Legal Studies in Europe (CELSE), the participants at the 2016 National Business Law Scholars Conference (NBLSC), and the participants at the18th Biennial Conference of International Academy of Commercial & Consumer Law 2016. The authors are also particularly grateful for outstanding support from librarians Nick Farris and Ann Bateson.
> ** Professor of Business & Financial Law, Tilburg Law School - Department of Business Law, Head of Governance, Philips Lighting, Netherlands.
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
# ABSTRACT
Disruptive innovation creates increasing regulatory challenges. The reason for this is simple: Regulation is usually reactive, responding to facts. However, we currently live in a world of data, not facts. Regulation must therefore be proactive and dynamically responsive to data and trends. This paper discusses the following questions: (1) Why should regulators be proactive?, (2) When should regulators respond?, (3) What should regulators respond to? and (4) How should regulators respond? Using a dataset comprising over 77,000 venture capital deals with over 35,000 companies in the United States from 2005 to 2015, the paper shows how investment data can provide important feedback on innovation trends and associated risks for regulators, optimize the timing of regulation, and support anticipatory rulemaking.
**_Keywords:_** _Disruptive Innovation, Venture Capital, Venture Investments, Dynamic Regulation, Feedback Effects, Optimized Information for Regulation, Anticipatory Regulation, Big Data_
**_JEL Classification_ :** K20, K23, K32, L43, L5, O31, O32
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
# TABLE OF CONTENTS
|I.|INTRODUCTION............................................................................................................................................. 4|
|---|---|
|II.|CHARACTERISTICS OFDISRUPTIVEINNOVATION............................................................................. 8|
|III.|DISRUPTIVEINNOVATION AS AREGULATORYCHALLENGE................................................. 13|
|_1._|_Ex Post Facts-Based (Trial-and-Error) Rulemaking ................................................................. 19_|
|_2._|_Stable and Presumptively Optimal Rules ....................................................................................... 19_|
|_3._|_Timing .......................................................................................................................................................... 20_|
|_4._|_Unknown Future Contingencies ........................................................................................................ 21_|
|IV.|DYNAMICREGULATION AS AREGULATORYSUPPLEMENT.................................................. 23|
|V.|DATA ASINPUT FORDYNAMICREGULATION................................................................................. 26|
|_1._|_VC Deal Characteristics ....................................................................................................................... 27_|
|_2._|_Industries Represented .......................................................................................................................... 34_|
|VI.|VENTURECAPITAL ASDYNAMICREGULATION OFDISRUPTIVEINNOVATION.............. 41|
|VII.|CONCLUSION......................................................................................................................................... 43|
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
# **I. Introduction**
We show that data derived from venture capital investments can function as a dynamic regulatory supplement that can help address the legal challenges presented by disruptive innovation. Venture capital’s financial allocations to innovative projects can provide feedback for dynamic regulation of disruptive innovation. We show how this approach can help optimize the timing of regulation and provide anticipatory rules. We use a PitchBook Data, Inc. dataset on venture capital investments with over 75,000 venture capital deals to illustrate that venture capital finance allocation and deal flow, among other venture capital related data, provide institution-specific decentralized information on innovation trends that can provide feedback effects for dynamic regulation of disruptive innovation. Venture capitalists’ investment allocations can measure both the risks associated with innovative developments as well as the economic opportunities associated with such innovation.
Disruptive innovation has shaped and reshaped industries throughout history. Disruptive innovation can take the form of new business models, new products or services, or new ways of delivering existing products and services to consumers, while addressing unmet consumer needs.<sup>1</sup> Disruptive innovation is characterized by scientific discoveries that change existing technological product paradigms and provide the foundation for more competitive new technologies and products to emerge.<sup>2</sup> I t is often associated with the emergence of completely new technologies, new combinations and applications of existing technologies, and the application of new technologies to specific societal problem areas, each precipitating a significant paradigm shift for product technology or creating entirely new paradigms.<sup>3</sup> The combination of these factors facilitates sudden exponential improvements in the value proposition for customers by enabling discontinuous innovations of processes, products, and services.<sup>4</sup>
> 1 DIRECTORATE FOR FINANCIAL AND ENTERPRISE AFFAIRS COMPETITION COMMITTEE, HEARING ON DISRUPTIVE INNOVATION (2015), https://goo.gl/3kSCi6.
2 The original “disruptive innovation” terminology came from the famous Bower and Christensen article in the _Harvard Business Review_ . Joseph L. Bower & Clayton M. Christensen, _Disruptive Technologies: Catching the Wave_ , HARV. BUS. REV., Jan.-Feb. 1995, at 43, 44 (“The technological changes that damage established companies are usually not radically new or difficult from a technological point of view. They do, however, have two important characteristics: First, they typically present a different package of performance attributes - ones that, at least at the outset, are not valued by existing customers. Second, the performance attributes that existing customers do value improve at such a rapid rate that the new technology can later invade those established markets.”). There is no widely accepted definition for discontinuous and disruptive innovation. However, scholars focus on different aspects, such as industrywide product technology factors to the gap between substitutable technological learning curves on a cost or performance basis, definitions of discontinuous innovation focus on customer behaviour, product newness, market factors, or a combination of these factors. For an overview of the disruptive innovation definitions used in the literature, _see_ Ronald N. Kostoff et al., _Disruptive Technology Roadmaps_ , 71 TECH. FORECASTING & SOCIAL CHANGE 141, 142-44 (2004).
> 3 Kostoff, _supra_ note 2.
> 4 _See_ Robert W. Veryzer, Jr., _Discontinuous Innovation and the New Product Development Process_ , 15 J. PROD. INNOV. MGMT. 304 (1998); Gary S. Lynn et al., _Marketing and Discontinuous Innovation: The Probe and Learn Process_ , 38 CAL. MGMT. REV. 8 (1996). _See also_ Steven T. Walsh, _Roadmapping a Disruptive Technology: A Case Study, The Emerging Microsystems and Top-Down Nanosystems Industry_ , 71 TECH. FORECASTING & SOC. CHANGE 161, 165 (2004) (“Two views of disruptive technologies arise from this literature. One emphasizes the ‘different’ nature of the technology, whereas the other
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
Disruptive innovation presents significant regulatory challenges. Disruptive technology convergence can bring with it unexpected and sudden changes in areas that are already quietly incubating. These new developments and trends are going to increasingly impact consumer preferences and behavior, products, the financial industry,<sup>5</sup> industry practices, delivery channels, and possible regulatory responses to changes in any of these. The ex-post facts-based and trial-and-error-rulemaking with stable and presumptively optimal rules<sup>6</sup> in the existing regulatory framework<sup>7</sup> often produces suboptimal regulatory outcomes that are no longer sustainable in an environment of exponential disruptive innovation. While ex-post facts-driven regulation with stable and presumptively optimal rules in the historically evolved legal infrastructure was
emphasizes the emerging high technology nature. Bower and Christensen champion the ‘different’ nature of disruptive technologies emphasizing that they may not be ‘radically new from a technological point of view’ but have superior ‘performance trajectories’ along critical dimensions that customers value. Abernathy and Clark and others argue that disruptive technologies’ radical newness or emergent character is an important element in its definition since robust manufacturing infrastructure is either limited or nonexistent.”).
> 5 Chris Brummer, _Disruptive Technology and Securities Regulation_ , 84 FORDHAM L. REV. 977, 977 (2015). (“This Article explains how technological innovation ‘disrupts’ not only capital markets but also the exercise of regulatory supervision and oversight. It provides the first theoretical account tracking the migration of technology across multiple domains of today’s securities infrastructure and argues that an array of technological innovations are facilitating what can be understood as the disintermediation of the traditional gatekeepers that regulatory authorities have relied on (and regulated) since the 1930s for investor protection and market integrity.”).
6 The existing regulatory infrastructure, including Congress, regulatory agencies, self-regulatory bodies, and the literature on regulation, relies almost exclusively on stable and presumptively optimal rules. Stable and presumptively optimal rules are created to address perceived regulatory issues identified by lawmakers through centralized information in the then-existing economic and market conditions. Stable and presumptively optimal rules, that have sufficient political support at the time of drafting, are tailored and drafted to attain permanent solutions for perceived regulatory issues. _See_ Wulf A. Kaal, _Evolution of Law: Dynamic Regulation in a New Institutional Economics Framework in_ FESTSCHRIFT ZU EHREN VON CHRISTIAN KIRCHNER 1211, 1212 (Wulf A. Kaal, Matthias Schmidt & Andreas Schwartze eds., 2014) [hereinafter Kaal in FS Kirchner]. (“[T]he institutional infrastructure for rulemaking was geared towards the creation of rules for governing a relatively stable society with less upward mobility and relatively stable economic and market environments.”); Wulf A. Kaal, _Dynamic Regulation of the Financial Services Industry,_ 48 WAKE FOREST L. REV. 791 (2013) [hereinafter Kaal, Wake Forest]; Wulf A. Kaal, _Dynamic Regulation via Governmental Contracts, in_ FESTSCHRIFT FÜR PETER NOBEL ZUM 65, 70 (Peter Sester et al. eds., 2015), http://ssrn.com/abstract=2517677 [hereinafter Kaal in Liber Amicorum Peter Nobel]; Wulf A. Kaal & Timothy Lacine, _The Effect of Deferred and Non-Prosecution Agreements on Corporate Governance: Evidence from 1993-2013,_ 70 BUS. LAW. 61 (2014) [hereinafter Kaal & Lacine 2014]. 7 _See_ KARL POPPER, THE POVERTY OF HISTORICISM (1957) _;_ Kaal in FS Kirchner _, supra_ note 6, at 1212 (“Since then, society and markets have evolved rapidly and are becoming increasingly more complex. The growing number of rule enactments, revisions, and revocations suggests that existing rules and institutional structures for rulemaking are becoming less capable of addressing the rapid pace of change.”).
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
sufficiently able to cope with its inherent collective action problems,<sup>8</sup> regulatory cycles,<sup>9</sup> trial-and-error rulemaking,<sup>10</sup> systemic constraints, and path dependencies, the exponential growth of disruptive innovation is likely to expose the depth of insufficiencies and design flaws of the existing regulatory and institutional infrastructure. More specifically, in an environment of exponential disruptive innovation, relevant information for rulemaking is less likely to materialize soon enough for traditional rulemaking to be effective, regulatory issues become increasingly complex, and unknown future contingencies increase substantially in the rulemaking process.<sup>11</sup> In short, exponential disruptive innovation has the potential to overwhelm the existing regulatory process.<sup>12</sup>
Examples of disruptive innovation creating regulatory challenges include humanrobot collaboration, constant body monitoring, and driverless cars.<sup>13</sup> Each of these examples illustrates the “pacing problem” between innovation and regulation,<sup>14</sup> e.g., innovation develops faster than applicable regulation. In perhaps the most timely and prominent example, the driverless car, regulatory developments cannot stay abreast of technological advances because: (1) the driverless car generates an enormous amount of data for ever accelerating alternative usage, creating unknown new legal issues related to data security and privacy concerns,<sup>15</sup> among many others; (2) driverless cars
> 8 John C. Coffee, Jr., _The Political Economy of Dodd-Frank: Why Financial Reform Tends To Be Frustrated and Systemic Risk Perpetuated_ , 97 CORNELL L. REV. 1019, 1020-22 (2012); Kaal in Liber Amicorum Peter Nobel, _supra_ note 6; Kaal, Wake Forest, _supra_ note 6; Kaal in FS Kirchner, _supra_ note 6. Smaller and better-organized special interest groups typically control dispersed investors and other dominant latent groups in the competition to influence the rulemaking process. MANCUR OLSON, JR., THE LOGIC OF COLLECTIVE ACTION (1965). The predominance of special interest groups in the rulemaking process can be temporarily overcome during and after crises when political entrepreneurs assume the transaction costs of organizing the otherwise disinterested latent groups. ELINOR OSTROM, GOVERNING THE COMMONS: THE EVOLUTION OF INSTITUTIONS FOR COLLECTIVE ACTION (1990).
> 9 Coffee, _supra_ note 8; Kaal in Liber Amicorum Peter Nobel, _supra_ note 6; Kaal in FS Kirchner, _supra_ note 6.
> 10 Kaal, Wake Forest _, supra_ note 6; Kaal in FS Kirchner, _supra_ note 6; Popper, _supra_ note 7; Christian Kirchner, _Evolution of Law_ **_̶_** _: Interplay Between Private and Public Rule-Making ─ A New Institutional Economics Analysis,_ 4 ERASMUS L. REV. 161 (2012).
> 11 See _infra_ notes 94-100 and accompanying text.
12 Because rulemakers are still attempting to comprehend the regulatory demands of the last wave of innovative solutions, technologies, and applications while the next wave of innovation is already in full force, exponential disruptive innovation is likely to intensify the frequencies of the regulatory sine curve that is recognized by the literature. Kaal, Wake Forest, _supra_ note 6; Coffee, _supra_ note 8, at 1029-37; John C. Coffee, Jr., _Systemic Risk After Dodd-Frank: Contingent Capital and the Need for Regulatory Strategies Beyond Oversight_ , 111 COLUM. L. REV. 795, 821 (2011).
> 13 Gavriel Avigdor & Rene Wintjes, _Trend Report: Disruptive Innovations and Forward-Looking Policies Towards Smart Value Chains_ 5-7 (European Union, Business Innovation Observatory No. 190, 2015), http://goo.gl/PUwnwn.
> 14 Wulf A. Kaal, _Dynamic Regulation for Innovation_ , in: PERSPECTIVES IN LAW, BUSINESS & INNOVATION, Mark Fenwick, Wulf A. Kaal, Toshiyuki Kono & Erik P.M. Vermeulen eds., New York Springer (2016), _see also_ Marchant 2011, Moses 2013, Butenko and Larouche 2015, Moses 2011, McFrath 2013, Desliver 2014, Hall and Khan 2003, Garrett 2015, Marchant 2011 (ALL CITED IN BIBLIOGRAPHY KAAL, DYNAMIC REGULATION FOR INNOVATION).
> 15 Mark van Rijmenam, _Self-driving Cars Will Create 2 Petabytes of Data, What Are The Big Data Opportunities for the Car Industry?,_ DATAFLOQ (July 19, 2016), https://datafloq.com/read/self-drivingcars-create-2-petabytes-data-annually/172 (“The self-driving car from Google already is a true data creator. With all the sensors to enable the car to drive without a driver, it generates nearly 1 Gigabyte every second. It uses all that data to know where to drive and how fast to drive. It can even detect a new cigarette butt
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
communicating among themselves, with other and ever increasingly newly developed electronic platforms, devices, applications and with the transport infrastructure requires regulators to safeguard telecommunication frequencies and protect against ever changing security threats, such as “car-hacking”;<sup>16</sup> (3) the data created by driverless cars creates the basis for rapidly evolving new innovations, platform business models, application developments etc., that create ever increasing unknown future contingencies for rulemaking. While these issues can be addressed over time, our main point is that by the time legal issues are addressed, new and different legal issues are created at an ever accelerating and unprecedented rate. The main reason why these issues cannot be effectively addressed in the future is associated with regulators using facts for rulemaking and not data.
We apply a data-driven approach that enables dynamic regulation as established by Kaal<sup>17</sup> to regulatory issues associated with disruptive innovation. The theory of dynamic regulation conceptualizes the study of regulatory phenomena in relation to preceding and succeeding events, using institution-specific and decentralized information to facilitate feedback effects for anticipatory rulemaking.<sup>18</sup> The dynamic regulatory framework allows rulemaking to overcome its historic constraints and path dependencies that traditionally focused the rulemaking process on stable and presumptively optimal rules. Evolving from a historically reactive process based only on preceding events/facts and driven by the agency and collective-action problems of rulemaking, rulemaking in a dynamic framework allows for the anticipation of future innovation trends and their associated contingencies. Most important for the context of this article, the rapid pace of technological developments in the telecommunications industry necessitated the use of dynamic elements in regulation.<sup>19</sup> The exponential growth of innovation discussed in this
thrown on the ground and it then knows that a person might appear all of a sudden from behind a corner or car. 1 Gigabyte per second, imagine the amount of data that will create every year: On average, Americans drive 600 hours per year in their car. That equals 2.160.000 seconds or approximately 2 Petabyte of data per car per year. With the amount of cars worldwide to surpass one billion, it is almost unimaginable how much data will be created when Google’s self-driving car will become common on the streets.”); Adrienne LaFrance, _How Self-Driving Cars Will Threaten Privacy,_ THEATLANTIC (Mar 21, 2016), http://www.theatlantic.com/technology/archive/2016/03/self-driving-cars-and-the-looming-privacyapocalypse/474600/ (“The companies building self-vehicles have been cagey, so far, about how they’re thinking about using individual data.”)
> 16 Jeffrey K. Gurney, _Driving into the Unknown: Examining the Crossroads of Criminal Law and Autonomous Vehicles_ , 5 WAKE FOREST J.L. & POL'Y 393, 433 (2015) (“In addition to physically interfering with an autonomous vehicle, people will be able to virtually interfere with the operation of an autonomous vehicle, also known as hacking.”); _see also_ Tom Simonite, _Your Future Self-Driving Car Will Be Way More Hackable_ , MIT TECH. REV. (Jan. 26, 2016), https://www.technologyreview.com/s/546086/yourfuture-self-driving-car-will-be-way-more-hackable/ (“’We are a long way from securing the nonautonomous vehicles, let alone the autonomous ones,’ said Stefan Savage, a computer science professor at the University of California, San Diego, at the Enigma security conference in San Francisco on Tuesday. The extra computers, sensors, and improved Internet connectivity required to make a car drive itself increase the possible weak points, he said. ‘The attack surface for these things is even worse,’ said Savage.”).
> 17 Kaal in Liber Amicorum Peter Nobel, _supra_ note 6; Kaal, Wake Forest, _supra_ note 6; Kaal in FS Kirchner, _supra_ note 6.
> 18 _Id_ at 11 _._
> 19 _See_ Johannes M. Bauer & Erik Bohlin, _From Static to Dynamic Regulation: Recent Developments in US Telecommunications Policy_ , 43 INTERECONOMICS 38, 40–42 (2008) (discussing competition in telecommunications); Paul W.J. de Bijl & Martin Peitz, _Dynamic Regulation and Entry in_
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
article<sup>20</sup> shows some parallels to the pace of technological developments in telecommunications markets.
This article has seven parts. After this brief introduction in Part I, Part II defines the phenomenon of disruptive innovation. Part III analyzes the regulatory challenges presented by disruptive innovation. Part IV introduces the theoretical framework for dynamic regulation before Part V introduces the data and evaluates different venture capital metrics that provide feedback for regulators on innovation. In Part VI we discuss how the information and feedback provided by venture capital data can be used as a dynamic supplement for regulation. Part VII concludes.
# **II. Characteristics of Disruptive Innovation**
Disruptive innovation is characterized by scientific discoveries that change the existing technological product paradigms and provide the foundation for more competitive new technologies and products to emerge.<sup>21</sup> By enabling discontinuous innovations of processes, products, and services, disruptive technological innovations facilitate exponential improvements in the value proposition for customers.<sup>22</sup> Disruptive technological innovation can be characterized by the emergence of completely new technologies, the new combination and application of existing technologies, and the application of new technologies to specific societal problem areas, each precipitating a significant paradigm shift for product technology or creating entirely new paradigms.<sup>23</sup> Disruptive innovation exemplifies Schumpeter’s “Creative Destruction,”<sup>24</sup> i.e., the creation of competitive strength through innovation that is followed by new demand in new markets while destroying older and less competitive technologies and existing markets that are based on older and less competitive technologies.<sup>25</sup>
_Telecommunications Markets: A Policy Framework_ , 16 INFO. ECON. & POL’Y 411, 411 (2004); Machiel van Dijk & Machiel Mulder, _Regulation of Telecommunication and Deployment of Broadband_ (CPB Netherlands Bureau for Economic Policy Analysis No. 131, Dec. 2005), <u>https://ideas.repec.org/p/cpb/memodm/131.html</u> (exploring whether telecommunication regulations encourage or hamper new technological development).
> 20 See _infra_ notes 21-46 and accompanying text.
> 21 _See supra_ note 2.
> 22 _See supra_ note 4.
> 23 Kostoff et al., _supra_ note 2, at 142; Steven T. Walsh & Jonathan D. Linton, _Infrastructure for Emergent Industries Based on Discontinuous Innovations_ , 12 ENG. MGMT. J. 23, 24 (2000).
> 24 Schumpeter described the “gaels of creative destruction,” often unleashed by technology, that periodically sweep through industries and sink weak and outdated firms. JOSEPH A. SCHUMPETER, CAPITALISM, SOCIALISM, AND DEMOCRACY 83-84 (3rd ed.1962) (“The opening up of new markets, foreign or domestic, and the organizational development from the craft shop and factory to such concerns as U.S. Steel illustrate the process of industrial mutation that incessantly revolutionizes the economic structure from within, incessantly destroying the old one, incessantly creating a new one. This process of Creative Destruction is the essential fact about capitalism. It is what capitalism consists in and what every capitalist concern has got to live in. . . . Every piece of business strategy acquires its true significance only against the background of that process and within the situation created by it. It must be seen in its role in the perennial gale of creative destruction; it cannot be understood irrespective of it or, in fact, on the hypothesis that there is a perennial lull.”)
> 25 Aron S. Spencer & Bruce A. Kirchhoff, _Schumpeter and New Technology Based Firms: Towards a Framework for How NTBFs Cause Creative Destruction_ , 2 INT’L ENTREPRENEURSHIP & MGMT. J. 145, 148 (2006) (“However, the new entrant is able to find or create a market for the innovation, where it matures, and improves in performance. When the performance achieves rough parity with the old product, the new product sweeps through the market, offering equal or superior performance along with a smaller size and
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
The literature on management has studied the implications of disruptive technologies and innovation since the mid 1990s.<sup>26</sup> The literature warned businesses that disrupter startups would attract new lower-end consumers by offering inexpensive substitutes for products and would then gradually move upmarket by attracting higherend consumers, suggesting that businesses act quickly when disrupters appear and either acquire the disrupter or incubate a competing business that embraces the disruptive technology.<sup>27</sup> A market leader’s lack of investment in disruptive technologies often results in the abrupt loss of market dominance and often even total replacement in such markets.<sup>28</sup> Market leaders often shortsightedly refuse to cannibalize their market dominance through the use of disruptive technologies. This enables small entrepreneurial firms with no established customer base to take advantage of disruptive technologies and redefine such markets.<sup>29</sup> Market leaders, successful institutions, and managers fail when they do not distinguish sustaining technologies from disruptive technologies.<sup>30</sup>
reduced price. The old product, and frequently the firm that produced it, exit the market. It is quite clear that this phenomenon is, in fact, ‘creative destruction’.”).
> 26 CLAYTON M. CHRISTENSEN, THE INNOVATOR’S DILEMMA: WHEN NEW TECHNOLOGIES CAUSE GREAT FIRMS TO FAIL (1997); GEOFFREY MOORE, CROSSING THE CHASM (1991).
> 27 Bower & Christensen, _supra_ note 2; LARRY DOWNES & PAUL NUNES, BIG BANG DISRUPTION (2013). _Contra_ Andrew A. King & Baljir Baatartogokh, _How Useful Is the Theory of Disruptive Innovation_ ?, 57 MIT SLOAN MGMT. REV. 77 (2015) (“In summary, although Christensen and Raynor selected the 77 cases as examples of the theory of disruptive innovation, our survey of experts reveals that many of the cases do not correspond closely with the theory. In fact, their responses suggest that only seven of the cases (9%) contained all four elements of the theory that we asked about.”).
> 28 Bower & Christensen, _supra_ note 2, at 43 (“One of the most consistent patterns in business is the failure of leading companies to stay at the top of their industries when technologies or markets change.”).
> 29 _Id_ . at 51 (“Small, hungry organizations are good at placing economical bets, rolling with the punches, and agilely changing product and market strategies in response to feedback from initial forays into the market.”); Constantinos D. Charitou & Constantinos C. Markides, _Responses to Disruptive Strategic Innovation_ , 44 MIT SLOAN MGMT. REV. 55, 56 (2003) (“Second, disruptive strategic innovations usually start out as small and low-margin businesses. That’s why they rarely gain support or long-term commitment from established competitors. The innovations are small and are not attractive until they start growing.”); _Contra_ Larry Downes & Paul Nunes, _Big-Bang Disruption_ , HARV. BUS. REV., MAR. 2013, at 44, 46, https://hbr.org/2013/03/big-bang-disruption/ (“The strategic model of disruptive innovation we’ve all become comfortable with has a blind spot. It assumes that disrupters start with a lower-priced, inferior alternative that chips away at the least profitable segments, giving an incumbent business time to start a skunkworks and develop its own next-generation products. That advice hasn’t been much help to navigation-product makers like TomTom, Garmin, and Magellan. Free navigation apps, now preloaded on every smartphone, are not only cheaper but better than the stand-alone devices those companies sell. And thanks to the robust platform provided by the iOS and Android operating systems, navigation apps are constantly improving, with new versions distributed automatically through the cloud.”).
> 30 Sustaining technologies are technologies that improve the performance of existing products through close cooperation with customers. By contrast, disruptive technologies provide value parameters not recognized by the mainstream market and in the short-run can even show less product performance in comparison with the existing products that are valued by the mainstream market. Kostoff et al., _supra_ note 2, at 144 (“The near-term individual rewards from sustaining technologies that yield short-term lowrisk payoffs displace the longer term social benefits that could result from proactive high-risk high-payoff disruptive technology selection. The procedural problem is that technology selection decisions, especially in large established commercial and government organizations, are increasingly being made by larger and more inclusive committees, a process traditionally steeped in tradition and conservatism. Revolutionary disruptive concepts are less likely (on average) to receive committee approval than evolutionary sustaining concepts.”).
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
Successful disruptive products of the last fifteen years displayed shared core characteristics that were facilitated by growth and advancement in core disruptive technologies. Revolutionizing products in the past few decades that were based on disruptive technological innovation were typically smaller in size, lighter, cheaper, more convenient and flexible, more reliable, had better unit performance with higher efficiency, and were more user friendly than prior products.<sup>31</sup> These characteristics typically require significant advances and growth in artificial intelligence and technologies including micro- or nanotechnologies, as well as materials and component technologies.
Disruptive innovation via so-called big-bang disruptions differs from moretraditional innovation. Big-bang disruptors not only often offer a much cheaper product in comparison with the established products, but they are also better integrated with other products and services and are often more inventive.<sup>32</sup> In today’s age of social media, bigbang disruptors can successfully exploit consumers’ advanced access to product information and consumers’ ability to contribute to products and share such contributions and products improvements via social media and other outlets.<sup>33</sup> Internet fads and products can be accessible to consumers in the entire world within a matter of days.<sup>34</sup> Big-bang disrupters typically launch businesses without a foundation, using cloud computing, open platforms built on the internet, and fast-cycling mobile devices. Bigbang disrupters often produce multiple new products to identify which products may take hold in the market. While most such products will fail, the payoff associated with the unconstrained growth of those products that succeed is substantial and facilitates and often accelerates other forms of disruptive innovation.<sup>35</sup>
Big data is a significant driver of disruptive innovation. The idea of N=All, facilitated by big data, allows researchers to understand correlations that are completely unprecedented and to revolutionize our world.<sup>36</sup> Big data in the form of digitized data
> 31 Kostoff et al., _supra_ note 2, at 143; Walsh, _supra_ note 4, at 170 (“Bower and Christensen champion the ‘different’ nature of disruptive technologies emphasizing that they may not be ‘radically new from a technological point of view’ but have superior ‘performance trajectories’ along critical dimensions that customers' value. Abernathy and Clark and others argue that disruptive technologies' radical newness or emergent character is an important element in its definition since robust manufacturing infrastructure is either limited or nonexistent.”).
> 32 Downes & Nunes, _supra_ note 29, at 46 (“The first key to survival is understanding that big-bang disruptions differ from more-traditional innovations not just in degree but in kind. Besides being cheaper than established offerings, they’re also more inventive and better integrated with other products and services. And today many of them exploit consumers’ growing access to product information and ability to contribute to and share it.”).
> 33 Steve Denning, _The Business Disease with No Cure: Big Bang Disruption_ , FORBES (Feb. 22, 2014, 3:55PM), http://www.forbes.com/sites/stevedenning/2014/02/22/the-business-disease-without-a-cure-bigbang-disruption/ (“And, in today’s world of ‘near perfect market information’ and social media, the best quickly becomes a function of social consensus, not objective reality. That is why we say marketing’s main function has shifted from selling to creating social consensus.”).
> 34 For instance, consumers downloaded an ad-supported version of a game called ”Angry Birds” over a million times within the first 24 hours of its release. Seven months thereafter, consumers downloaded that game more than 200 million times. Downes & Nunes, _supra_ note 29, at 46.
> 35 _Id_ . at 48 (“ In the bizarro world of big-bang disrupters, it is perfectly rational to churn out dozens of new products and see which ones take hold. Like venture capital investments, most will fail outright. But just one success can pay off big.”).
> 36 KENNETH CUKIER & VIKTOR MAYER-SCHONBERGER, BIG DATA: A REVOLUTION THAT WILL TRANSFORM HOW WE LIVE, WORK AND THINK (2013).
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
that grows at exponential rates and can be captured and manipulated electronically draws on several core sources including the internet of things, public records, social media, and cameras, as well as satellite tracking.<sup>37</sup> Big data benefits not only industry and researchers; it also increases consumer choice through publicly available websites providing big data analyses intended to support consumers’ decision-making processes. Despite the overwhelming benefits associated with big data, big data’s use of such disparate sources creates challenges for the integration of data and normalizing.<sup>38</sup> The literature mostly takes issue with the big data inaccuracies that occur when collecting data from databases that merely require generalized accuracy.<sup>39</sup> Big data is often not the output of instruments designed to generate valid and reliable data suitable for scientific analysis. Foundational data issues of construct validity, measurement, reliability, and data dependencies are the same regardless of data quantities.<sup>40</sup> Some critique the implicit
> 37 _Id_ .; EXEC. OFFICE OF THE PRESIDENT, BIG DATA: SEIZING OPPORTUNITIES, PRESERVING VALUES 53 (2014), https://goo.gl/xtDerj (“Big data technologies, together with the sensors that ride on the ‘Internet of Things,’ pierce many spaces that were previously private. Signals from home WiFi networks reveal how many people are in a room and where they are seated.”); James Manyika et al., _Big Data: The Next Frontier for Innovation, Competition, and Productivity_ 15 (McKinsey Global Institute Report, May 2011), http://goo.gl/GBzeLR (“We estimate that new data stored by enterprises exceeded 7 exabytes of data globally in 2010 and that new data stored by consumers around the world that year exceeded an additional 6 exabytes. To put these very large numbers in context, the data that companies and individuals are producing and storing is equivalent to filling more than 60,000 US Libraries of Congress. If all words spoken by humans were digitized as text, they would total about 5 exabytes—less than the new data stored by consumers in a year. The increasing volume and detail of information captured by enterprises, together with the rise of multimedia, social media, and the Internet of Things will fuel exponential growth in data for the foreseeable future.”).
> 38 Manyika et al., _supra_ note 46, at 12 (“To capture value from big data, organizations will have to deploy new technologies (e.g., storage, computing, and analytical software) and techniques (i.e., new types of analyses). The range of technology challenges and the priorities set for tackling them will differ depending on the data maturity of the institution. Legacy systems and incompatible standards and formats too often prevent the integration of data and the more sophisticated analytics that create value from big data. New problems and growing computing power will spur the development of new analytical techniques. There is also a need for ongoing innovation in technologies and techniques that will help individuals and organizations to integrate, analyze, visualize, and consume the growing torrent of big data.”); Christian Bizer et al., _The Meaningful Use of Big Data: Four Perspectives—Four Challenges_ , SIGMOD REC., Dec. 2011, at 56, 57. (“My challenge is meaningful data integration in the real, messy, often schema-less, and complex Big Data World of databases and the (Semantic) Web using multi-disciplinary, multi-technology methods.”).
> 39Bizer et al., _supra_ note 47, at 57 (“The Web is an open medium in which everybody can publish data on the Web. As the classic document Web, the Web of Data contains data that is outdated, conflicting, or intentionally wrong (SPAM). Thus, one of the main challenges that Linked Data applications need to handle is to assess the quality of Web data and determine the subset of the available data that should be treated as trustworthy.”); IAN AYRES, SUPER CRUNCHERS: WHY THINKING-BY-NUMBERS IS THE NEW WAY TO BE SMART 60-63 (2007) (identifying big data phenomenon); Andrea Lancichinetti et al., _HighReproducibility and High-Accuracy Method for Automated Topic Classification_ , 5 PHYSICAL REV. X 011007 (2015) (detailing the inaccuracy problems associated with big data text analysis/machine reading (called LDA)).
> 40 _See_ David Lazer et al., _The Parable of Google Flue: Traps in Big Data Analysis_ , 343 SCIENCE 1203 (2014); Gary Marcus & Ernest Davis, _Eight (No, Nine!) Problems with Big Data_ , N.Y. TIMES (Apr. 6, 2014), http://www.nytimes.com/2014/04/07/opinion/eight-no-nine-problems-with-big-data.html (“Reliable statistical information can be compiled about common trigrams, precisely because they appear frequently. But no existing body of data will ever be large enough to include all the trigrams that people might use, because of the continuing inventiveness of language.”); Tim Harford, _Big Data: Are We Making a Big_
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
assumption of big data researchers that big data is a substitute for traditional data collection and analysis rather than a mere supplement<sup>41</sup> while others see big data and small data not as mutually exclusive but as reinforcing and supporting each other.<sup>42</sup> Big data shortcomings can largely be addressed with artificial intelligence, and the combination of big data and artificial intelligence opens up significant additional big data applications.<sup>43</sup>
Artificial intelligence (AI) presents another significant field of disruptive innovation. AI is different from a regular computer algorithm. AI tries to emulate human
_Mistake?_ , FIN. TIMES (Mar. 28, 2014), http://goo.gl/2dyox2 ("Statisticians have spent the past 200 years figuring out what traps lie in wait when we try to understand the world through data. The data are bigger, faster and cheaper these days – but we must not pretend that the traps have all been made safe. They have not.”).
> 41 Jeff Leek, _Why Big Data Is in Trouble: They Forgot About Applied Statistics_ , SIMPLYSTATS (May 7, 2014), http://goo.gl/2dyox2 (“Statistical thinking has also been conspicuously absent from major public big data efforts so far.”).
> 42 Lazer et al., _supra_ note 49, at 1205 (“There is a tendency for big data research and more traditional applied statistics to live in two different realms—aware of each other's existence but generally not very trusting of each other. Big data offer enormous possibilities for understanding human interactions at a societal scale, with rich spatial and temporal dynamics, and for detecting complex interactions and nonlinearities among variables. We contend that these are the most exciting frontiers in studying human behavior. However, traditional ‘small data’ often offer information that is not contained (or containable) in big data, and the very factors that have enabled big data are enabling more traditional data collection. The Internet has opened the way for improving standard surveys, experiments, and health reporting _._ Instead of focusing on a ‘big data revolution,’ perhaps it is time we were focused on an ‘all data revolution,’ where we recognize that the critical change in the world has been innovative analytics, using data from all traditional and new sources, and providing a deeper, clearer understanding of our world.”).
> 43 Jo Ann S. Barefoot, _Disrupting Fintech Law_ , FINTECH L. REP., Mar,-Apr. 2015, at 6 (“Terms like ‘machine learning’, ‘deep learning,’ and ‘neural networks’ are entering the lexicon, referring to the fastgrowing ability of machines to gather and analyze vast volumes of information in ways, and at speeds, hitherto unimaginable. …Computers today can take on a question or task; search all the digitized information in the world; analyze what might be responsive to the goal; suggest possibilities; learn from feedback; and endlessly improve results. In the medical realm, for instance, machines can diagnose cancer with accuracy rivaling and surpassing that of human pathologists, and can ‘collaborate’ with humans to enhance the outcomes that either could reach alone. Importantly, machines have ‘noticed’ things they were not even asked to find, such as characteristics of tissue outside a malignancy, bringing such observations to the attention of doctors who then have pursued new pathways of study and medical diagnosis. Computers can conduct searches in seconds, hours, or days that would take humans hours, months, or even years to execute. They efficiently synthesize images, data, and words, in any language, from across the globe, and then present analysis and conclusions in conversational language and useful graphics.”); Daniel E. O’Leary, _Artificial Intelligence and Big Data_ , IEEE INTELLIGENT SYS., Mar.-Apr. 2013, at 96, 97 (“AI researchers have long been interested in building applications that analyze unstructured data, and in somehow categorizing or structuring that data so that the resulting information can be used directly to understand a process or to interface with other applications. As an example, Johan Bollen and Huina Mao [footnote omitted] found that stock market predictions of the Dow Jones Industrial average were improved by considering the overall ‘sentiment’ of the stock market— this is an unstructured concept, but based on structured data generated from Google.”); Maryam M. Najabadi et al., _Deep Learning Applications and Challenges in Big Data Analytics_ , 2 J. BIG DATA 11 (2015) (“[P]erforming discriminative tasks in Big Data Analytics one can use Deep Learning algorithms to extract complicated nonlinear features from the raw data, and then use simple linear models to perform discriminative tasks using the extracted features as input. This approach has two advantages: (1) extracting features with Deep Learning adds nonlinearity to the data analysis, associating the discriminative tasks closely to Artificial Intelligence, and (2) applying relatively simple linear analytical models on the extracted features is more computationally efficient, which is important for Big Data Analytics.”).
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
thought processes and rational human behavior through self-learning and storage of experiences.<sup>44</sup> Because it emulates human behavior, AI can act differently in the same situations, depending on the actions previously performed. Many sectors of the economy are already significantly affected by the rapid advances in AI during the past decade.<sup>45</sup> The exponential development of AI and the associated disruptive innovation pose substantial challenges for policy makers in education, financial markets, labor markets, and other areas.<sup>46</sup>
# **III. Disruptive Innovation as a Regulatory Challenge**
Regulatory concerns associated with disruptive technologies will likely increase in the coming years. Disruptive innovative technology often does not fit into existing legal categories created by recalcitrant regulatory structures.<sup>47</sup> Examples of existing regulatory concerns that may be exacerbated by future developments include the possible hacking of digital devices either to obtain information or to take control, the remote controlling of a driverless car, the hacking of pacemakers of influential elites, and the remote controlling of an individual’s computer to demand a ransom, among many others.<sup>48</sup> AI may not only
> 44 Robert D. Hof, _10 Breakthrough Technologies 2013: Deep Learning_ , MIT TECH. REV. (Apr. 23, 2013), http://www.technologyreview.com/featuredstory/513696/deep-learning/ (“With massive amounts of computational power, machines can now recognize objects and translate speech in real time. Artificial intelligence is finally getting smart.”); STUART RUSSELL & PETER NORVIG, ARTIFICIAL INTELLIGENCE: A MODERN APPROACH 5 (2013) (describing the different major definitions for AI); Paulius Cerka et al., _Liability for Damages Caused by Artificial Intelligence_ , 31 COMPUTER L. & SEC. REV. 376, 378 (2015) (“This belief began to change due to constant improvement of computer systems. Intellect, i.e. the ability to know, understand, and think, may not only be innate (natural), but also artificially created. AI is a broad area that includes such subfields as: (i) natural language systems; (ii) machine learning; (iii) simulation of senses; (iv) neural networks; (v) computer games; (vi) expert systems; and (vii) robotics.”).
> 45 Craig E. Karl, _The Three Breakthroughts That Have Finally Unleashed AI on the World_ , WIRED (Oct. 27, 2014, 6:30AM), http://www.wired.com/2014/10/future-of-artificial-intelligence/ (“Over the past five years, cheap computing, novel algorithms, and mountains of data have enabled new AI-based services that were previously the domain of sci-fi and academic white papers. [Examples include the:] Self-Driving Car…Body Tracker…Personal Photo Archivist…Universal Translator…Smarter News Feed.”).
> 46 ERIK BRYNJOLFSSON & ANDREW MCAFEE, THE SECOND MACHINE AGE: WORK, PROGRESS, AND PROSPERITY IN A TIME OF BRILLIANT TECHNOLOGIES 205-28 (2014) (discussing policy in the face of projected automation of low-skill labor); Tess Townsend, _Peter Diamandis: A.I. Will Lead to Massive Disruption Across Industries_ , INC. (Sept. 24, 2015), http://www.inc.com/tess-townsend/diamandisartificial-intelligence.html (“He said self-driving cars will render car insurance and the need for more roads obsolete. He anticipates his children will never drive. He projected that advances in camera technology will lead to cameras woven into clothes, biometric sensing will ‘massively disrupt’ medicine, and satellites will be able to watch raw materials entering factories and finished products leaving them, enabling smart AI to extract financial performance data ahead of the markets.”).
47 Examples include the question of how to categorize those who participate in the sharing economy into the standard employee-independent contractor framework, how crowdfunding works with the regulatory structure of private securities offerings, or how cryptocurrencies (such as BitCoin) fit with regulatory structure around currencies and bank reporting requirements. _See_ Felix Salmon, _When Disruption Meets Regulation_ , REUTERS (Jan. 30, 2014), http://blogs.reuters.com/felix-salmon/2014/01/30/when-disruptionmeets-regulation/; Balaji S. Srinivasan, _Regulation, Disruption, and the Technologies of 2013_ , http://goo.gl/sjZ3tF (providing an overview of the major conflicts between the traditional regulatory structures and innovative disruptors, with quotes from Elon Musk, Airbnb founders, other top entrepreneurs).
> 48 Mark Harris, _Researcher Hacks Self-Driving Car Sensors_ , IEEE SPECTRUM (Sept. 4, 2015, 19:00 GMT), http://goo.gl/rrlnP6; Aarti Shanani, _Ransomware: When Hackers Lock Your Files, To Pay or Not To Pay_ ,
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
fundamentally challenge traditional employment law structures,<sup>49</sup> AI and automation may eventually play important roles in causing large dislocations in labor markets.<sup>50</sup> In finance, it remains unclear how to treat crowdfunding technology that allows firms and individuals to raise money through bottom-up funding rather than traditional capital markets.<sup>51</sup> While acknowledging positive traits associated with crowdfunding technology, many commentators are concerned about the possibility of fraud, the breakdown of traditional protections for small investors, privacy violations, and intellectual property issues.<sup>52</sup>
AI and its disruptive capabilities present significant regulatory challenges in the existing regulatory framework. AI’s ability to accumulate and learn from experiences, to make individual decisions, and to act without supervision creates the possibility of causing damage to other technologies or humans.<sup>53</sup> Should AI unknowingly cause damage, any forms of compensation would have to be addressed through existing legal provisions in the existing regulatory framework. However, national and international law do not (currently) recognize AI as a subject of law. Without legal personality, AI cannot be personally liable for damages.<sup>54</sup> With autonomous AI playing an expanding role in
NPR ALLTECHCONSIDERED (Dec. 8, 2014, 6:27AM ET), http://goo.gl/8Ybj8o; Tarun Wadhwa, _Yes, You Can Hack a Pacemaker (and Other Medical Devices Too_ ), FORBES (Dec. 6, 2012, 8:31AM), http://goo.gl/oglPhS.
49 There are major issues of whether to classify Uber and Lyft drivers as independent contractors or employees. Uber’s innovative model essentially requires them to be non-employees, but many claim such classification harms existing firms that have to comply with California’s extensive labor law requirements. _Berwick v. Uber Tech., Inc_ ., No. 11-46739 EK, 2015 WL 4153765 (CA Dep’t of Labor Jun. 3, 2015) (deciding Uber workers were indeed employees, and Uber must comply with California labor law in this regard). See also Mike Isaac & Natasha Singer, _California Says Uber Driver Is Employee, Not a Contractor_ , N.Y. TIMES, Jun. 17, 2015, http://www.nytimes.com/2015/06/18/business/uber-contestscalifornia-labor-ruting-that-says-drivers-should-be-employees.html?_r=0. 50 John Komlos, _Has Creative Destruction Become More Destructive?_ (NBER Working Paper No. 20379), http://www.nber.org/papers/w20379 (suggesting that the creative destruction aspect of capitalism is being amplified, especially in labor markets, by the exponential pace of technology).
> 51 Thomas L. Hazen, _Crowdfunding or Fraudfunding? Social Networks and the Securities Laws--Why the Specially Tailored Exemption Must Be Conditioned on Meaningful Disclosure_ , 90 N.C. L. REV. 1735, 1736 (2012) (“Social networks have been used as a medium for financing films and other forms of art, as well as for charitable solicitations. These and similar fundraising endeavors are known as crowdfunding. Crowdfunding is the fundraising analog to crowdsourcing, which refers to mass collaboration efforts through large numbers of people, generally using social media or the Internet.”).
> 52 _See generally id_ .; Letter from William F. Galvin, Secretary of the Commonwealth of Massachusetts, to Elizabeth M. Murphy, Secretary, U.S. Securities and Exchange Commission 1-2 (Aug. 8, 2012), http://www.sec.gov/comments/jobs-title-iii/jobstitleiii-121.pdf (“Crowdfunding represents a significant departure from long-established rules for public offerings of securities. . . . Crowdfunding is designed to permit the smallest investors to participate in securities offerings by early-stage companies. Our experience is that these investors are especially vulnerable in the small-company segment of the market.”).
> 53 Cerka et al., _supra_ note 53, at 376 (“The ability to accumulate experience and learn from it, as well as the ability to act independently and make individual decisions, creates preconditions for damage. Factors leading to the occurrence of damage identified in the article confirm that the operation of AI is based on the pursuit of goals. This means that with its actions AI may cause damage for one reason or another; and thus issues of compensation will have to be addressed in accordance with the existing legal provisions.”).
> 54 _Id_ . (“In the absence of direct legal regulation of AI, we can apply article 12 of United Nations Convention on the Use of Electronic Communications in International Contracts, which states that a person (whether a natural person or a legal entity) on whose behalf a computer was programmed should ultimately be responsible for any message generated by the machine. Such an interpretation complies with a general
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
society, an increasing number of scientists and entrepreneurs suggest that government regulation may be necessary to reduce the risks to the public associated with the rapid advances in AI.<sup>55</sup> While some favor an indirect form of AI regulation based on differential tort liability,<sup>56</sup> others oppose rigid regulation of AI because regulation may inhibit the socially beneficial innovations associated with AI, or they suggest such regulation is impossible in the face of such a powerful and exponentially growing technology.<sup>57</sup> AI is currently not able to match human-level reasoning,<sup>58</sup> but it can still have an impact on the service industries, among others.<sup>59</sup>
A regulatory response to disruptive innovation requires policy makers to balance legitimate policy objectives such as consumer protection, privacy, and public welfare while avoiding regulatory capture and policies that predominantly protect incumbent firms. In the startup phase and before gaining substantial market share, disruptive firms often create products or provide services that either do not fit into the existing regulatory framework and the protections provided thereby, or such disruptive products and services do not or do not fully comply with existing regulatory protections. In the latter case, disruptive firms often argue that because of the unprecedented nature of the disruptive product or service the existing rules do not or should not apply or that compliance with
rule that the principal of a tool is responsible for the results obtained by the use of that tool since the tool has no independent volition of its own. So the concept of AI-as-Tool arises in the context of AI liability issues, which means that in some cases vicarious and strict liability is applicable for AI actions.”). 55 Sharon Gaudin, _A.I. Researchers Say Elon Musk’s Fears ‘Not Completely Crazy_ ’, COMPUTERWORLD (Oct. 29, 2014, 1:16PM PT), http://goo.gl/3K91LD (“I think we should be very careful about artificial intelligence," Musk said when answering a question about the state of AI. "If I were to guess at what our biggest existential threat is, it's probably that …With artificial intelligence, we are summoning the demon. In all those stories with the guy with the pentagram and the holy water, and he's sure he can control the demon. It doesn't work out.”); Rory Cellan-Jones, _Stephen Hawking Warns Artificial Intelligence Could End Mankind_ , BBC NEWS (Dec. 2, 2014), http://www.bbc.com/news/technology-30290540 (“He told the BBC: ‘The development of full artificial intelligence could spell the end of the human race.’"). 56 Matthew U. Scherer, _Regulating Artificial Intelligence Systems: Risks, Challenges, Competencies, and Strategies_ , 29 HARV. J. L. & TECH. 353 (2016).
> 57 _You, Robot_ ?, ECONOMIST (Sept. 1, 2012), http://www.economist.com/node/21560986 (“Regulators must tread carefully. Noel Sharkey, a computer scientist at the University of Sheffield, observes that overly rigid regulations might stifle innovation. But a lack of legal clarity leaves device-makers, doctors, patients and insurers in the dark. The RoboLaw researchers hope to square this circle when they deliver their findings in 2014. So far, though, they seem to have more questions than answers.”); John Danaher, _Is Regulation of Artificial Intelligence Possible_ ?, H+ MAG. (Jul. 15, 2015), http://goo.gl/vMw8et (“Despite these worries, debates about the proper role of government regulation of AI have generally been lacking. There are a number of explanations for this: law is nearly always playing catch-up when it comes to technological advances; there is a decidedly anti-government libertarian bent to some of the leading thinkers and developers of AI; and the technology itself would seem to elude traditional regulatory structures.”).
> 58 Richard Waters, _Artificial Intelligence: Machine v. Man_ , FIN. TIMES (Oct. 31, 2015), http://goo.gl/8gaWQn (“Their effort is prompted by a fear of what will happen when computers match humans in intelligence. At that point, humans would cede leadership in technological development, since the machines would be capable of improving their own designs by themselves. And with the accelerating pace of technological change, it wouldn’t be long before the capabilities – and goals – of the computers would far surpass human understanding.”).
> 59 Harry Surden, _Machine Learning and Law_ , 89 WASH. L. REV. 87, 88 (2014) (“In the last few decades, researchers have successfully used machine learning to automate a variety of sophisticated tasks that were previously presumed to require human cognition. These applications range from autonomous (i.e., selfdriving) cars, to automated language translation, prediction, speech recognition, and computer vision. Researchers have also begun to apply these techniques in the context of law.”).
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
the existing rules is not necessary to protect the public because more appropriate consumer protections are available through the disruptive firm.<sup>60</sup> If disruptive firms do not comply with existing rules or create their own exemptions because the existing regulatory framework does not apply and appropriate rules are not available, public policy goals can be undermined and incumbent firms that continue to be subject to existing rules encounter often severe competitive disadvantages.<sup>61</sup> To counteract such disadvantages, incumbent firms that are exposed to a perceived threat of disruptive innovation from disruptive firms may attempt to use the existing regulatory process to create obstacles for disruptive firms to compete.
The anticipated level of innovation in the next thirty years<sup>62</sup> makes it increasingly less likely that rulemakers will be able to effectively protect the public while harnessing innovation for the benefit of society. The combination of harm and possible benefits that are associated with disruptive innovation can either equally affect both consumers and providers or it can benefit a subset of consumers and providers while another subset suffers adverse consequences; it can also create tradeoffs for subsets of providers and consumers. As accelerating disruptive technology creates new complexity for society at a rapidly accelerating pace, especially in significant growth areas such as robotics, nanotechnology, and biotechnology,<sup>63</sup> society increasingly struggles with analyzing the wealth of new, beneficial, and ever more detailed information received from disruptive technologies.<sup>64</sup> The negative and positive impact of disruptive innovation will trigger political and policy pressures for and against various regulatory responses to change. The political system is, however, less likely to be able to deal with the challenges of disruptive innovation because the increasing complexity of innovation-driven regulatory issues causes political and policy makers’ confusion over rapidly-emerging disruptive change, making a coherent political and policy solution to these challenges less likely.
> 60 Benjamin G. Edelman & Damien Geradin, _Efficiencies and Regulatory Shortcuts: How Should We Regulate Companies Like Airbnb and Uber?_ 19 Stan. Tech. L. Rev. 293, 295-96 (2016), http://journals.law.stanford.edu/stanford-technology-law-review/print/volume-19/issue-2/efficiencies-andregulatory-shortcuts-how-should-we-regulate-companies (“[W]e observe that many platform operators advertise their services as ‘sharing.’ For example, short-term property rental service Airbnb says its service lets hosts ‘share their homes with guests’ while transportation service Lyft says it offers ‘ridesharing.’ The term ‘sharing’ partially captures some aspects of these companies’ activities, e.g., employing a single resource for multiple purposes, such as using a vehicle both for an owner’s personal needs and to transport paying passengers. Moreover, the key efficiencies generally do not come from ‘sharing’ but from the market structure that platforms facilitate, including casual service providers who avoid the fixed cost and, often, regulation associated with traditional service.”); Andy Kessler, _The Weekend Interview with Brian Chesky: The ‘Sharing Economy’ and Its Enemies_ , WALL ST. J., Jan. 18-19, 2014, at A11. [Uber cofounder Brain Chesky said of the current regulatory structure]: “There were laws created for businesses, and there were laws for people. What the sharing economy did was create a third category: people as businesses,” to which the application of existing laws is often unclear.).
> 61 See Edelman & Geradin, _supra_ note 60, at 23 (“Most regulatory regimes require full service to disfavored groups, including racial minorities, low-income users, and low-income regions. Software platforms tend to circumvent these requirements, either through decentralized decision-making that favors individual preferences over government mandates, or through software implementations that otherwise do not require compliance.”).
> 62 _See supra_ notes 30-55 and accompanying text.
> 63 _See infra_ Table 4.
> 64 John O. McGinnis, _Accelerating Ai_ , 104 NW. U. L. REV. 1253, 1269 (2010) (“The problem now is that the information available to be processed may be swelling beyond human capacity to achieve sound social decisionmaking without the aid of AI.”).
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
The existing regulatory infrastructure cannot sufficiently distinguish and harness beneficial innovation. Rulemakers’ inability to address regulatory issues associated with disruptive innovation will likely generate high levels of legal uncertainty and inconsistency<sup>65</sup> that inhibit innovation during technological transition periods; and technological transition is likely going to be a permanent state in the age of disruptive innovation, which exacerbates the uncertainty and inconsistency created by rulemakers’ inability to react timely and adequately to disruptive innovation. In the existing regulatory framework, regulators and commenters are engaged in a long, drawn-out feedback process that involves hearings, proposed rules, the submission of comment letters, and finally agency lawyers finalizing a rule after considering the comments. The current regulatory framework lacks a mechanism that succinctly and anticipatorily informs rulemakers of beneficial innovative ideas. In fact, the existing rulemaking process prohibits ex parte communications and insufficiently integrates brainstorming and ideas across industries, and therefore may actually undermine innovation.
Policy makers and scholars are debating possible policy responses to address the regulatory challenges associated with disruptive innovation. Some suggest that the regulatory challenges associated with disruptive innovation can only be effectively addressed through principles-based regulation because the existing rulemaking procedures lack sufficient speed and flexibility.<sup>66</sup> Others see disruptive innovation predominantly as a competition law (unfair competition) and antitrust issue.<sup>67</sup> Consider the online transportation network company, Uber, and the online marketplace for vacation homes, Airbnb. Early intervention,<sup>68</sup> rulemaking petitions, and retrospective review may be more appropriate remedies for the effects of disruptive innovation that
> 65 Barefoot, _supra_ note 52, at 18 (“[The United States has] five federal agencies that directly examine and supervise financial institutions (Comptroller of the Currency, FDIC, Federal Reserve Board, CFPB, and National Credit Union Administration (NCUA)). At least 20 federal agencies regulate some aspect of financial products, from mortgages and student loans to retirement funds, and/or have enforcement mandates that include financial services, as with the Department of Justice (DOJ) and the FTC. Much of this complexity is multiplied by fifty at the state level, plus insurance products are almost entirely regulated by the states. . . . All or most of these regulatory bodies have at least fragmentary responsibility relating to the broad, overarching consumer protection themes discussed above, such as privacy, data security, UDAP, fair lending, and the like. As these and other regulatory standards shift and escalate under pressure from fast-changing technology, it seems inevitable that the complexity of this structure will exacerbate uncertainty and inconsistency in regulatory expectations, and will militate against modernization that might address problems efficiently and serve consumers well.”).
> 66 Barefoot, _infra_ note 52, at 9.
> 67 Inge Graef et al., _How Google and Others Upset Competition Analysis: Disruptive Innovation and European Competition Law_ (Conference Paper, 25<sup>th</sup> European Regional Conference of the International Telecommunications Society, Brussels, Belgium, Jun. 22-25, 2014), http://econstor.eu/bitstream/10419/101378/1/795226780.pdf.
> 68 Nathan Cortez, _Regulating Disruptive Innovation_ , 29 BERKELEY TECH. L.J. 175, 201-04 (2014) (“With novel technologies, agencies also risk paralysis by analysis. New technologies often present unforeseen risks if under-regulated and dramatic opportunity costs if overregulated. The inclination of most regulators is to avoid both extremes. . . . Regulators also tend to overvalue gathering information about new technologies, which further distorts decisions on regulatory timing. . . . [E]arly interventions can benefit both regulated industry and regulatory beneficiaries. . . . [E]arly interventions may also benefit from a more objective regulatory atmosphere, before parties become entrenched and adversarial.”).
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
also can help address the regulatory sine curve.<sup>69</sup> Assuming that regulatory innovation is subject to an interdependent ecological system, others propose networks for the diffusion of improved regulatory technology.<sup>70</sup> Recognizing that ex ante regulation has diminished in significance in the context of regulating disruptive innovation and artificial intelligence (AI), others see research, development, and public operation as the only ways to identify harmful innovation.<sup>71</sup> Chris Brummer suggests a form of regulatory experimentalism including selectively adopting trial periods for new regulatory approaches.<sup>72</sup> This form of micro-level regulatory experimentation could later be expanded to “system-wide enterprise zones for financial market compliance and regulatory adaptation.”<sup>73</sup>
An emerging literature explores the use of informal regulation.<sup>74</sup> Wu suggests that traditional rulemaking is regularly impracticable in rapidly developing industries where highly informal methods can be justified.<sup>75</sup> Rather than relying on traditional regulatory means in the existing regulatory framework, Wu suggests that regulatory agencies confronting disruptive innovations should utilize “threats” for the disruptive industries, packaged in guidance documents, warning letters etc., because such threats can help avoid regulation that is inappropriately calibrated or premature.<sup>76</sup> Threats are less
> 69 Reeve T. Bull, _Building a Framework for Governance: Retrospective Review and Rulemaking Petitions_ , 67 ADMIN. L. REV. 265, 293-306 (2015) (urging regulatory agencies to expand the use of petitions for rulemaking and retrospective review to achieve smarter regulation).
> 70 Jonathan B. Wiener, _The Regulation of Technology, and the Technology of Regulation_ , 26 TECH. IN SOC’Y 483, 496 (2004) (“Fourth, we should foster networks for the diffusion of regulatory innovations. Just as private technology diffusion is crucial to economic prosperity, so the diffusion of improved public regulatory technology is crucial to social progress.”).
> 71 Scherer, _infra_ note 65, at 357.
> 72 Brummer, _supra_ note 5, at 1044 (“So for example, the ban on general solicitation could be lifted for five years, with a sunset provision triggered absent some additional regulatory blessing. Or more onerous investigatory requirements on portals could be lifted where other (statistical or algorithmic) technologies are used to test or limit exposures of retail investors in crowdfunded securities, but likewise set to expire absent some regulatory finding that they insufficiently track or identify breaches in prudential requirements for investors.”).
> 73 _Id_ . at 1047 (“So, for example, avenues of special information sharing and expedited assistance could be opened with targeted industry members for the fulfillment of desired regulatory objectives. Or investment tax credits could be provided for market participants that undertake strategies to bolster investor protection or market stability.”).
> 74 _See e.g_ ., Tim Wu, _Agency Threats_ , 60 DUKE L.J. 1841 (2011).
> 75 _Id_ . at 1841 (“There are three main ways in which agencies regulate: rulemaking; adjudication; and informal tools of guidance, also called nonlegislative or interpretative rules. Over the last two decades, agencies have increasingly favored the use of the last of these three, which can include statements of best practices, interpretative guides, private warning letters, and press releases.”).
> 76 _Id_ . at 1849 (“Of the three options, the first--making law--may be the worst alternative. What sounds attractive is the prospect of an orderly, planned approach to the future. The problem is that, with so little known about the industry, issuing specific rules based on guesses about the future runs a grave risk of creating a bad law, or at least a law that is much worse than one issued after more development. Such lawmaking suffers from all of the defects that Friedrich Hayek identified with central planning--impressive in a world of perfect information, but terrible in this world.”); _Contra_ Cortez, _supra_ note **Error! Bookmark not defined.** , at 175 (arguing that “a flexible initial posture based primarily on ‘threats’ can calcify, creating weak defaults that lead to suboptimal regulation in the long term. . . .If agencies are concerned about regulating prematurely or in error, then they can experiment with timing rules, alternative enforcement mechanisms, and other variations on traditional interventions. If agencies do choose to proceed by making threats, then they should use them as a short-term precursor to more decisive, legally binding action, as the FCC did, and avoid relying on them as a long-term crutch, as the FDA did.”); Jerry
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
burdensome than traditional regulatory approaches, more flexible, can be better finetuned, and signal regulatory intent appropriately.<sup>77</sup>
# _1. Ex Post Facts-Based (Trial-and-Error) Rulemaking_
The ex-post facts-based, trial-and-error-rulemaking in the existing regulatory framework<sup>78</sup> often produces suboptimal regulatory outcomes that may no longer be sustainable as disruptive innovation accelerates and causes unprecedented societal change. The ex-post facts-based approach to rulemaking has functioned in the existing regulatory framework because optimal requirements for rules do become clear when stable and presumptively optimal rules emerge as suboptimal. A prerequisite for rulemaking is thus the availability of the information on optimized rule requirements. In an age of exponential innovation, that information may not materialize at all or may not materialize in time for traditional rulemaking to be effective. In other words, because facts-based rulemaking did not anticipate regulatory issues created by innovation, rulemakers may not at all or much too late realize what new regulatory demands apply to a given innovation and associated regulatory issue.
# _2. Stable and Presumptively Optimal Rules_
The existing regulatory infrastructure, including Congress, regulatory agencies, selfregulatory bodies, and the literature on regulation, relies almost exclusively on stable and presumptively optimal rules.<sup>79</sup> Such stable and presumptively optimal rules are created to address perceived regulatory issues identified by lawmakers through centralized information in the then-existing economic and market conditions. Stable and presumptively optimal rules are tailored and drafted to attain permanent solutions for perceived regulatory issues that can be addressed with sufficient political support at the time of drafting.<sup>80</sup> The underlying conditions for rulemaking and the corresponding requirements for optimal and stable rules are, however, subject to constant and everincreasing amounts of change.<sup>81</sup> The exponential growth of technology, the associated exponential growth of innovation, and the resulting growth in the complexity of
> Brito _, "Agency Threats" and the Rule of Law: An Offer You Can't Refuse_ , 37 HARV. J.L. & PUB. POL'Y 553, 554 (2014) (“I conclude that not only is Wu's thesis wrong, but it is also dangerous. Part I of this article reviews Wu's essay, defining what constitutes an agency threat and when it is justified according to Wu. Part II critiques Wu's thesis by showing, among other things, that it is based on a false dilemma; that it assumes an unwarranted level of knowledge on the part of regulators; that it assumes--contrary to evidence-that regulators are good proxies for the public interest; and that it ignores the costs of eschewing the regulatory process.”) _._
> 77 Wu, _supra_ note **Error! Bookmark not defined.** , at 1843 (“Both agency and industry will sometimes share an interest in an informal and flexible regime that resembles an unenforceable ‘letter of intent’ in the world of private contracts.”)
> 78 Popper, _supra_ note 7 _;_ Kirchner, _supra_ note 10.
> 79 See Kaal in FS Kirchner, _supra_ note 6, at 1212 (“[T]he institutional infrastructure for rulemaking was geared towards the creation of rules for governing a relatively stable society with less upward mobility and relatively stable economic and market environments.”); Kaal, Wake Forest, _supra_ note 6; Kaal in Liber Amicorum Peter Nobel, _supra_ note 6; Kaal & Lacine 2014, _supra_ note 6.
> 80 Kaal in FS Kirchner, _supra_ note 6, at 1218.
> 81 _Id._
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
regulatory issues will likely require ever-increasing numbers of governance adjustments via stable and presumptively optimal rules,<sup>82</sup> making it increasingly less likely that rulemakers will be able to effectively protect the public via stable and presumptively optimal rules. If rulemakers cannot adequately protect their constituents, regulatory supplements facilitating anticipatory rulemaking may be justified.
The existing regulatory infrastructure with stable rules and stable rulemaking processes is subject to significant systemic constraints and path dependencies. Historically, a comparatively small, stable, and less interconnected society with limited upward mobility and relatively stable economic and market environments for rulemaking created an institutional infrastructure for rulemaking that produces stable and presumptively optimal rules to address a limited set of readily identifiable regulatory issues.<sup>83</sup> Because of the limited scope of regulatory issues and the need for consensusdriven institutional designs, the historically evolved infrastructure for rulemaking could cope with its inherent collective action problems,<sup>84</sup> regulatory cycles,<sup>85</sup> and trial-and-error rulemaking,<sup>86</sup> without major disruptions. The exponential growth of technology and its associated disruptive innovation<sup>87</sup> are likely to expose the depth of issues related to the path dependencies in the existing regulatory infrastructure and the associated design flaws of the institutional infrastructure.
# _3. Timing_
The timing of regulation in an environment of disruptive innovation should be a primary concern for policy makers. If policy makers regulate too early, they risk inhibiting innovation; if they withhold regulation too long, they may harm consumers and markets if regulatory inertia sets in around the disruptive product or service. The time frame for rulemaking in the existing regulatory infrastructure may be inadequate to address regulatory challenges associated with disruptive innovation. In the existing regulatory framework, regulators and commenters are engaged in a long, drawn-out feedback process that involves hearings, proposed rules, the submission of comment
> 82 MAEVE P. CAREY, COUNTING REGULATIONS: AN OVERVIEW OF RULEMAKING, TYPES OF FEDERAL
> REGULATIONS, AND PAGES IN THE FEDERAL REGISTER 17 (Congressional Research Service Report R43056, 2015), https://fas.org/sgp/crs/misc/R43056.pdf (“As the data show, the number of pages has increased since publication of the _Federal Register_ began. The number of pages reached a peak in 1980 at 87,012 pages; decreased to 47,418 pages in 1986; then increased again and has been approximately between 65,000 and 85,000 pages for the past two decades.”).
> 83 Kaal in FS Kirchner, _supra_ note 6, at 1212 (“[T]he institutional infrastructure for rulemaking was geared towards the creation of rules for governing a relatively stable society with less upward mobility and relatively stable economic and market environments.”).
> 84 Coffee, _supra_ note 8, at 1019-82; Kaal in Liber Amicorum Peter Nobel, _supra_ note 6; Kaal, Wake Forest, _supra_ note 6; Kaal in FS Kirchner, _supra_ note 6; Ostrom, _supra_ note 8 (The predominance of special interest groups in the rulemaking process can be temporarily overcome during and after crises when political entrepreneurs assume the transaction costs of organizing the otherwise disinterested latent groups); Olson, _supra_ note 8 (Smaller and better-organized special interest groups typically control dispersed investors and other dominant latent groups in the competition to influence the rulemaking process).
> 85 Coffee, _supra_ note 8; Kaal in Liber Amicorum Peter Nobel, _supra_ note 6; Kaal in FS Kirchner, _supra_ note 6.
> 86 Kaal in Liber Amicorum Peter Nobel, _supra_ note 6; Kaal in FS Kirchner, _supra_ note 6; Kirchner, _supra_ note 10, at 161-172; Popper, _supra_ note 7.
> 87 _See supra_ notes 21-46 and accompanying text.
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
letters, and finally agency lawyers finalizing a rule after considering the comments<sup>88</sup> . This existing rulemaking process often prohibits ex parte communications,<sup>89</sup> involves very little brainstorming, and undermines innovation.<sup>90</sup> While rulemakers may be able to update regulations and regulatory guidance to address regulatory issues created by disruptive innovation, given the exponential nature of disruptive innovation,<sup>91</sup> they are less able to _adequately_ update existing rules in the present regulatory framework. The speed of product innovation alone makes formal rulemaking in the existing regulatory infrastructure challenging. Formal rulemaking is simply too time-consuming.<sup>92</sup> The speed of product innovation makes it possible to bring a new product to market while formal rulemaking in the existing regulatory infrastructure, taking months and often years of regulatory procedure, is still dealing with the last product launch. New regulations pertaining to an innovative product could be obsolete before they are finalized.<sup>93</sup>
# _4. Unknown Future Contingencies_
The existing regulatory infrastructure with stable and presumptively optimal rules is
> 88 Lynn E. Blais & Wendy E. Wagner, _Emerging Science, Adaptive Regulation, and the Problem of Rulemaking Ruts_ , 86 TEX. L. REV. 1701, 1706 (2008) (“[A]gencies systematically engage in excess data gathering, protracted analysis of the data and associated public comments, and extraordinarily detailed explanation of the bases and purposes of their final rules in an attempt to insulate their policies from judicial reversal. . . . This hyperformalization of notice-and-comment rulemaking has several consequences. The most obvious is the increase in time and resources required to promulgate each rule.”).
> 89 5 U.S.C. § 557(d)(1) (2012); _Ex Parte Communications in Informal Rulemaking Proceedings (Recommendation 77-3_ ), 1 C.F.R. § 305.77 (2014) (listing other advantages associated with restraining ex parte communications, including reducing possibility of unfair influence over decision makers and affording interested parties opportunity to respond to information relied upon in the decisionmaking process).
> 90 Peter H. Schuck, _When the Exception Becomes the Rule: Regulatory Equity and the Formulation of Energy Policy Through an Exceptions Process_ , 1984 DUKE L.J. 163, 197 (1984) (“Hastily prepared, overbroad rules can be disastrous for those who technically are covered by the rules but to whom the rules should not, in justice or sound policy, be applied. Inflexible application of such rules may quickly create grave competitive distortions, perhaps even driving firms out of business before the rules can be refined or eliminated.”); Blais & Wagner, _supra_ note 88, at 1705 (“Important statutory directives remain unimplemented years after the deadline for implementation has passed. Agencies are increasingly turning to even more informal methods--which lack adequate opportunities for public participation and evade meaningful judicial oversight--to promulgate important policies. And, not surprisingly, agencies are increasingly reluctant to revisit rules after enactment, even if the factual or policy predicates underlying them have changed.”).
> 91 _See supra_ notes 21-46 and accompanying text.
> 92 Cass R. Sunstein, _Is the Clean Air Act Unconstitutional_ ?, 98 MICH. L. REV. 303, 371 (1999) (“With respect to systematic effects: A great deal of attention has been paid to the phenomenon of the ‘ossification’ of notice-and-comment rulemaking, and indeed a high priority, for the future of administrative law, is to devise means to overcome the problem. Originally intended as a quick and effective alternative to formal, on-the-record rulemaking, executive and especially judicial innovations have converted notice-andcomment rulemaking into an exceptionally time-consuming affair, often consuming many years, frequently half a decade and more. In fact EPA estimates that informal rulemaking typically takes five years.”); Thomas O. McGarity, _Some Thoughts on "Deossifying" the Rulemaking Process_ , 41 DUKE L.J. 1385, 1385 (1992) (“Although informal rulemaking is still an exceedingly effective tool for eliciting public participation in administrative policymaking, it has not evolved into the flexible and efficient process that its early supporters originally envisioned. During the last fifteen years the rulemaking process has become increasingly rigid and burdensome.”).
> 93 Barefoot, _supra_ note 43, at 10.
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
largely incapable of addressing the ever-increasing unknown future contingencies associated with disruptive innovation. Because it lacks anticipatory rulemaking capabilities, the existing regulatory system addresses regulatory issues only ex-post, if and when they materialize and if core constituents are sufficiently burdened to precipitate enough political pressure for lawmakers to act.<sup>94</sup> Evidence exists that the suboptimal expost timing of rulemaking in the existing regulatory infrastructure regularly requires expedited rulemaking that results in suboptimal regulatory outcomes.<sup>95</sup> Rulemakers often discount or willingly accept the suboptimalities of rules associated with unknown future contingencies. Temporary acceptance of suboptimal rules increases temporary rule certainty and predictability, but it also precipitates the inevitable need for rule revision, amendments, and repeals<sup>96</sup> In the existing regulatory framework, if and when the inevitable future contingencies materialize, unveiling suboptimal sets of existing rules, rulemakers perpetuate a costly and uncertainty-increasing process of rule revisions, amendments, and repeals to address such shortcomings. Despite its insufficient anticipatory capabilities and all the downsides that are associated with it, stable and presumptively optimal rules remain the uniform response to perceived regulatory issues.<sup>97</sup>
Given the pace of innovation,<sup>98</sup> future contingencies in rulemaking are likely to substantially increase, making the dynamic anticipation of future contingencies more important for rulemaking. The current process of rule revisions, amendments, and repeals to address inevitable shortcomings of stable rules is costly, time-consuming, and cannot, in our estimation, keep track of future innovations and corresponding regulatory needs. Exponential innovation has the potential to overwhelm the existing regulatory process because rulemakers are still attempting to comprehend the regulatory demands of the last wave of innovative solutions, technologies, and applications while the next wave of innovation is already in full force.<sup>99</sup> In effect, exponential innovation is likely to intensify
> 94 _Id._ ; _see also supra_ note 3.
> 95 Kaal Wake Forest, _supra_ note 6; Coffee, _supra_ note 8; Roberta Romano, _Regulating in the Dark_ 1 (Yale Law & Econ. Research Paper No. 442, 2012), http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1974148 (“Foundational financial legislation is typically adopted in the midst or aftermath of financial crises, when an informed understanding of the causes of the crisis is not yet available. Moreover, financial institutions operate in a dynamic environment of considerable uncertainty, such that legislation enacted even under the best of circumstances can have perverse unintended consequences, and regulatory requirements correct for an initial set of conditions can become inappropriate as economic and technological circumstances change. Furthermore, the stickiness of the status quo in the U.S. political system renders it difficult to revise legislation, even though there may be a consensus to do so.”); STEPHEN M. BAINBRIDGE, CORPORATE GOVERNANCE AFTER THE FINANCIAL CRISIS 269 (2012) (“Just as a ratchet wrench works only in one direction, the size and scope of government tends to move in only one direction—upwards—because the interest groups that favored the changes now have an incentive to preserve the new status quo, as do the bureaucrats who gained new powers and prestige. Hence, each crisis has the effect of ratcheting up the long-term size and scope of government.”).
> 96 Kaal in FS Kirchner, _supra_ note 6, at 1217 (describing “ _ex-ante_ experimentation” in rulemaking).
> 97 _See_ Kaal in Liber Amicorum Peter Nobel, _supra_ note 6; Kaal, Wake Forest, _supra_ note 6; Kaal and Lacine 2014, _supra_ note 6; Kaal in FS Kirchner, _supra_ note 6.
> 98 _See supra_ notes 30-55 and accompanying text.
> 99 Romano, _supra_ note 95, at 1 (“Moreover, financial institutions operate in a dynamic environment of considerable uncertainty, such that legislation enacted even under the best of circumstances can have perverse unintended consequences, and regulatory requirements correct for an initial set of conditions can become inappropriate as economic and technological circumstances change”); Shrupti Shah, Rachel Brody
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
the frequencies of the regulatory sine curve that are recognized by the literature.<sup>100</sup>
# **IV. Dynamic Regulation as a Regulatory Supplement**
Economists have used the concept of dynamic regulation in the context of innovation and learning by doing,<sup>101</sup> regulation of quality,<sup>102</sup> principal-agent and adverse selection problems,<sup>103</sup> and continuing regulatory relationships.<sup>104</sup> Dynamic elements in regulation have previously also been described as “dynamic games,”<sup>105</sup> “regulatory dialectic,” and as a “dynamic adjustment process.”<sup>106</sup> The rapid pace of technological developments in the telecommunications industry necessitated the use of dynamic elements in regulation.<sup>107</sup>
The theory of dynamic regulation conceptualizes the study of regulatory phenomena in relation to preceding and succeeding events, using institution-specific and decentralized information to facilitate feedback effects for anticipatory rulemaking.<sup>108</sup> The dynamic regulatory framework allows rulemaking to overcome the historic constraints and path dependencies that traditionally focused the rulemaking process on stable and presumptively optimal rules. Evolving from a historically reactive process based only on preceding events and driven by the agency and collective-action problems of rulemaking, rulemaking in a dynamic framework allows for the anticipation of future innovation trends and associated contingencies.
Shortcomings in the existing rulemaking process<sup>109</sup> can be addressed with dynamic and anticipatory elements in rulemaking.<sup>110</sup> The collective action problem of rulemaking,
& Nick Olson _, The Regulator of Tomorrow: Rulemaking and Enforcement in an Era of Exponential Change_ (2015), http://dupress.com/articles/us-regulatory-agencies-and-technology/ (“Regulators are on the front lines of nearly every controversy resulting from a new technology—either being told to get out of the way or being called upon for solutions when incidents arise. Data breaches, 3D-printed guns, clashes between taxis and ridesharing services, and many other events that make news headlines also pose real challenges to regulators. In a fast-moving and increasingly complex world, regulators are finding it harder and harder to balance the need to protect citizens and fair markets with the need to avoid impeding innovation.”).
> 100 Kaal Wake Forest, _supra_ note 6; Coffee, _supra_ note 8.
> 101 _See_ Tracy R. Lewis & Huseyin Yildirim, _Learning by Doing and Dynamic Regulation_ , 33 RAND J. ECON. 22, 23–24 (2002) (“Innovation is more rapid when current service is increased, enabling the supplier to accelerate his rate of learning.”).
> 102 Stephane Auray et al., _Dynamic Regulation of Quality_ , 42 RAND J. ECON. 246, 246 (2011).
> 103 JOHN M. LITWACK, DYNAMIC REGULATION, DEMAND INFORMATION AND MARKET PRICES (1992).
> 104 David P. Baron & David Besanko _, Commitment and Fairness in a Dynamic Regulatory Relationship,_ 54 REV. ECON. STUD. 413, 413 (1987); David P. Baron & David Besanko, _Regulation and Information in a Continuing Relationship,_ 1 INFO. ECON. & POL’Y 267, 267 (1984).
> 105 Edward J. Kane, _Extracting Nontransparent Safety Net Subsidies by Strategically Expanding and Contracting a Financial Institution’s Accounting Balance Sheet_ , 36 J. FIN. SERV. RES. 161, 161 (2009).
> 106 Edward J. Kane, _Interaction of Financial and Regulatory Innovation,_ 78 AM. ECON. REV. 328, 333 (1988) (“The regulatory dialectic posits a dynamic adjustment process that in the long run enforces a ‘law of one regulatory burden.’" Precisely because inefficient patterns of regulation impose excessively burdensome costs either on regulatees, their customers, or the general taxpayer, the burdened parties must be expected sooner or later to develop avoidance strategies by which to throw these burdens aside.”).
> 107 _See supra_ note 16.
> 108 _See_ Kaal in FS Kirchner, _supra_ note 6; Kaal in Liber Amicorum Peter Nobel, _supra_ note 6; Kaal, Wake Forest, _supra_ note 6; Kaal and Lacine 2014, _supra_ note 6.
> 109 _See supra_ notes 49-55.
> 110 Kaal Wake Forest, _supra_ 6; Kaal in FS Kirchner, _supra_ note 6.
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
problems associated with trial-and-error rulemaking, and problems associated with regulatory cycles derive largely from the nature of stable and presumptively optimal rules.<sup>111</sup> Rulemaking with dynamic elements increases the use of institution-specific, decentralized, and timely information<sup>112</sup> to allow for adaptive rulemaking. Adaptive rulemaking helps overcome the collective action problem of rulemaking because with fewer stable rules latent majority groups and dominant minority groups have fewer opportunities to influence the continuously and timely adapting rulemaking process. Similarly, regulatory cycles and trial-and-error rulemaking become less prevalent because adaptive rulemaking processes supplement stable rules with adaptive capabilities that make rule revisions less prevalent and minimize trial-and error rulemaking.
Feedback effects are an integral part of dynamic forms of regulation.<sup>113</sup> Feedback effects are described in the theory of lawmaking as rulemakers’ reactions to institutional change<sup>114</sup> or private actors’ reactions and counteractivities to institutional constraints.<sup>115</sup> A dynamic regulatory framework facilitates feedback effects that help enhance the availability of institution-specific and decentralized information to support the rulemaking process.<sup>116</sup> Feedback effects can occur in several settings including intra- and interjurisdictional feedback processes, feedback processes between outcomes and institutions, feedback processes between different public and private rulemakers, and feedback processes between rules and rulemaking processes.<sup>117</sup> Competition between legislators and regulators from different jurisdictions can facilitate interjurisdictional
111 The collective action problem of rulemaking, inherent in smaller and better-organized interest groups and dominant latent groups competing to influence the rulemaking process, pervades when regulatory outcomes are stable. Regulatory cycles, inherent in existing rule adoption, revisions, and revocations, derive from stable and presumptively optimal rules that cannot timely adapt to future contingencies and corresponding regulatory needs. Trial-and-error-rulemaking is necessitated if rules are presumptively optimal and stable. Trial-and-error-rulemaking could become obsolete if dynamic elements in rulemaking processes systematically anticipated future contingencies and corresponding regulatory needs.
112 Information on the functioning of financial institutions, information pertaining to how financial institutions, or decision makers in financial institutions, actually act and how they are expected to react to unforeseen contingencies in the future helps incorporate dynamic elements into financial regulation. Several mechanisms can increase the information for rulemaking in a more timely fashion, including but not limited to: (1) contingent capital, (2) governmental contracts in the form of non-and deferred prosecution agreements, (3) venture capital finance allocation, and (4) crowdfunding. _See supra_ note 6. 113 _See supra_ note 6.
> 114 Jean-Jacques Dethier _, Governance and Economic Performance: A Survey_ 49 (Discussion Papers on Development Policy No. 5, Center for Development Research, Bonn, April 1999), http://goo.gl/xzHwxK.
> 115 Christian Kirchner, _Market Organizations: A New-Institutional Perspective_ , 151 J. INST. THEORETICAL & ECON. 260, 265 (1995) (“By introducing various examples of the way in which private actors have tried to escape the onerous results of regulation, and showing that law courts often have favoured such ‘inventions,’ the paper developed a hypothesis of how public and private actors act and react in the complex process of forming the institutional arrangements of market.”).
> 116 Kaal in FS Kirchner, _supra_ note 6, at 1214 (“Rules can be adaptable if institutions and rulemaking processes integrate dynamic elements that produce timely, relevant, and decentralized information for rulemaking.”).
117 Rules as outcomes are the result of the institutional design of rulemaking and reinforce the institutional design. Existing rules create a feedback effect for the rulemaking process itself. Rules with suboptimal characteristics are the results of institutional arrangements and reinforce suboptimal institutional arrangements and rulemaking processes. Stable and presumptively optimal rules reinforce a rulemaking process with an institutional structure that perpetuates stable elements in rules.
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
feedback processes.<sup>118</sup> The unrestricted exchange of information between public- and private rulemakers can create regulatory synergies that increase the availability of relevant, decentralized, and timely information for rulemaking<sup>119</sup> and facilitate feedback effects.<sup>120</sup> Existing rules can provide feedback effects for the rulemaking process itself. Through feedback effects, rulemakers in a dynamic regulatory framework can adopt rules that are adaptable to future states of the world _._
Feedback processes in a dynamic regulatory framework facilitate the enhancement of information for regulation. Dynamic regulatory tools as a regulatory supplement enable rulemakers to adapt to regulatory contingencies if and when they arise because a feedback effect provides relevant, timely, decentralized, and institution-specific information ex-ante. By increasing the availability of information ex-ante, dynamic regulatory tools help lower unforeseen contingencies in the rulemaking process. Improved information for rulemaking also helps maintain certainty in the rulemaking process.
Dynamic elements in rulemaking can help optimize the timing of rule enactment. A core problem for most regulation is its inaccurate and delayed timing. Because of the collective action problem of regulation, path dependencies, and political inertia, among many other reasons, regulation is mostly reactive, following business cycles, and not preemptive, e.g., enacted before crises when governance improvements are most needed.<sup>121</sup> Instead of the traditional rule revisions ex-post after rules have emerged as suboptimal, feedback effects in dynamic regulatory processes can increase the
118 Including synergies and exchanges between public rulemakers and public regulators as well as intrajurisdictional feedback processes between private and public rulemakers in a given jurisdiction. Christian Kirchner et al., _Regulatory Competition in EU Corporate Law After Inspire Art: Unbundling Delaware’s Product for Europe_ , 2 EURO. COMPANY & FIN. L. REV. 159 (2015) _;_ Kirchner et al _., Europäischer vs. US-amerikanischer Wettbewerb der Gesellschaftsrechte - Abschied vom alten DelawareVerbundprodukt. Die Aktiengesellschaf_ , 13 DIE AKTIENGESELLSCHAFT 469 (2012).
119 Private rulemakers can have a comparative advantage over public rulemakers because, unlike their public counterparts, they often can produce necessary, comparable, decentralized, and institution-specific information for rulemaking. Unlike public rulemakers, private rulemakers often have access to decentralized information and can more readily react to emerging, decentralized, and institution-specific information. Interaction between public rulemakers and private rulemakers, including the exchange of emerging information, creates a feedback process between them that increases the availability, timeliness, and quality of information available to the public rulemaker. Based on the optimized information for rulemaking, the public rulemaker can take regulatory action when the relevant information for rulemaking becomes available. The interplay between public and private rulemakers and the associated feedback process enables a learning process and optimization process for the public rulemaker.
> 120 _Id._ The rulemaking process is enhanced through the competition between private- and public rulemakers. Competition between different public rulemakers can require public rulemakers to meet consumers’ and legal addressees’ quality expectations and preferences. Consumer choice can add a dynamic element to the rulemaking process because public rulemakers in a given jurisdiction can adjust rulemaking to consumer choice once consumers have opted out of a suboptimal regulatory regime in that jurisdiction. Consumer choice creates a feedback effect for the public rulemaker. It can thus facilitate appropriate information for rulemaking and enable anticipation and modification of the public rulemaker’s next action in the rulemaking process. Feedback effects may exist between public rulemakers and parties who are subject to informal rules if public rulemakers observe the preferences and efficient solutions of parties who are subject to informal rules. In effect, parties who are subject to informal rules are signaling their preferences and efficient solutions to the public rulemaker. Public rulemakers can benefit from the additional insights such signaling may produce for the public rulemaking process.
> 121 _See supra_ text accompanying notes 70–72
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
availability of relevant information for rulemaking ex-ante and anticipate necessary revisions before rules emerge as suboptimal.<sup>122</sup> By identifying possible contingencies and necessary rule revisions with optimized information ex-ante, dynamic regulatory supplements make adaptive rulemaking the focal point for rulemaking in a dynamic framework.<sup>123</sup>
Industries and regulatory agencies can benefit from dynamic regulation as a regulatory supplement, an informal and flexible approach to regulatory issues. While the Administrative Procedure Act (APA)<sup>124</sup> institutes regulatory procedures to protect regulated industries,<sup>125</sup> requiring a full rulemaking process to apply,<sup>126</sup> industries and regulatory agencies often prefer unenforceable rules and no judicial involvement.<sup>127</sup> Both industries and regulators are affected by costly, untimely, overextended, and ossified rulemaking or adjudicative procedures and the associated regulatory uncertainty.<sup>128</sup> Dynamic regulatory supplements would not violate mandated regulatory procedures in the APA<sup>129</sup> as the full lawmaking process still applies.<sup>130</sup> Dynamic regulation functions as a regulatory supplement to help optimize the existing rulemaking process.
# **V. Data as Input for Dynamic Regulation**
The data used for this study provides general guidance on innovation trends. The venture capital (VC) investment allocation in the deals encapsulated by the dataset provides a window to the market for innovative products. The VC investments result from VC deal flow screening. Venture capital’s assessment of innovative products and initiatives generates institution- and industry-specific decentralized information on innovation trends. While deal flow, that is, the totality of potential deals and business plans screened by venture capitalists, would provide an optimal assessment of innovation trends, deal flow data is not available. We therefore use venture capital investment allocations to assess innovation trends. To illustrate venture capital’s ability to identify innovation trends, we use a PitchBook Data, Inc. dataset on venture capital deals in the United States. The dataset comprises 77,508 deals involving 37,298 companies from 2005 through 2015. It includes all VC deals and all VC stages.
Figure 1 below shows that since 2009 the number of VC deals and the total capital invested increased substantially from around $30 billion in total capital invested to around $80 billion invested. This increase illustrates VC investors’ attempts to capitalize on increasing innovation in multiple industries.
> 122 Kaal, Wake Forest, _supra_ note 6; Kaal in FS Kirchner, _supra_ note 6.
> 123 _Id_ .
> 124 Administrative Procedure Act, 5 U.S.C. §§ 551-559, 701-06 (2012).
> 125 See 5 U.S.C. §§ 551-559 (2012).
> 126 In the typical process, the APA requires that for a rule to become effective, the proposed rule must go through what is known as notice-comment rulemaking, often a lengthy process in which the public is given an opportunity to comment on the proposed rule, and the agency must consider such comments before promulgating a final regulation. 5 U.S.C. § 553(b), (c) (2012).
> 127 Wu, _supra_ note 22, at 1843 (“Both agency and industry will sometimes share an interest in an informal and flexible regime that resembles an unenforceable ‘letter of intent’ in the world of private contracts. The costs of a slow-moving, ossified rulemaking or adjudicatory procedure, with its accompanying uncertainty and litigation costs, fall on both industry and agency.”).
> 128 McGarity, _supra_ note 89, at 1387-96.
> 129 Administrative Procedure Act, 5 U.S.C. §§ 551-559, 701-706 (2012).
> 130 S _upra_ note 124.
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
<!-- Start of picture text -->
$90.00B 14,000<br>12,870<br>$80.34B<br>$80.00B<br>11,099 $69.65B 10,929 12,000<br>$70.00B<br>10,000<br>8,951<br>$60.00B<br>$50.00B $44.55B 7,324 $41.38B $45.25B 8,000<br>5,951<br>$40.00B $35.98B $36.78B<br>6,000<br>$28.76B 4,409 4,847 4,841 $27.74B $32.41B<br>$30.00B<br>$23.64B<br>2,962 3,325 4,000<br>$20.00B<br>2,000<br>$10.00B<br>$0.00M 0<br>2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015<br>Total Capital Invested Deal Count<br><!-- End of picture text -->
Figure 1: VC Capital Invested <u>and Deal Count, United States, 2005-2015. The dataset comprises 77,508 deals</u> involving 37,298 companies. Data provided by PitchBook.
# _1. VC Deal Characteristics_
Figure 2 below shows that since 2010 both the early stage and late stage venture capital internal rate of return (IRR) has been positive and increasing steadily across different VC stages and overall. Figure 2 covers several relevant VC stages, giving us the opportunity to determine which stage we can best focus on for purposes of identifying more optimal VC investment allocation signals for rulemaking.
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
<!-- Start of picture text -->
25.00%<br>20.00%<br>15.00%<br>10.00%<br>5.00%<br>0.00%<br>-5.00%<br>-10.00%<br>Venture Capital<br>-15.00%<br>Venture Capital -<br>Early Stage<br>-20.00%<br>Venture Capital -<br>Later Stage<br>-25.00%<br>2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015<br><!-- End of picture text -->
Figure 2: VC Fund IRR by VC Fund Type, 2005-2015. Dataset consists of 6,157 VC funds from 2005-2015. Data provided by PitchBook.
Figure 3 below shows that since 2014 the latest round of VC investment in our dataset (19<sup>th</sup> Round) has become increasingly important. This could suggest that some of the established innovative products are continuing to receive VC financing rather than new startups. However, early stage and angel investments also show an increase starting in 2011 as compared with the overall VC rounds in our dataset.
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
<!-- Start of picture text -->
$90.00B 19th Round<br>18th Round<br>$80.00B 17th Round<br>16th Round<br>$70.00B 15th Round<br>14th Round<br>$60.00B<br>13th Round<br>12th Round<br>$50.00B<br>11th Round<br>10th Round<br>$40.00B<br>9th Round<br>8th Round<br>$30.00B<br>7th Round<br>6th Round<br>$20.00B<br>5th Round<br>4th Round<br>$10.00B<br>3rd Round<br>2nd Round<br>$0.00M<br>1st Round<br>2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015<br>Angel<br><!-- End of picture text -->
Figure 3: VC Capital Invested by VC Round, United States, 2005-2015. Dataset comprises 77,508 deals involving 37,298 companies. Data provided by PitchBook.
Figures 4, 5, and 6 below show that most industries in the dataset are substantially represented up to Series F. Big Data applications and SaaS are disproportionally high represented in Series I to K. The rest of the industries represented up to Series F play no or a much less significant role in Series I to K. This data seems to suggest that VC funds perceive Big Data and SaaS as the most promising businesses applications that justify continuing investments. Additionally, Table 1 shows that among the Top 25 companies by highest VC series investment deals involving SaaS and Big Data are among the top represented industry verticals. This data is consistent with the rest of the data in Figures 8-11 below. A more granular assessment of the VC investments in Big Data and SaaS can provide much needed feedback for regulators on associated regulatory needs.
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
<!-- Start of picture text -->
Ser ies K<br>Series J<br>Series I<br>Ser ies H<br>Ser ies G<br>Series F<br>Series E<br>Ser ies D<br>Series C<br>Ser ies B<br>Ser ies A<br>adodtnnt.<br>Seed<br>100 200 300 400 500 600 700 800 006 1000<br>Wearables Security SAAS Robotics Oncology Nano Mobile Marketing Manufactur ingWellness ■Life Sci<br>IOT InfrastructureHealthTech FinTech EdTech E-Commerce CleanTech Big Data Ad Tech 3d Printing<br><!-- End of picture text -->
Figure 4: VC Investments by Periods/Stages and Industry from 2005 through 2015. Periods/stages include Seed and Series A through Series K. Industries represented include: 3d Printing, AdTech, Big Data, CleanTech, E-Commerce, EdTech, Fin Tech, Health Tech, Infrastructure, Internet of Things, Life Sciences, LOHAS and Wellness, Manufacturing, Marketing, Mobile, Nano, Oncology, Robotics, SaaS, Security, and Wearables. The following ratios are calculated by dividing capital invested sum by the deal count for the cumulative deals from 2005 through 2015. Data provided by PitchBook.
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
<!-- Start of picture text -->
1000<br>900<br>800<br>700<br>600<br>500<br>400<br>300<br>200<br>100<br>0<br>3d Printing Ad Tech Big Data CleanTechE-Commerce EdTech FinTech HealthTechInfrastructure IOT Life Sci WellnessManufacturingMarketing Mobile Nano Oncology Robotics SAAS SecurityWearables<br>Seed Ser ies A Series BSeries C Series D SeriesESeriesFSeries G SeriesHSeriesISeriesJSeries K<br><!-- End of picture text -->
Figure 5: VC Investments by Periods/Stages and Industry from 2005 through 2015. Periods/Stages include Seed and Series A through Series K. Industries represented include: 3d Printing, AdTech, Big Data, CleanTech, E-Commerce, EdTech, Fin Tech, Health Tech, Infrastructure, Internet of Things, Life Sciences, LOHAS and Wellness, Manufacturing, Marketing, Mobile, Nano, Oncology, Robotics, SaaS, Security, and Wearables. The following ratios are calculated by dividing capital invested sum by the deal count for the cumulative deals from 2005 through 2015. Data provided by PitchBook.
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
<!-- Start of picture text -->
1800<br>1600<br>1400<br>1200<br>1000<br>800<br>600<br>400<br>200<br>0<br>Seed Series A Series B Series C Series D Series E Series F Series G Series H Series SeriesJ Series K<br>3d Printing Ad Tech B CleanTech E-Commerce EdTech FinTech HealthTech InfrastructureIOT ■Life Sci<br>■Wellness ManufacturingMarketing Mobile Nano Oncology Robotics ■SAAS ■Security Wearables<br><!-- End of picture text -->
Figure 6: VC Deal/Dollar Investment Ratios, Stage/Series by Industry Vertical from 2005-2015 and VC Investments for the Following Periods/Stages: Seed and Series A through Series K. Pitchbook provides data for the following industries: 3d Printing, AdTech, Big Data, CleanTech, E-Commerce, EdTech, Fin Tech, Health Tech, Infrastructure, Internet of Things, Life Sciences, LOHAS and Wellness, Manufacturing, Marketing, Mobile, Nano, Oncology, Robotics, SaaS, Security, and Wearables. The ratios are calculated by dividing capital invested sum by the deal count for the cumulative deals from 2005 through 2015. Data provided by PitchBook.
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
|**Company**<br>**Name**|**Industry**<br>**Vertical**|**Current**<br>**Financing**<br>**Status**|**Deal Date**|**% Acquired**|<br>**VC**<br>**Round**|**Series**|**Total**<br>**Invested**<br>**Equity**|**Year**<br>**Founded**|
|---|---|---|---|---|---|---|---|---|
|Palantir<br>Technologies|Big Data, SaaS|Venture<br>Capital-Backed|12/13/2015|4.29|12th<br>Round|Series K|880|2004|
|Palantir<br>Technologies|Big Data, SaaS|Venture<br>Capital-Backed|11/3/2014|6.91|10th<br>Round|Series I|635.71|2004|
|Palantir<br>Technologies|Big Data, SaaS|Venture<br>Capital-Backed|11/26/2014|0.33|11th<br>Round|Series J|400|2004|
|Motricity|Mobile|Corporate<br>Backed or<br>Acquired|12/31/2007|20.11|9th Round|Series I|185|2001|
|Onconova<br>Therapeutics<br>(ONTX)|Life Sciences,<br>Oncology|Private Equity-<br>Backed|8/10/2012||10th<br>Round|Series J|50|1998|
|Tocagen|Life Sciences,<br>Oncology|Venture<br>Capital-Backed|10/21/2015|12.98|9th Round|Series H|46.2|2007|
|TherOx|Life Sciences|Venture<br>Caital-Backed|7/21/2005|58.68|9th Round|Series I|40.8|1994|
|Revolution<br>Foods||p<br>Venture<br>Capital-Backed|11/19/2015|11.85|12th<br>Round|Series I|35.22|2006|
|Alien<br>Technology|Internet of<br>Things,<br>Manufacturing|Venture<br>Capital-Backed|3/12/2007|15.55|12th<br>Round|Series I|33|1994|
|OptiScan<br>Biomedical||Venture<br>Capital-Backed|4/14/2010||11th<br>Round|Series I|31.5|1994|
|BioNumerik<br>Pharmaceuticals|Life Sciences,<br>Oncology|Private Equity-<br>Backed|3/12/2009||10th<br>Round|Series K|30|1995|
|TherOx|Life Sciences|Venture<br>Caital-Backed|2/20/2008|28.33|10th<br>Round|Series J|29.99|1994|
|Alien<br>Technology|Internet of<br>Things,<br>Manufacturing|p<br>Venture<br>Capital-Backed|3/12/2007|15.89|<br>11th<br>Round|Series I|28.47|1994|
|Onconova<br>Therapeutics<br>(ONTX)|Life Sciences,<br>Oncology|Private Equity-<br>Backed|7/18/2012||9th Round|Series I|26.44|1998|
|Allconnect||Venture<br>Capital-Backed|2/21/2007|10.08|11th<br>Round|Series I|20|1998|
|NeuroPace|Life Sciences,<br>LOHAS &<br>Wll|<br>Venture<br>Capital-Backed|7/8/2013|3.91|<br>6th Round|Series I|18.05|1997|
|AFS<br>Technologies|eness|Private Equity-<br>Backed|8/29/2008||3rd Round|Series I|13|1985|
|Agraquest|Life Sciences,<br>Manufacturing|Formerly VC-<br>Backed|1/1/2009|11.32|13th<br>Round|Series J|10.2|1995|
|Agraquest|Life Sciences,<br>Manufacturing|Formerly VC-<br>Backed|6/18/2009|6.56|12th<br>Round|Series I|10.13|1995|
|CafePress<br>(PRSS)|E-Commerce|Venture<br>Capital-Backed|11/24/2008|1.39|3rd Round|Series I|8.3|1999|
|Agraquest|Life Sciences,<br>|Formerly VC-<br>|3/25/2011|6.89|14th<br>|Series K|7.5|1995|
|Exa (EXA)|Manufacturing|Backed<br>Formerly VC-<br>|5/21/2008|5.79|Round<br>8th Round|Series I|6|1991|
|Iconix<br>Biosciences|Manufacturing|Backed<br>Venture<br>Capital-Backed|1/24/2007|3.95|10th<br>Round|Series I|4|1997|
|Iconix<br>Biosciences|Manufacturing|Venture<br>Capital-Backed|12/18/2006|2.06|9th Round|Series I|2|1997|
|OptiScan<br>Biomedical||Venture<br>Capital-Backed|7/2/2010||12th<br>Round|Series J|0.7|1994|
Table 1: Top 25 Companies by Highest VC Series Investment. Data provided by PitchBook.
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
# _2. Industries Represented_
Figure 7 shows that the average number of deals of the top 25 United States venture capital investors between 2005 and 2015 focuses on information technology, mobile technology, and software as a service (SaaS). While the data in Figure 7 arguably only confirms media reporting on innovation trends between 2005 and 2015, the VC data used herein may provide earlier signals for regulators to identify possible areas of regulatory need in the face of increasingly disruptive innovation.
<!-- Start of picture text -->
$60.00 30<br>$50.00 25<br>$40.00 20<br>$30.00 15<br>$20.00 10<br>1<br>$10.00 5<br>$0.00 ■ _ _ 0<br>B2Fínancial Services B 2CEnergyMaterials & ResourcesHCI3D PrintingAdTechBig DataCleanTechE-CommerceEdTechFinTechHealthTechInfrastructureInternet of ThingsLife SciencesLOHAS & WeManufacturingMarketing Tech l ness NanotechnologyMobileRobotics and DronesOncologyWearables & Quantified SelfSaaSSecurity<br>Aver age No. of Deals Aver age Deal Size<br><!-- End of picture text -->
Figure 7: Sector/Industry Allocations (both number of deals and deal size) for the Top 25 United States Venture Capital Investors between 2005 and 2015. Investors include the following: Y Combinator, 500 Startups, Techstars, SV Angel, New Enterprise Associates, Western Technology Investment, Intel Capital, Kleiner Perkins Caufield & Byers, Sequoia Capital, Andreessen Horowitz, GV, Plug and Play Ventures, First Round Capital, Accel Partners, Khosla Ventures, Draper Fisher Jurvetson, Individual Investors, David McClure, Greylock Partners, MassChallenge, General Catalyst Partners, Lerer Hippeau Ventures, Bessemer Venture Partners, National Science Foundations, Lightspeed Venture Partners. Data provided by PitchBook.
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
Figure 8 shows that since 2014 VC deals in the United States increasingly focused on information technology. While this confirms the trends suggested in Figure 7, and Figures 7 and 8 together arguably only confirm media reporting on innovation trends from 2005 through 2015, the VC data used herein, and especially a more granular examination of the data by regulators, may provide earlier signals for regulators to identify possible areas of regulatory need in the face of increasingly disruptive ~~innovation.~~
<!-- Start of picture text -->
$90.00B<br>$80.00B<br>$70.00B<br>$60.00B Materials & Resources<br>■ Informa@on Technology<br>$50.00B<br>■ Healthcare<br>$40.00B ■ Financial Services<br>$30.00B ■ Energy<br>■ B2C<br>$20.00B<br>■ B2B<br>$10.00B<br>$0.00M _<br>2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015<br><!-- End of picture text -->
Figure 8: VC Capital Invested by Primary Industry, United States, 2005-2015. Dataset comprises 77,508 deals involving 37,298 companies. Data provided by PitchBook.
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
Figure 9 shows that between 2006 and 2013, VC investment allocations focused on applications in the context of the Internet of Things. Since 2010, however, both SaaS and Wearables have taken over the primary focus in VC investments in this dataset. This is consistent with Figures 7 and 8. The main point here is that a more granular examination of these trends can provide guidance to regulators to help identify possible areas of regulatory need in the face of increasingly disruptive innovation.
<!-- Start of picture text -->
$120.00B<br>Wearables & Quan;fied<br>Security<br>$100.00B<br>SaaS<br>Robo;cs and Drones<br>$80.00B<br>Oncology<br>Nano-technology<br>$60.00B<br>Mobile<br>Marke;ng Tech<br>$40.00B<br>Manufacturing<br>LOHAS & Wellness<br>$20.00B<br>Life Sciences<br>Internet of Things<br>$0.00M<br>Infrastructure<br>2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015<br>HealthTech<br><!-- End of picture text -->
Figure 9: VC capital invested by industry vertical United States 2005-2015. Dataset comprises 77,508 deals involving 37,29 ~~8 companies. Data provided by PitchBook.~~
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
Figure 10 shows the relative impact of VC investments by industry. This data examination is included to suggest that using investment ratios by industry, among other more granular examinations of trends, can help inform rulemakers on innovation and the corresponding need for rulemaking.
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
400 350 300 250 200 150 100 50 0 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 All 3D Prin<ng AdTech BigData CleanTech E-Commerce EdTech FinTech HealthTech Infrastructure Internet of Things Life Sciences LOHAS & Wellness Manufacturing Marke<ng Tech Mobile Nano-technology Oncology Robo<cs and Drones SaaS Security Wearables & Quan<fied Self
Figure 10: VC Investment Ratio by Industry. The investment ratio is calculated as VC investment amount divided by VC deals per year and industry. Dataset comprises 77,508 deals involving 37,298 companies. Data provided by PitchBook.
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
Figure 11 shows that the majority of VC investments in the majority of industries represented are later stage VC investments, following early seed VC investments in the same industries. This data may be interpreted and perhaps discounted as VCs simply following up with later stage investments in companies they had invested seed capital in at an earlier stage. Accordingly, the data could suggest risk-averse behavior by VC funds. At the same time, however, VC funds are not likely to continue with later stage investments if contractual performance obligations/thresholds of their portfolio companies were not fulfilled to a substantial degree.
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
<!-- Start of picture text -->
45<br>40<br>35<br>30<br>25<br>20<br>15<br>10<br>5<br>Pre/Accelerator/Incubator Angel Seed Early Seed VC Later Stage VC Restart Angel Restart Early Stage Restart Late Stage<br>3d Printing ■AdTech BigData CleanTech ■E-Commerce EdTech<br>Fin Tech ■HealthTech Infrastructure ■ loT ■Ufe Sciences LOHAS & Wellness<br>Manufacturing Marketing Tech Mobile Nano ■ Oncology Robotics<br>■SoaS ■Security Wearables & Quantified Self<br><!-- End of picture text -->
Figure 11: VC Deal Type Ratios by Industry Vertical and Periods/Stages, 2005-2015: Pre/Accelerator/Incubator, Angel, Seed, Early Seed VC, Later Stage VC, Restart Angel, Restart Early Stage, and Restart Late Stage. Pitchbook provides data for the following industries: 3d Printing, AdTech, Big Data, CleanTech, E- Commerce, EdTech, Fin Tech, Health Tech, Infrastructure, Internet of Things, Life Sciences, LOHAS and Wellness, Manufacturing, Marketing, Mobile, Nano, Oncology, Robotics, SaaS, Security and Wearables. The data runs from 2005 through 2015. The following ratios are calculated by dividing capital invested sum by the deal count for the cumulative deals between 2005 through 2015. Data provided by PitchBook.
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
# **VI. Venture Capital as Dynamic Regulation of Disruptive Innovation**
VC capital allocations are a potential treasure trove of valuable information for regulators. While VC capital allocations offer only one among other possible ways for regulators to identify areas that may require their attention in the future and facilitate anticipatory rulemaking, the data analysis in this study shows that venture capitalists’ finance allocation and the implicit assessment of innovative products, businesses, and initiatives generates highly relevant institution- and industry-specific decentralized information on innovation trends. As such, venture capitalists’ innovation-driven finance allocation can provide feedback on innovation trends for regulators and rulemakers, optimize the timing of regulation, and facilitate anticipatory rulemaking. Venture capitalists’ investment allocations can provide regulators and rulemakers with a measurement of risks associated with innovative developments as well as the economic opportunities associated with such innovation.
The VC finance allocation can create feedback effects for dynamic regulation of innovation. The theory of lawmaking describes feedback effects as rulemakers’ reactions to institutional change or private actors’ reactions and counteractivities to institutional constraints. This paper shows that data on VC finance allocation can distill institutional changes of companies that received VC financing. As such, the regulators’ reaction to institutional change could be informed by VC financing allocations. VC feedback effects provide relevant, timely, decentralized, and institution-specific information ex-ante.
VC’s finance allocations can help optimize the timing of rulemaking. A core problem for regulation of disruptive innovation is inaccurate and delayed timing.<sup>131</sup> Regulation is mostly reactive, following business cycles, rather than proactive. Data on VC investments allows regulators to see where innovation trends are going and the possible risks they entail before disruptive innovation materializes. Should regulators consider data on VC investments and their possible implications for disruptive innovation in rulemaking processes or legislative proposals, it is conceivable that the identification of disruptive innovation and associated legal concerns could be addressed in a more timely fashion. By identifying possible contingencies and necessary rule revisions with optimized information from VC investments ex-ante, before disruptive innovation creates regulatory issues, regulators could anticipate regulatory needs.
Venture capitalists’ innovation-driven finance allocations can facilitate anticipatory rulemaking. Through the feedback effects that are facilitated by venture capitalists’ finance allocations towards innovative products, rulemakers can obtain timely and decentralized industry- and entity-specific information that allows them to adapt rules to anticipate regulatory issues. For instance, information on industry-specific VC investments allows regulators to anticipate regulatory needs in certain industries that are associated with the highest levels of disruptive innovation. Disruptive innovation may here be quantified with the number of VC deals or the VC dollar investments in such industries. Multiple metrics with VC investment information can be examined. The broad data on VC investments allows for a detailed analysis of innovation trends with anticipatory qualities that could be directly utilized by regulators, including, as needed, on a much more granular level. While this paper does not provide a granular evaluation
> 131 _See supra_ notes 121-123 and accompanying text.
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
of the data, regulators would be able to obtain the relevant innovation data presented herein and analyze areas of concern or regulatory need more granularly. For instance, a more granular approach to the data could suggest that regulators and rulemakers should start taking robotics and drones much more seriously. Figure 12 shows that we have seen more and more investments in these areas since 2011. Of course, one could argue that the decrease in investments in the number of the later stage rounds in the United States indicates that the technological development in this area is stalling. However, a closer look at the industry shows that companies relocate the later stage development to other countries due to the regulatory disconnect that exists in the United States.
<!-- Start of picture text -->
160.0<br>140.0<br>120.0<br>100.0 Grants<br>Later stage VC<br>80.0<br>Early stage VC<br>60.0 Seed<br>Angel<br>40.0<br>Pre/Accelerator/Incubator<br>20.0<br>.0<br>2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016<br><!-- End of picture text -->
Figure 12: VC Investments in Robotics and Drones in the United States. The total dataset comprises 77,508 deals involving 37,298 companies. Data provided by PitchBook.
Critics may suggest that the market often undervalues venture-capital-financed innovation in products and ideas. We realize that a trend for venture-capital-financed technology companies, among others, to stay private<sup>132</sup> and market undervaluation of formerly venture-capital-financed companies<sup>133</sup> suggest that the innovation potential identified by venture-capital-finance allocation may not always be shared by the market at large. However, our objective in this article is not to discuss the long-term potential of venture-capital-financed innovation. Rather, we suggest that venture capitalists’ innovation-driven finance allocation can provide feedback for regulators and rulemakers, optimize the timing of regulation, and facilitate anticipatory rulemaking.
> 132 Nicole Bullock, Gavin Jackson & Jennifer Hughes, _Global Listings Drop as Tech Groups Stay Private_ , FIN. TIMES, January 2/3 2016, at 9.
> 133 William R. Kerr & Ramana Nanda, _Financing Innovation_ (Harvard Bus. Sch. Working Paper 15-034), https://perma.cc/NJ4D-SSP5 (“[M]odel of managerial myopia in which shareholders cannot properly evaluate investments in long-term innovative projects due to asymmetric information, and therefore markets tend to undervalue the stocks of companies engaged in such projects.”).
HOW TO REGULATE DISRUPTIVE INNOVATION – _FROM FACTS TO DATA_
Critics may also suggest that venture-capital-financed innovation is not a reliable measure of innovation that requires regulatory action. In other words, critics may allege that even if regulators can acquire the depth of information needed to engage in anticipatory rulemaking, there may be risk of waste of scarce regulatory resources because VC funds make many investments, not all of which turn out to be successful. Moreover, if companies that received VC financing are still incubating, it may be largely unclear what regulatory issues these companies might raise in practice. First, the use of VC finance allocation is only one of several evolving data available to identify future regulatory needs. Second, over time, regulators will have an ever-increasing access to high-quality, decentralized, and real-time information for rulemaking. If and when multiple data analyses together point unanimously in the direction of a certain demand for regulatory action, regulators should take anticipatory measures after cross-validation and triangulation. Currently no regulatory processes and data evaluation exist that could facilitate such analyses and anticipatory regulatory action.
# **VII. Conclusion**
This paper adds to increasing evidence that companies that received VC investments have─and continue to─outrun regulation and regulatory efforts. Companies that received VC investments drive innovation trends in the United States and abroad. This paper also shows that the existing regulatory processes are suboptimally equipped to address the challenges of exponential disruptive innovation. The notice and comment procedures of the APA<sup>134</sup> and the SEC are only some of the examples illustrating the suboptimal regulatory response rates with regards to regulatory challenges presented by VC-backed companies and the associated disruptive innovation.
The data presented in this paper shows that VC investment allocation decisions provide relevant broad and general guidance on innovation trends and associated regulatory issues. The paper shows that the use of VC investment allocation data can help facilitate anticipatory regulation associated with disruptive innovation, but it emphasizes that VC investment allocation data is only one form of the emerging data sources that can be used as signaling for regulatory process optimization. Other data sources are and may increasingly become available. With the existing and emerging data and a much more granular examination of these data trends, regulators can further increase their ability to examine and anticipate regulatory needs.
> 134 _See supra_ notes 124-130 and accompanying text.