Seth C. Oranburg

    ENCOURAGING ENTREPRENEURSHIP AND
     INNOVATION THROUGH REGULATORY
            DEMOCRATIZATION

                        Seth C. Oranburg*

                            I. INTRODUCTION

Entrepreneurship provides a path to prosperity

or many people.1 In particular, women and minorities pre — er entrepreneurship as their path to achieve the American Dream. 2 In their striving, their startups and small businesses bene — it our entire society.3 Entrepreneurial innovation has a positive impact on social wel — are.4 For these reasons, the — ederal government has implemented numerous policies designed to support small businesses and promote startup innovation. 5

However, these policies appear to be inadequate. Recent studies have shown that startups and small businesses are less success — ul

* © Seth C. Oranburg, Assistant Pro --- essor o ---  Law, Duquesne University.
 1 Entrepreneurship is the ability to “make something” o ---  onesel --- . As coined in 1931 by John T. Adams, entrepreneurship symbolized the democratization o ---

opportunity that is the American dream. Rebecca Gill, The Evolution o

Organizational Archetypes: From the American to the Entrepreneurial Dream, 80 COMM. MONOGRAPHS 331, 337 (2013). 2 See Alicia M. Robb, Entrepreneurial Per — ormance by Women and Minorities: The Case o — New Firms, 7 J. DEVELOPMENTAL ENTREPRENEURSHIP 383, 383–97 (2002). 3 Entrepreneurship positively impacts social wel — are in two ways: by major innovations that shock the equilibrium through creation o — a new product or process, which is also re — erred to as Schumpeterian entrepreneurship or creative destruction, and by minor innovations that bring the market price close to equilibrium, which may be called Kirzner entrepreneurship. See Samuel Bostaph, Schumpeter vs. Kirzner on Entrepreneurs, MISES INST. (May 16, 2019), https://mises.org/wire/schumpeter-vs-kirzner-entrepreneurs [https://perma.cc/D2GY-YRFP]; see also in — ra Part III. 4 See Bill Conerly, Innovation Bene — its Society, Not Just the Rich, FORBES (Oct. 21, 2018, 7:39 AM), https://www. — orbes.com/sites/billconerly/2018/10/21/innovation-bene — its- beyond-the- — ilthy-rich/#22d61 — eb6a6a [https://perma.cc/8CYK-GT4B]; see also William D. Nordhaus, Schumpeterian Pro — its in the American Economy: Theory and Measurement 1, 34 (Nat’l Bureau o — Econ. Research, Working Paper No. 10433, 2004), https://www.nber.org/papers/w10433.pd —

[https://perma.cc/7AAU-RFXG] (“Using data

rom the U.S. non — arm business section, I estimate that innovators are able to capture about 2.2 percent o — the total social surplus — rom innovation.”). 5 See Lewis D. Solomon & Garry S. Grossman, Tax and Non-Tax Policies to Promote Capital Formation: Stimulating High Technology in the 1980’s, 1 AM. J. TAX POL’Y 63, 115 (1982).

 Electronic copy available at: https://ssrn.com/abstract=3435274

than large, incumbent — irms.6 Despite what the shows Shark Tank and Silicon Valley depict, outside o — certain high-tech — ields, American entrepreneurship is declining. 7

The decline o

innovation could well be the result o — over- regulation. It is axiomatic that innovation is harder, slower, and less success — ul in highly regulated industries. 8 Legal scholars have suggested that this phenomenon might be explained by the distributional e —

ects o — regulation: although well-intentioned lawmakers might try to devise regulatory regimes that support entrepreneurs, the larger incumbents in the market are more able to in — luence and bene — it — rom the regulatory process.9 There — ore, regulations by nature are at odds with startup innovation.10

This argument is incomplete. It is right inso

ar as studies show that higher regulation is correlated with lower innovation. 11 However, this argument is wrong in that it collapses all di —

erent sorts o — regulations into one.

See Jeremy Quittner, The American Dream is Dying: How Entrepreneurs Can Change That, INC. (Jan. 14, 2015), https://www.inc.com/jeremy- quittner/milstein-commission-and-salvaging-the-american-dream.html [https://perma.cc/E6UV-Z2PC]; Steve Case, Can Startups Save the American Dream?, CASE FOUND. (Jan. 14, 2015), https://case --- oundation.org/blog/can- startups-save-american-dream/ [https://perma.cc/Z9XJ-AUJ2].
See Philip Aldrick, With Startups Declining, the American Dream is Beginning to Fade, TIMES (July 2, 2018, 5:00 PM), https://www.thetimes.co.uk/article/with-startups-declining-the-american-dream- isbeginning-to- --- ade-k8grswhv5 [https://perma.cc/NTS7-ABC2].
See Naoimi Fried, Innovating in a Highly Regulated Industry Like Health Care, HARV. BUS. REV. (June 12, 2017), https://hbr.org/2017/06/innovating-in-a- highly-regulated-industry-like-health-care           [https://perma.cc/J535-PTXN] (“Innovating in regulated industries takes longer than in other industries. . . . Expect setbacks. Innovation is rarely easy, especially in regulated industries.”).
See Mirit Eyal-Cohen, The Cost o ---  Inexperience, 69 ALA. L. REV. 859, 863–64 (2018) (“[R]egulations have the potential o ---  a ---

ecting newcomers more perversely than old-timers. . . . This observation reveals regressive regulatory barriers.”). 10 See Victor Fleischer, Regulatory Arbitrage, 89 TEX. L. REV. 227, 280 (2010) (“Firms that can better manage transaction costs can better manage regulatory costs, shi — ting the burden o — those regulatory costs on to those that cannot.”). 11 See, e.g., ORG. FOR ECON. CO-OPERATION & DEV., REGULATORY REFORM

AND INNOVATION 19, https://www.oecd.org/sti/inno/2102514.pd

[https://perma.cc/S994-R9PV]. For example, a long-term international study by the Organization — or Economic Co-operation and Development (OECD) — ound that — ar — ewer patents were issued — or telecommunications innovations in countries where the telecommunications industry is subject to more regulation— Germany and France—as compared to countries where telecommunications are subject to greater competition—the United States and Japan. See id.

 Electronic copy available at: https://ssrn.com/abstract=3435274

This Article will show that not all regulations are created equal. Instead, regulations can be more precisely categorized across two axes: complex versus simple, and rules versus standards.12 This more nuanced analysis o — regulation, split into — our categories, reveals that some categories o — regulation are more likely to impact entrepreneurial innovation than others. Moreover, this provides some hope that smart regulation can provide an optimal middle ground between over-regulation and over-deregulation.13 This requires a more detailed analysis o — the impact o — complex standards, simple standards, simple rules, and complex rules.

Complex standards are especially burdensome

or startups and are likely to have the highest impact on entrepreneurial innovation. 14 This is because the cost o — compliance with complex standards exhibits strong economies o — scale.15 This economic e —

ect gives large incumbents a signi — icant advantage over small startups.

Simple standards are not much better

or most small businesses, inso — ar as the regulator’s discretion and the resulting uncertainty make it di —

icult to develop o —

-the-rack solutions to regulatory problems.16 Some startups, however, sometimes engage in what Jordan Barry and Elizabeth Pollman term “regulatory entrepreneurship,” where startups enter a legal gray area with the speci — ic intent o — clari — ying or changing the law in a way that

avors their business model. 17 But this technique only works — or startups that are highly scalable, closely connected to customers, and have mass appeal. 18

Simple rules, unlike simple standards, can o

ten be made to be equally burdensome — or startups and incumbents. Simple rules are, by de — inition, easy to understand and comply with. There is no great advantage to haveing much experience in dealing with regulation by simple rules. Access to regulators, — or example, is o —

See in --- ra Part IV.
See in --- ra Part V.
See, e.g., supra notes 9–10 and accompanying text.
See in --- ra Part IV.
See Elizabeth Pollman & Jordan M. Barry, Regulatory Entrepreneurship, 90 S. CAL. L. REV. 383, 392 (2017).
Id. (“[S]ome companies pursue a line o ---  business that has a legal issue at its core—a signi --- icant uncertainty regarding how the law will apply to a main part o ---  the business operations, a need  --- or new regulations in order  --- or products to be

easible or pro — itable, or a legal restriction that prevents the long-term operation o — the business. For these entrepreneurs, political activity is generally a major component o — their business models. Essentially, these companies are in the business o — trying to change or shape the law. We term such businesses ‘regulatory entrepreneurs,’ and this class o — business activity ‘regulatory entrepreneurship.’”). 18 See id. at 442–47.

 Electronic copy available at: https://ssrn.com/abstract=3435274

little use where rules are straight — orward enough to be equally en — orced. Richard A. Epstein also noted that simpli — ying rules decreases error costs o — getting the rule wrong, and that simple rules tend to be the most e —

icient. 19

Complex rules, on the other hand, demand some expensive analysis, at least the — irst time they are encountered. 20 Over time, complex rules can be — igured out. 21 This advantages larger and older — irms, which encounter the same rules time and again. 22 This would indeed lead to a disparate negative impact on small and young — irms—entrepreneurs.23

Complex rules would have a disparate impact on startups, except that technology can disrupt the boundaries o — a — irm. Innovative regulatory products make it easier to “rent” than “buy” compliance solutions.24 The technological innovation o — “regulatory democratizations” makes regulatory in — ormation and compliance costs accessible — or a wider range o — startups and small businesses.25 Entrepreneurial innovations o — these regulatory compliance solutions are identi — ied in The Startup Study26 and are described in the case studies in this Article. 27

To put this another way, technology reduces the errors and administrative costs o — complex rules. As a result o — regulatory democratization, complexity o — rules is less problematic where technology has made it relatively easy to comply, even with very complex rule-type regulations.28 The case studies show six o —

these innovative responses to regulatory challenges, where technology has enabled scalable solutions to regulatory compliance problems that make the impact o — certain regulations less disparate upon startups.29

 19 See RICHARD A. EPSTEIN, SIMPLE RULES FOR A COMPLEX WORLD 21–36 (1995). In economic terms, there is a “great trade-o ---

. . . between social incentives and administrative costs” and legal systems should “minimize the sum o — administrative (including error) costs and the costs associated with the creation o — poor incentives — or individual action.” Id. at 30, 32. 20 C — . Eyal-Cohen, supra note 9, at 874. 21 See id. 22 See id. at 872–74. 23 C — . Carol Go — orth, Securities Treatment o — Tekenized O —

ering Under U.S. Law, 46 PEPP. L. REV. 405, 410 (2019) (noting the disparate impact o — aggressive regulation on startups in the cryptotransaction sphere). 24 See in — ra Part III. 25 See in — ra Part III. 26 See in — ra Part II. 27 See in — ra Part III. 28 See in — ra Section IV.B.3. 29 See in — ra Part III.

 Electronic copy available at: https://ssrn.com/abstract=3435274

This Article is not the — irst to recognize that regulations have a disparate impact upon startups, but its new — inding o — regulatory democratization contributes to the conversation on how to make regulations that encourage entrepreneurial innovation. In particular, regulatory democratization sheds new light on three proposed policy changes.

A current hot topic in the regulatory literature is to create “regulatory sandboxes.”30 These sandboxes should be called regulatory sand traps because they are going to mire startups in the shi — ting sands o — vague standards and vast regulatory discretion. Admission to regulatory sandboxes is entirely up to the regulator’s discretion,31 and this advantages companies that have relationships with regulators.32 That tends to be old and large companies. 33 Moreover, since the sandboxes have no rules o — which to speak, it is impossible to create any type o — technological approach to this particular regulatory apparatus.34

Another proposal is to mandate in

ormation sharing between competitors regulated networks.35 In addition to the obvious antitrust issues presented by requiring competitors to share competitively sensitive in — ormation, this proposal is sure to back — ire. Moreover, practical questions remain unresolved. How would such an in — ormation sharing network be en — orced? How would be the cost o — incentivizing competitors to share appropriate in — ormation? Who even knows what is the right in — ormation to share? This game is not worth the candle.

Tax incentives, however,

or startups might be worth exploring

urther. Tax rules and potential tax savings through them can be made available to small businesses and even individual proprietors. Thanks to technological solutions to tax problems, in — ormation about potential tax savings can be shared and even incorporated into business planning — airly easily. Tax breaks — or startups in their early years and — or small businesses throughout their li — etimes

See The Role o ---  Regulatory Sandboxes in Fintech Innovation, FINEXTRA, (Sept. 10, 2018), https://www. --- inextra.com/blogposting/15759/the-role-o --- - regulatory-sandboxes-in- --- intech-innovation        [https://perma.cc/Q28F-Y2SQ]; in --- ra Section V.A. See generally Hillary J. Allen, Regulatory Sandboxes, 87 GEO. WASH. L. REV. 579 (2019).
See The Role o ---  Regulatory Sandboxes in Fintech Innovation, supra note 30.
See Allen, supra note 30, at 589.
See id. at 588–89.
See id. at 641–42.
See in --- ra Section V.B.




 Electronic copy available at: https://ssrn.com/abstract=3435274

could o —

set the disparate impact o — regulatory costs and encourage entrepreneurial innovation.

In conclusion, this Article

inds that regulations can and should be designed to encourage entrepreneurial innovation, by — ocusing on making regulatory compliance equal in cost — or — irms large and small, young and old. On the horizon are new tools — or measuring the impact o — regulation. This Article theorizes that rules and standards will have a very di —

erent impact on regulation, and this variable should be considered in regulatory empirical analysis.

                        II. THE STARTUP STUDY

The discovery o

regulatory democratization arose through The Startup Study. The author o — this article spent two years conducting this — ieldwork study that was designed to generate and develop theories about the impact o — law and regulation on entrepreneurship and innovation. While most scholars seem to agree that the laws are not optimal — or entrepreneurship,36 there is almost nothing published on the impact o — regulation on startup innovation. The Startup Study was designed to categorically address this issue at its largest scale.

Setting aside conclusory and somewhat political claims either that regulations harm innovation and should be limited to ex post en — orcement,37 or that the — ree market is dangerous to society at large and should be controlled by the — ederal government, 38 there

36 See supra notes 8–10, 14–23 and accompanying text.
 37 See, e.g., Arch G. Woodside, Man-Ling Chang & Cheng-Feng Cheng, Government Regulations o ---  Business, Corruption, Re --- orms, and the Economic Growth o ---  Nations, 11 INT’L J. BUS. & ECON. 127, 128 (2012) (“The claim appears  --- requently in the United States that government regulations sti --- le business growth.”); see also, e.g., STEVE FORBES & ELIZABETH AMES, HOW CAPITALISM WILL SAVE US: WHY FREE PEOPLE AND FREE MARKETS ARE THE BEST ANSWER IN TODAY’S ECONOMY xi (2009) (“Government is good at maintaining order. But it lacks the imagination and creativity to produce the kind o ---  innovations that have always created jobs and driven genuine growth.”); Marc Bourreau & Pinar Doğan, Regulation and Innovation in the Telecommunications Industry, 25 TELECOMM. POL’Y 167, 168 (2001) (“Generally, regulation can a ---

ect these innovative activities via two di —

erent channels. Firstly, price regulations (or more speci — ically, the regulation o —

interconnection charges and retail prices) alter industry pro

its, hence the incentives to innovate. Secondly, both price and entry regulations change the terms o — entry, and hence innovation decisions regarding new entry.”); Stephan J. Goetz, Richard C. Ready & Brad Stone, U.S. Economic Growth vs. Environmental Conditions, 27 GROWTH & CHANGE 97, 97 (1996) (“Economists generally agree that environmental regulations reduce economic growth. Standard analysis reveals a shi — t in the supply curve to the le — t as — irms comply with regulations, leading to reduced output at higher prices[.]”). 38 See generally Joseph E. Stiglitz, Government Failure vs. Market Failure: Principles o — Regulation, in GOVERNMENT AND MARKETS: TOWARD A NEW THEORY OF REGULATION 13, 13, 15–16 (Edward J. Balleisen & David A. Moss

 Electronic copy available at: https://ssrn.com/abstract=3435274

are very — ew law review articles discussing the impact o —

regulations on innovation generally 39 or scienti

ically.40

For these reasons, the Author, and co-investigators Liya Palagashvili41 and Richard A. Epstein,42 proposed a research

eds., 2009) (“The notion that markets, by themselves, lead to e

icient outcomes has, in short, no theoretical justi — ication . . . .”). 39 Much o — the legal scholarship on regulation and innovation deals with speci — ic regulations. Indeed, a substantial portion o — it pertains to antitrust or patent law. For example, excellent papers have been written on the impact o —

mandatory unbundling o

telecommunications networks on innovation in incumbent local exchange carriers, see, e.g., Thomas M. Jorde, J. Gregory Sidak & David J. Teece, Innovation, Investment, and Unbundling, 17 YALE J. ON REG. 1 (2000), or the impact o — government power to enjoin mergers on innovation in business models, see, e.g., Rachel S. Tennis & Alexander Baier Schwab, Business Model Innovation and Antitrust Law, 29 YALE J. ON REG. 307, 308–10, 348–50 (2012) (recommending policy re — orm — or mergers between — irms with innovative business models). It is unsurprising that discussions about entrepreneurship should come up in the literature on competition law and policy, whose design is to — oster innovation. Likewise, the impact on innovation comes up in the patent context in large part because patent law is supposed to incentivize innovation by granting a temporary monopoly to the innovator. See Talya Ponchek, The Emergence o — the Innovative Entity: Is the Patent System Le — t Behind?, 16 J. MARSHALL REV. INTELL. PROP. L. 66, 71, 84, 112 (2016). However, regulations beyond those intended to directly a —

ect incentives to innovate o — ten have an indirect on entrepreneurial innovation. 40 The entrepreneurship literature is still developing with regard to the impact o — regulation on startup innovation. In this literature, the impact o —

regulation on entrepreneurship is o

ten accounted — or only indirectly by its composition within more general or only partially overlapping — actors such as industry uncertainty. While it is generally acknowledged that government, law, and regulation impact startup’s entry—the impact o — regulation on entrepreneurship—much o — the scholarly attention in the entrepreneurship literature has instead — ocused on intrinsic — actors such as the individual characteristics o — entrepreneurs—risk pre — erences—or non-regulatory industry

actors such as industry advertising intensity, industry density, and industry pro — itability. See Stephanie A. Fernhaber, Patricia P. McDougall & Benjamin M. Oviatt, Exploring the Role o — Industry Structure in New Venture Internationalization, 31 ENTREPRENEURSHIP THEORY & PRAC. 517 (2007) (providing a comprehensive literature review that de — ined nearly twenty industry structure variables—none o — which pertain to industry regulation); see, e.g., Thomas J. Dean & G. Dale Meyer, Industry Environments and New Venture Formations in U.S. Manu — acturing: A Conceptual and Empirical Analysis o —

Demand Determinants, 11 J. BUS. VENTURING 107 (1996) (industry structure’s impact on new venture — ormation); Patricia P. McDougall, Richard B. Robinson, Jr. & Angelo S. DeNisi, Modeling New Venture Per — ormance: An Analysis o —

New Venture Strategy, Industry Structure, and Venture Origin, 7 J. BUS. VENTURING 267 (1992) (industry structure’s impact on new venture per — ormance); Patricia Phillips McDougal et al., The E —

ects o — Industry Growth and Strategic Breadth on New Venture Per — ormance and Strategy Content, 15 STRATEGIC MGMT. J. 537 (1994) (industry structure’s impact on new venture strategic behavior); William R. Sandberg & Charles W. Ho — er, Improving New Venture Per — ormance: The Role o — Strategy, Industry Structure, and the Entrepreneur, 2 J. OF BUS. VENTURING 5, 5–28 (1987) (industry structure’s impact on new venture per — ormance). 41 Liya Palagashvili, Assistant Pro — essor o — Economics, State University o —

New York-Purchase; Research Fellow and Program A

iliate Scholar, The Classical Liberal Institute at New York University. 42 Richard A. Epstein, Laurence A. Tisch Pro — essor o — Law and Director o —

the Classical Liberal Institute at New York University, James Parker Hall

 Electronic copy available at: https://ssrn.com/abstract=3435274

project called The Startup Study: The Impact o — Regulation on Innovation. The researchers received a grant — rom the John Templeton Foundation to interview — ounders, — unders, and other startup market participants about their experiences with regulations; create qualitative datasets about our — indings — rom these interviews; publish scholarship based on the analysis o — our

indings; develop and issue a quantitative survey based on our qualitative — indings; and analyze and publish articles on both the qualitative — indings and the quantitative results — rom the survey. The Startup Study thus seeks to — ill voids in the law and entrepreneurship literature through its combination o — theoretical and qualitative approaches. This Article re — lects the primary legal

indings — rom the qualitative phase o — The Startup Study.

The Startup Study began with a qualitative phase in which the main goal was theory development. To achieve this research goal, the researchers selected unstructured interviews as a research tool. The use o — unstructured interviews — or the purpose o — theory development is well-established in the anthropology and sociology literatures.43 Fieldwork appears to be much less common in the legal literature; however, there are precedents — or the use o —

unstructured interviews

or qualitative analysis in legal scholarship as well.44

The unstructured interview technique is a well-established means o — understanding the complex behavior o — people without imposing any a priori categorization.45 This approach allowed The Startup Study to begin with a broad scope o — inquiry designed to maximize potential — or theory development. Fieldwork researchers — ollowed best practices to promote consistency across the interview sessions,

Pro

essor o — Law Emeritus and Senior Lecturer at the University o — Chicago, and Peter and Kirsten Bed — ord Senior Fellow at the Hoover Institution. 43 See, e.g., Yan Zhang & Barbara M. Wildemuth, Unstructured Interviews, in APPLICATIONS OF SOCIAL RESEARCH METHODS TO QUESTIONS IN INFORMATION AND LIBRARY SCIENCE 222, 223–24 (Barbara M. Wildemuth ed., 2009) (“Unstructured interviews can be very use — ul in studies o — people’s in — ormation seeking and use. They are especially use — ul — or studies attempting to — ind patterns, generate models, and in — orm in — ormation system design and implementation.” “[T]he purpose o — inquiry is theory development rather than theory testing.”). 44 See, e.g., Stewart Macaulay, Lawyers and Consumer Protection Laws, 14 L. & SOC’Y REV. 115 (1979) (conducting qualitative research through unstructured interviews o — approximately 100 lawyers — rom across — ive Wisconsin counties to learn about the impact o — the Magnuson-Moss Warranty Act on the practice o — law, 15 U.S.C. §§ 2301–2312 (2018)); Jean Braucher, Lawyers and Consumer Bankruptcy: One Code, Many Cultures, 67 AM. BANKR. L.J. 501 (1993) (conducting qualitative research through unstructured interviews o — consumer debt lawyers, chapter 13 trustees, and other participants in the bankruptcy system). 45 KEITH F. PUNCH, INTRODUCTION TO SOCIAL RESEARCH: QUANTITATIVE AND QUALITATIVE APPROACHES 185–86 (1998).

 Electronic copy available at: https://ssrn.com/abstract=3435274

including the establishment o — aide memoire by setting an agenda o — topics that might be covered in the interviews. 46

From May 2017 to December 2017,

ieldworkers Dr. Liya Palagashvili and Seth C. Oranburg traveled across the United States 47 and interviewed eighty-eight participants in the domestic startup ecosystem. O — these, — orty- — ive were with startups, twelve were with investors, and twenty-one were with other startup market participants. In addition, these researches also traveled internationally and interviewed startups in Israel and London.48

Subjects were chosen via

our methods: geographical sampling, 49 judgment sampling,50 opportunity sampling,51 and snowball

See generally ROBERT G. BURGESS, IN THE FIELD: AN INTRODUCTION TO FIELD RESEARCH 108 (Martin Bulmer ed., 1984); Charles Briggs, Interview, 9 J. LINGUISTIC ANTHROPOLOGY 137, 137–38 (2000); Terence McCann & Eileen Clark, Using Unstructured Interviews with Participants Who Have Schizophrenia, 13 NURSE RESEARCHER 7, 11–13 (2005).
To provide a national scope  --- or The Startup Study, interviews were conducted in Austin, Texas; Boston, Massachusetts; the Denver-Boulder Metro Area, Colorado; Los Angeles, Cali --- ornia; the Miami Metro Area, Florida; New York City, New York; Omaha, Nebraska; Pittsburgh, Pennsylvania; the San Diego Metro Area, Cali --- ornia; and the Silicon Valley-San Francisco Metro Area, Cali --- ornia.
International interviews included: 11 in Tel Aviv; 8 in Jerusalem; and 2 in other regions o ---  Israel; and 11 in London, UK.
Researchers created a list o ---  startup hub cities based on multiple lists o ---

top cities

or startups published by institutions such as the Kau —

man Foundation

or Entrepreneurship and trade publications such as Inc. Magazine, Entrepreneur, etc. See ROBERT FAIRLIE, ARNOBIO MORELIX & INARA TAREQUE, EWING MARION KAUFFMAN FOUND., 2017 KAUFMAN INDEX OF STARTUP ACTIVITY: METROPOLITAN AREA AND CITY TRENDS 9–13 (2017), https://www.kau —

man.org/wp- content/uploads/2019/09/2017_Kau —

man_Index_Startup_Activity_Metro_Repo rt_Final.pd — [https://perma.cc/Q6U4-H8VN]; Emily Canal, The 10 Hottest Startup Cities in America, INC (Aug. 14, 2019), https://www.inc.com/emily- canal/top-10-cities-success — ul-businesses-2019-inc5000.html [https://perma.cc/7EEP-3GRY]; Angela Ruth, The Top 7 Cities Competing with Silicon Valley — or Tech Entrepreneurs, ENTREPRENEUR (Aug. 24, 2017), https://www.entrepreneur.com/article/299173 [https://perma.cc/F4XJ-727X]. Researchers then used CrunchBase Pro—a proprietary database o — startups that contains details including the names and emails o —

ounders and executives—to search — or startups in those cities that met the chosen criteria. See generally CRUNCHBASE, https://www.crunchbase.com [https://perma.cc/7VX6-D46H]. Criteria included: privately held, technology startups, less than seven years old, that had received outside — unding but had not passed the Series C stage. From this list, researchers and their sta —

cold called, emailed, and sent LinkedIn messages to these executives and — ounders. They then described The Startup Study and requested a live interview on speci — ic dates when the — ieldworkers would be in that city. I — these initial e —

orts did not generate a response, the researchers leveraged personal connections and networking to secure interviews. 50 Researchers sought out speci — ic — irms in targeted cities that were well- known — or being in — luential in the startup marketplace. Judgement sampling led researchers to seek out interviews with venture capital — irms, accelerators, and incubators that are known to be key players in local and national startup ecosystems. 51 Researchers spent a considerable amount o — time in each city, where they

 Electronic copy available at: https://ssrn.com/abstract=3435274

sampling.52 Researchers contacted these subjects via phone, email, LinkedIn, Twitter, Facebook, and other social media plat — orms and requested a one-hour in-person interview. Most interviews were indeed conducted on-site and in-person, although a — ew were conducted by con — erence call or telecon — erence plat — orms, such as Zoom or Skype.

Almost all the interviews were recorded

or — urther analysis, except in a — ew cases where the interviewee requested that the interview not be recorded. A — ter each interview, a research assistant read the researcher’s interview notes, listened to the recording,53 and dra — ted a memorandum that identi — ied the key issues, the topics addressed, and details that were discussed. The research assistant also provided some suggestions — or — urther investigation. The researchers than reviewed each o — the memoranda and compiled them into — ive distinct datasets: GenTech, 54 FinTech,55 MedTech,56 Israel, and London.

these events, the researchers met — ounders and executives o — startups who were not initially identi — ied through the geographic sampling, but who expressed knowledge about the impact o — regulation on startups. Researchers took these — ortunate meetings as opportunities to conduct additional interviews. 52 A — ter some interviews, interviewees o —

ered to connect the researchers with other potential interviewees. Prior to accepting this invitation, researchers determined whether these proposed connections would make a valuable contribution to The Startup Study. I — so, researchers — ollowed up with a subsequent interview based on this snowball e —

ect. 53 In the very — ew cases where a recording was not available, notes taken during the interview were later expanded and re — ined by the interviewer. 54 GenTech—as used in this article—re — ers to general technology — irms, which includes most technology startups—so — tware, IT, database, B2B, media, social networks, arti — icial intelligence, etc. A key criterion — or GenTech is the absence o — any highly speci — ic or especially onerous regulations. GenTech startups and small businesses are still subject to general regulations such as employment, immigration, patent, tax, etc. See Daniel McKenzie, Startup Law A to Z: Regulatory Compliance, EXTRA CRUNCH (Apr. 4, 2019, 12:17 PM), https://techcrunch.com/2019/04/04/startup-law-a-to-z-regulatory-compliance/ [https://perma.cc/A86J-PW5X]. 55 FinTech re — erences — inancial technology — irms, which include blockchain, crowd — unding, cryptocurrency, insurance, micro — inance, payment systems, and robo-trading companies. See Julia Kagain, Financial Technology – Fintech, INVESTOPEDIA (June 25, 2019), https://www.investopedia.com/terms/ — / — intech.asp [https://perma.cc/D5RA- 4SL5]. FinTech — irms generally have to deal with additional regulations — rom the Financial Industry Regulatory Authority (FINRA), the Securities and Exchange Commission (SEC), and state — inancial regulators. See id. 56 MedTech re — ers to — irms that require FDA approval prior to selling products in the US market. See Scott Gottlieb, Advancing Policies to Promote Sa — e, E —

etive MedTech Innovation, FDA (Dec. 11, 2018), https://www. — da.gov/news- events/ — da-voices/advancing-policies-promote-sa — e-e —

ective-medtech- innovation [https://perma.cc/FL7S-FN7T]; Kevin O’Kee —

e, The Rise o — Medtech,

 Electronic copy available at: https://ssrn.com/abstract=3435274

To encourage the interviewees to speak — rankly about their experiences, the researchers agreed to maintain their con — identiality. Accordingly, the people interviewed are not identi — ied in this Article. However, a common thread that wove throughout their stories was their use o — technology to comply with certain regulations. But — or that technology, competition in that regulated industry would be impossible. This Article will now

ocus on that technology, which it terms “regulatory democratizations.”

               III. REGULATORY DEMOCRATIZATIONS

Discovering the phenomenon o

regulatory democratizations was a primary — inding o — The Startup Study. This Section will — irst de — ine and discuss regulatory democratization (RD) in general. Then it will present case studies o — companies engaging in RD as speci — ic illustrations o — the general concept.

RDs are technological solutions

or regulatory compliance problems. RD only arises in regulated industries where technology is able to make regulatory compliance cheaper — or a large number o —

irms. In other words, companies will develop RD products when they can sell scalable compliance solutions, and companies will purchase RD technologies when they can make regulatory compliance more a —

ordable.

RD seems to depend on new technologies that have only recently become available. The startups described below are powered by virtual servers, arti — icial intelligence, and other cutting-edge technologies. In particular, the general availability o — cloud computing power seems essential — or RD. The common thread binding these necessary technologies together is that they all make it easier to enter new markets and scale up production. For

MDDI (June 10, 2011), https://www.mddionline.com/rise-medtech [https://perma.cc/HDP9-M33F]. This includes Class II medical devices and Class III pharmaceuticals. See O’Kee —

e, supra; Classi — y Your Medical Device, FDA, https://www. — da.gov/medical-devices/overview-device-regulation/classi — y-your- medical-device [https://perma.cc/3BFP-5ZWP]. It does not include “digital health” companies like Fitbit who make devices that relate to health but do not claim to treat any particular disease or medical condition. See FDA Says It Won’t Regulate FitBit, Many Other Fitness Wearables, ADVISORY BOARD (Aug. 18, 2016, 8:17 AM), https://www.advisory.com/daily-brie — ing/2016/08/18/ — da-says- it-wont-regulate- — itbit [https://perma.cc/A7EQ-KTAX]. See generally CTR. FOR DEVICES & RADIOLOGICAL HEALTH, U.S. FOOD & DRUG ADMIN., GENERAL WELLNESS: POLICY FOR LOW RISK DEVICES (2019), https://www. — da.gov/media/90652/download [https://perma.cc/V8MS-55JW].

 Electronic copy available at: https://ssrn.com/abstract=3435274

example, cloud computing makes it easy to “rent” instead o — “buy” processing power.57

RD companies use this processing power to

urther make it cheaper — or small — irms to “rent” compliance solutions, instead o —

“buying” complex and customized solutions

or each — irm. Lowering the “rental” cost o — regulatory compliance makes compliance a —

ordable — or more young and small companies. RD thus mitigates some o — the disparate impact o — regulatory burdens

aced by entrepreneurs. In this way, RD can have a dramatic e —

ect on regulated markets by — acilitating entry and competition—so long as underlying regulations are computable.

RD is distinct

rom traditional regulatory consulting services. In the traditional model, a consulting — irm provides customized pro — essional advice to a single individual or an organization — or a

ee.58 Consultancy may include advice on strategy, operations, management, taxation, human resources, etc. 59 Traditional regulatory consulting is the analogue to the digital solution o — RD.

A regulatory consulting

irm provides customized advice as to how an individual or — irm can navigate a given regulation. Further, these consulting — irms have a distinct comparative advantage 60 in assessing, analyzing, and digesting regulatory in — ormation. Due to the principle o — microeconomic specialization, 61 they are better

57 See Steve Ranger, What is Cloud Computing? Everything You Need to Know About the Cloud, Explained, ZDNET (Dec. 13, 2018, 12:24 PM), https://www.zdnet.com/article/what-is-cloud-computing-everything-you-need- to-know- — rom-public-and-private-cloud-to-so — tware-as-a/ , [https://perma.cc/B5LQ-6TH8]. 58 See, e.g., ACCESTRA, https://www.accestra.com [https://perma.cc/6WYH- FML3]; Regulatory A —

airs, BEAUFORT, https://www.beau — ortcro.com/regulatory-a —

airs [https://perma.cc/Y7EC-Y9KT]. 59 See Jenni — er Wilson, Consulting Services Can Grow Your Firm—i —

Mangaged Properly, J. ACCT. (Mar. 7, 2016), https://www.journalo — accountancy.com/newsletters/2016/mar/consulting- services-can-grow- — irm.html [https://perma.cc/XWF2-89C3]. 60 David Ricardo expanded on Adam Smith’s theory o — absolute advantage. See DAVID RICARDO, ON THE PRINCIPLES OF POLITICAL ECONOMY, AND TAXATION 66, 388–402 (3d ed. 1821). Ricardo argued that even i — a country has an absolute advantage in producing both goods, the other may have a comparative advantage and should produce the good associated with the lower opportunity cost to increase economic wel — are. Id. 61 Firms that are better suited to analyze regulation regimes have an endogenous comparative advantage because o — internal — actors such as training, resources, and investments in equipment. These — irms can sell their services to other — irms, which results in an e —

icient division o — labor and microeconomic specialization. Sherwin Rosen, Substitution and Division o — Labour, 45 ECONOMICA 235, 235 (1978) (“[T]he packaging o — work activities into bundles is itsel — the endogenous outcome o — economic decisions.”).

 Electronic copy available at: https://ssrn.com/abstract=3435274

suited to distill this in — ormation and tailor legal strategies — or other

irms, rather than a — irm trying to navigate the applicable regulation regime in-house.

For example, NAMSA is a medical research organization that helps — irms comply with FDA regulations.62 An organization can hire NAMSA to provide a risk assessment o — whether a new medical device is likely to be considered a biological risk under,

or example, ISO 10993-1 or the EU Medical Device Directive. 63 For these services, a — irm like NAMSA might charge up to $1,000 per hour.64

Due to such high

ees, many startups and small businesses, especially ones that “run lean,”65 cannot a —

ord to hire traditional regulatory consultants. With the option o — compliance o —

the table, startups and small businesses are le — t with two options: proceed with the development o — a product that might not comply with regulations or abandon the project completely. Accordingly, some startups begin their li — e in will — ul ignorance or even deliberate violation o — regulatory requirements, while other small businesses are stillborn in regulatory limbo.

RD solves the dilemma o

una —

ordable noncompliance by o —

ering a way — or startups and small businesses to comply with regulations a —

ordably. Unlike a traditional consulting — irm, which o —

ers customized expertise to individual organizations at a relatively high price,66 an RD — irm o —

ers a technological solution that can be employed by a large number o — startups — or a relatively low price. To accomplish this, RD — irms’ business models rely on economies o — scale.67

  See     Our    History:     From     CRO      to    MRO,     NAMSA, https://www.namsa.com/our-history [https://perma.cc/B9WF-B8WQ].
See Medical Device Regulatory Consulting, NAMSA, https://www.namsa.com/services/consulting/regulatory [https://perma.cc/XC4Q- V2K7]; Biological Sa --- ety Services, NAMSA, https://www.namsa.com/services/testing/biological-sa --- ety-services [https://perma.cc/WXJ6-8TQJ]; EU MDR & IVDR Planning Resources, NAMSA, https://www.namsa.com/mdr-ivdr-resources [https://perma.cc/79YZ- N4WX].
Hanh Nguyen, How Much Should You Pay  --- or a Compliance Consultant?, LINKEDIN (Aug. 31, 2018), https://www.linkedin.com/pulse/how-much-should- you-pay-compliance-consultant-hanh-nguyen/ [https://perma.cc/5GJM-SRWK].
For a discussion on the lean startup model, see Steve Blank, Why the Lean Start-Up Changes Everything, HARV. BUS. REV. (May 2013), https://hbr.org/2013/05/why-the-lean-start-up-changes-everything [https://perma.cc/VL2E-V2YY].
See supra notes 58–64 and accompanying text.
The concept o ---  economies o ---  scale dates back as  --- ar as Adam Smith, who




 Electronic copy available at: https://ssrn.com/abstract=3435274

RD is thus distinct — rom the traditional regulatory consultancy. Traditionally, a consulting — irm prepares a customized regulatory solution — or a unique client at a high cost. 68 A customized solution may include expensive long-term political investments, such as lobbying and hobnobbing with regulators. 69 Accordingly, only large and longstanding — irms could a —

ord regulatory consultancy. Now, under RD, technologies o —

er scalable solutions to regulatory problems — or a large number o —

irms. Unlike traditional consulting — irms, which are generally too expensive — or small — irms to employ, RD — irms o —

er scalable regulatory compliance solutions that are a —

ordable and sometimes even — ree — or the smallest — irms. Thus, RD makes it easier — or small and young

irms to compete with large and old ones.

Technological solutions that work

or a wide range o — startups and small businesses are not easy to create. I — they were, then traditional consulting — irms would be obsolete. Yet the traditional consulting industry is clearly necessary, as evidenced by the — act that it generates about $506 billion in annual revenues.70 However, technology, especially cloud computing and arti — icial intelligence (AI), is making technological RD solutions more — easible.

RD not only changes how much regulatory costs are incurred by small business. RD also changes when those costs are incurred. By trans — orming up- — ront costs to pay-as-you-go costs, RD makes it much easier to run a cash-strapped small business in a highly regulated environment.

Traditional regulatory costs—such as paying a consultant to determine whether a proposed product would be in compliance

explained how the division o

labor can result in lower costs o — production. See Reem Heakal, What Are Economies o — Scale? INVESTOPEDIA (Aug. 14, 2019), https://www.investopedia.com/insights/what-are-economies-o — -scale [https://perma.cc/22NN-HY3E]. A simple example o — economies o — scale in ordinary li — e is the lower costs a consumer incurs when buying in bulk. Id. In general, as a good or service becomes more standardized, or “commoditized,” that commodity becomes less expensive. See id. 68 See Soren Kaplan, The Business Consulting Industry Is Booming, and It’s About to Be Disrupted, INC. (Sept. 11, 2017), https://www.inc.com/soren- kaplan/the-business-consulting-industry-is-booming-and-it.html [https://perma.cc/C4AE-H5R7]. 69 See, e.g., Public Policy & Regulation, HOLLAND & KNIGHT, https://www.hklaw.com/en/services/practices/regulatory-and-government- a —

airs/public-policy-regulation [https://perma.cc/7FG7-FLT7]. 70 Consulting Industry Market Research, PLUNKETT RES., https://www.plunkettresearch.com/industries/consulting-market-research [https://perma.cc/32EE-GJHF].

 Electronic copy available at: https://ssrn.com/abstract=3435274

with a given regulation—are incurred up — ront.71 Costs — or FDA pharmaceutical trials, — or example, are incurred be — ore the — irst pill is sold.72 To a —

ord the massive up- — ront costs, many pharmaceutical companies try to raise billions o — dollars via an initial public o —

ering in order to — inance drug trials. 73

Startups and small businesses that lack access to capital markets

ind it very di —

icult to pay such huge up- — ront costs.74 Entrepreneurs may decide not to enter a market where the initial price o — admission is too high.

Yet, RD provides an alternative to raising capital to “buy” compliance at a high up- — ront cost. RD can trans — orm a high initial cost into a lower but ongoing cost. In other words, RD trans — orms up- — ront, — ixed regulatory costs into ongoing, variable regulatory costs. Such technologies can make it possible — or smaller and younger companies to compete with larger and older ones. To put this in terms o — economics, RD technology — lattens the average

See Consulting Fees and Rates: How Much Should I Charge?, CONSULTING,                   https://www.consulting.com/consulting- --- ees-rates [https://perma.cc/9S3A-N3LW].
See Matthew Herper, The Cost o ---  Developing Drugs is Insane. That Paper That Says Otherwise is Insanely Bad, FORBES (Oct. 16, 2017, 10:58 AM), https://www. --- orbes.com/sites/matthewherper/2017/10/16/the-cost-o --- - developing-drugs-is-insane-a-paper-that-argued-otherwise-was-insanely- bad/#23cecea82d45 [https://perma.cc/W2RN-QYAT].
See generally DAVID THOMAS & CHAD WESSEL, BIO, 2019 EMERGING

THERAPEUTIC COMPANY TREND REPORT (2019), http://go.bio.org/rs/490- EHZ- 999/images/BIO%202019%20Emerging%20Company%20Trend% 20Report.pd — [https://perma.cc/9679-TPGM]. It is well-known in the industry that pharmaceutical companies are highly dependent on access to capital. For early-stage private companies, the majority o — this investment comes in the — orm o — venture capital until the eventual listing on a public exchange. This initial public o —

ering is the — irst o —

      what can be many rounds o ---

inancing — rom public investors through — ollow-on public o —

erings, — inancings that can provide timely access to capital a — ter key clinical or regulatory milestones. Id. at 4. 74 See Michael R. Wade & Jialu Shan, The Battle — or Digital Disruption: Startups v. Incumbents, IMD (Mar. 2016), https://www.imd.org/research- knowledge/articles/the-battle- — or-digital-disruption-startups-vs-incumbents [https://perma.cc/ZX5W-KNZ6]. There are at least two reasons why — ixed costs have a disparate impact on startups versus incumbents. See id. First, the startup generally needs to pay that cost with debt or equity, whereas an incumbent also has the option o — paying via cash — lows. See id. Second, startups have more uncertainty as to whether they will succeed in a new market, as compared with incumbents who have greater certainty as to whether they will continue to operate in an existing market. The uncertainty o — return on investment — or startups adds to their cost. See id.

 Electronic copy available at: https://ssrn.com/abstract=3435274

cost curve o — regulatory compliance. This makes a more level playing — ield — or competition between large and small — irms in highly regulated markets.

Figure 1. Regulatory democratization “

lattens” the average cost curve o — regulatory compliance. Without regulatory democratization, this curve has economies o — scale that advantage larger — irms, who spread the cost o — compliance over a greater quantity o — goods or services sold, over smaller — irms.

RD is not entirely new. A

ter all, online — orms have helped entrepreneurs save legal costs — or decades, and hornbooks also democratize access to the legal process.75 One might even argue that the printing press is a technological innovation that enabled a sort o —

i — teenth century RD.76

But this Article argues that the use o

cloud computing and arti — icial intelligence technologies has enabled RD to make a higher degree o — impact on competition in highly regulated industries.

Next, this Part will introduce

ive case studies o — startups that o —

er RD technologies—Paubox, AirMap, Avalara, Cognigo, and Metomic. In addition, Amazon Web Services provides compliance technologies — or these and other regulations, and it — orms the hardware backbone — or many startups’ so — tware solutions,

75 See Daniel W. Linna, Jr., What We Know and Need to Know About Legal Startups, 67 S.C. L. REV. 389, 394–95, 410 (2016).
76 See generally Steven Kreis, The Printing Press, HIST. GUIDE (May 2, 2016), http://www.historyguide.org/intellect/press.html [https://perma.cc/78GJ- 2EL5].

 Electronic copy available at: https://ssrn.com/abstract=3435274

including RD.77 The startups described below were not necessarily interviewed by the researchers in The Startup Study. However, these startups came to the attention o — the researchers through the process o — theory development via unstructured interviews with startup market participants.

1. Paubox

Paubox is a startup that developed an RD solution

or doctors’ o —

ices.78 Doctors’ o —

ices must comply with many regulations including the Health Insurance Portability and Accountability Act (HIPAA).79 HIPAA regulates the storage and transmission o —

protected health in

ormation (PHI). 80 Essentially, HIPAA requires PHI to be sent in encrypted — ormats such that doctors’ o —

ices generally cannot email patient records without investing tens o —

thousands o

dollars to install and maintain their own secure email servers.81

Paubox was

ounded in 2015 to o —

er a solution whereby doctors’ o —

ice could send secure emails via a simple app called Paubox.82 Instead o — paying a huge up — ront cost — or private email servers, Paubox users pay a low monthly rate, based on their usage.83 This makes it much cheaper — or small doctors’ o —

ices to email PHI, helping small doctors’ o —

ices compete on the dimension o — patient

E.g., Hot AWS EdStart Startups: GlyphEd, Exploros, and WayUp, AMAZON            WEB           SERVICES         (May           1,       2018), https://aws.amazon.com/blogs/publicsector/hot-aws-edstart-startups-glyphed- exploros-and-wayup/ [https://perma.cc/N5SZ-BJX5].
    About      Us,       PAUBOX,        https://www.paubox.com/about [https://perma.cc/G9BL-7S96].
See Who Must Comply with HIPAA Privacy Standards, U.S. DEP’T HEALTH & HUM. SERVICES (Dec. 19, 2002), https://www.hhs.gov/hipaa/ --- or- pro --- essionals/ --- aq/190/who-must-comply-with-hipaa-privacy- standards/index.html [https://perma.cc/L7NH-PLH4].
See What is PHI? U.S. DEP’T HEALTH & HUM. SERVICES (Feb. 26, 2013), https://www.hhs.gov/answers/hipaa/what-is-phi/index.html [https://perma.cc/XT7A-KKMZ].
    HIPAA       Compliance         --- or      Email,        HIPAA     J., https://www.hipaajournal.com/hipaa-compliance- --- or-email [https://perma.cc/GW56-JAZQ]. For an inexhaustive list o ---  HIPAA email compliance costs, see Best HIPAA Compliant Email Encryption Services, TOTAL HIPAA COMPLIANCE (May 1, 2019), https://www.totalhipaa.com/hipaa- compliant-email-encryption-services [https://perma.cc/PZ57-CZEL].
See Paubox HIPAA Compliant Email Demo, PAUBOX (May 28, 2015), https://paubox.wistia.com/medias/d17vqybqog?wvideo=d17vqybqog [https://perma.cc/5QS8-DRFD].
    Paubox        Encrypted        Email        Pricing,      PAUBOX, https://www.paubox.com/solutions/encrypted-email/pricing [https://perma.cc/MK8E-PUTM].




 Electronic copy available at: https://ssrn.com/abstract=3435274

communication with large healthcare systems that can more easily a —

ord to build their own secure PHI transmission in — rastructure.

HIPAA is a typical example o

an anachronistic regulation. When HIPAA was passed in 1996, email was — ar less — requently used as compared to today.84 Facsimile, on the other hand, was generally considered a secure transmission method under HIPAA without taking other steps.85 As a result, many healthcare organizations relied on — ax transmission and eschewed email — or many years. 86

Perhaps this reliance on

ax transmission made sense in 1996, but the means o — communication have changed signi — icantly since then. For example, in 2000, approximately 12 billion emails were sent per day.87 In 2018, approximately 281 billion emails were sent per day,88 and estimates suggest that daily email tra —

ic will increase to 306 billion per day in 2020 and to 347 bilion per day in 2023.89 Meanwhile, many observers recognize that HIPAA is the

 84 See Jim Grubbs, E-mail and Instant Messaging, in 1 THE INTERNET ENCYCLOPEDIA 660, 661 (Hossein Bidgoli ed., 2004), http://www.encyclopedias.biz/dw/The%20Internet%20Encyclopedia,%20Volum e%201.pd ---  [https://perma.cc/4TGC-SA9L]. In 1996, when HIPAA was passed into law, an average o ---  about 300 million emails were sent every day. Id. at 661.
 85 See Shi --- ali Arora, Jenni --- er Yttri & Wendy Nilsen, Privacy and Security in Mobile Health (mHealth) Research, 36 ALCOHOL RES. 143, 144 (2014), https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4432854/pd --- /arcr-36-1-143.pd ---

[https://perma.cc/SY8X-ERRJ] (“At the time these rules were introduced, clinical health in — ormation existed primarily in the — orm o — handwritten patient health records. In — ormation generally was shared between care providers over the phone, by — ax or in person. Consequently, initial regulations and guidelines

ocused on the challenges surrounding protecting in — ormation in these limited- sharing — ormats.”). 86 Yvonne Li, The Slow Disappearance o — the Fax Machine in Healthcare, HEALTH IT EXCHANGE (Dec. 3, 2014, 1:31 PM), https://web.archive.org/web/20180317190600/https://searchhealthit.techtarget.co m/healthitexchange/CommunityBlog/the-slow-disappearance-o — -the- — ax- machine-in-healthcare [https://perma.cc/43WE-ZZ8B] (“Fax, short — or

acsimile, involves the transmission o — scanned printed material over phone lines, typically to a telephone number connected to a printer or other device. Although — ax reliance peaked in the 1980s, which is also when it took o —

in healthcare o —

ices, its invention dates all the way back to 1843.”). 87 Oliver J. Chiang, The Decade in Data, FORBES (Dec. 28, 2009, 6:00 AM), https://www. — orbes.com/2009/12/27/broadband-text-messages- technology-cio-network-data.html#4324c86b5 — c3 [https://perma.cc/5AC6- R7TP]. 88 THE RADICATI GROUP, INC., EMAIL STATISTICS REPORT, 2018-2022 EXECUTIVE SUMMARY 3 (2018), https://www.radicati.com/wp/wp- content/uploads/2018/01/Email_Statistics_Report,_2018- 2022_Executive_Summary.pd — [https://perma.cc/V87M-EPXP]. 89 Id.; Jessica Clement, Number o — E-mail Users Worldwide 2017-2024, STATISTA (Mar. 25, 2020), https://www.statista.com/statistics/255080/number- o — -e-mail-users-worldwide [https://perma.cc/AHK8-UF2V].

 Electronic copy available at: https://ssrn.com/abstract=3435274

reason why the healthcare industry continues to rely on archaic — ax technology.90

Part o

the reason why healthcare was stuck in the technological stone age was the sheer cost o — implementing modern communications solutions that are HIPAA compliant. HIPAA compliance costs create signi — icant overhead, including cyber liability insurance,91 technical errors and omissions insurance, 92 HIPAA training — or new employees and ongoing HIPAA training,93 HIPAA audits,94 encryption — or laptops and other data- related processes,95 and other costs.

These costs can be insurmountable

or small doctors’ o —

ices, which is precisely why Paubox’s solution is designed — or and adopted by these small businesses. These small doctors’ o —

ices appreciate that Paubox’s encrypted email is a plug-and-play solution.96 Small businesses may not have the technical sta —

or computing in — rastructure — or complex adoption, but Paubox makes it easy — or small doctors’ o —

ices to send PHI via email while complying with HIPAA.97 In a world where the — ax machine is a dying breed o — communication device, Paubox helps small doctors’ o —

ices be more competitive with large — irms.

Sophie Haigney, The Fax Is Not Yet Obsolete, ATLANTIC (Nov. 18, 2018), https://www.theatlantic.com/technology/archive/2018/11/why-people- still-use- --- ax-machines/576070 [https://perma.cc/L2VA-VG93].
Cyber Liability: Complying with HIPAA Regulations, JAMISON (Dec. 31, 2015, 4:03 PM), https://www.jamisongroup.com/blog/cyber-liability-complying- with-hipaa-regulations.aspx [https://perma.cc/H8YS-WXCG].
Have Clients in Healthcare? Understand the Pro --- essional Liability Exposures, INSUREON (Jan. 12, 2015), https://www.insureon.com/blog/have- clients-in-healthcare-understand-the-pro --- essional-liability-exposures [https://perma.cc/K349-WG7T].
     HIPAA         Training       Requirements,        HIPAA      J., https://www.hipaajournal.com/hipaa-training-requirements [https://perma.cc/8FQA-E9CS].
See Perry Price, The Cost o ---  a HIPAA Audit, REVATION SYSTEMS (June 1, 2019), https://revation.com/hipaa-audit-costs [https://perma.cc/Q8FF-PASV].
 Mobile Data Security and HIPAA Compliance, HIPAA J., https://www.hipaajournal.com/mobile-data-security-and-hipaa-compliance/ [https://perma.cc/H876-VZF3].
See Customer Stories, PAUBOX, https://www.paubox.com/customers [https://perma.cc/PZY2-5CGN]; Rick Kuwahara, Paubox Named a High Per --- ormer in Three Categories in G2 Crowd’s December 2018 Report, PAUBOX (Dec.      17,    2018),     https://www.paubox.com/blog/paubox-named-a-high- per --- ormer-in-three-categories-in-g2-crowds-december-2018-report [https://perma.cc/W4AC-HN3F].
See Arianna Etemadiah, How to Encrypt Your Email and Why You Should, PAUBOX (Sept. 18, 2017), https://www.paubox.com/blog/how-to-encrypt-email [https://perma.cc/6Q66-7SW2].




 Electronic copy available at: https://ssrn.com/abstract=3435274

Interestingly, Paubox relies on AWS98 to provide its HIPAA- compliant email service. 99 AWS is Payment Card Industry (PCI), Sarbanes-Oxley Act (SOX), and HIPAA compliant. 100 So long as AWS/EC2 users write code that does not break compliance, users can scale their business using AWS cloud services. 101 This is an example o — how AWS makes it possible — or Schumpeterian entrepreneurs—innovators who disrupt existing industries and thereby change market equilibria,102 like Hoala Greevy, — ounder and CEO o — Paubox103—and startups, like Paubox, to compete with large incumbent — irms. Paubox would not be able to o —

er a competitive HIPAA compliant email solution i — it had to bear the cost o — buying its own web servers.

2. AirMap

AirMap is a startup whose goal is to help make drones part o

everyday li

e by helping drone operators navigate the regulations

or low-altitude operation o — unmanned aerial vehicles (UAVs).104 This requires navigating a complex and dynamic web o —

regulations. The Federal Aviation Administration (FAA) creates reports that are supposed to explain how to comply with the regulations on UAVs,105 but, through our study, we learned that the rule-making committees do not appreciate dissent — rom the consensus opinion within the industry. Generally, the committee

See in --- ra Section III.6.
Evan Fitzgerald, Paubox Joins Amazon’s AWS Partner Network (APN), PAUBOX (June 7, 2017), https://www.paubox.com/blog/paubox-amazon-aws- partner-network [https://perma.cc/WEG7-6RHG].
AMAZON WEB SERVS., AMAZON WEB SERVICES: RISK AND COMPLIANCE 5                                                                          (2017), https://d0.awsstatic.com/whitepapers/compliance/AWS_Risk_and_Compliance_ Whitepaper.pd ---  [perma.cc/7RKP-X7JQ].
  See    Amazon      EC2     FAQs,     AMAZON WEB SERVICES, https://aws.amazon.com/ec2/ --- aqs [https://perma.cc/X9AC-5VZY]. “EC2” re --- ers to Amazon Elastic Compute Cloud, a “web service that provides secure, resizable compute capacity in the cloud.” Id.
See JOSEPH A. SCHUMPETER, CAPITALISM, SOCIALISM, AND DEMOCRACY 71–75 (Routledge 2010) (1943). Joseph Schumpeter introduced the concept o ---

entrepreneurship in conjunction with the idea o

creative destruction in a chapter titled “The Process o — Creative Destruction” in his seminal monograph. See id. 103 About Us, supra note 78. 104 Lora Kolodny, AirMap Raises $26 Million to Manage Air Tra —

ic as Drone Use Surges, TECHCRUNCH (Feb. 23, 2017, 4:33 AM), https://techcrunch.com/2017/02/23/airmap-raises-26-million-to-manage-air- tra —

ic-as-drone-use-surges/ [https://perma.cc/7FY8-484Y]. 105 See, e.g., MICRO UNMANNED AIRCRAFT SYS. AVIATION RULEMAKING COMM., ARC RECOMMENDATIONS FINAL REPORT (2016), https://www. — aa.gov/uas/resources/public_records/media/Micro-UAS-ARC- FINAL-Report.pd — [https://perma.cc/8KSE-F2NT] (recommending a regulatory

ramework — or the FAA in the classi — ication and operation o — UAVs).

 Electronic copy available at: https://ssrn.com/abstract=3435274

only listens to major players such as Boeing, which means their reports and recommendations do not re — lect the needs and use cases o — the UAV community at large. 106

Since access to the FAA rulemaking process is dominated by a

ew large incumbent players who do not want the rules to change and do not want innovative startup competitors to be able to try new business models, disruption o — the aviation industry by small startups seemed impossible. AirMap, however, provides these small startups with a plat — orm that it terms “collaborate acceleration,” whereby startups may collectively engage in educating a regulatory agency and encourage them to try a more innovative approach.107 AirMap thus engages in a unique sort o —

regulatory democratization by making it easier

or small startups to in — luence agency decision-making.

AirMap also builds so

tware that re — lects the more common sort o —

regulatory democratization. Their Discover

unction allows users to view airspace advisories, airspace requirements, and even weather conditions be — ore — light. 108 “The AirMap Plat — orm includes a Tile Map Service that allows developers to visually display airspace and advisories to pilots as an overlay.”109 They also have a — unction that allows users to go mobile with airspace mapping via iOS or Android.110 According to our interviews, it would be prohibitively expensive — or individual drone operators to obtain this in — ormation. AirMap lowers the relative cost o —

compliance

or small startups by showing all pilot advisories and displays o — the airspace, 111 whereas the large — ixed costs in

 106 See Natalie Kitroe ---

& David Gelles, Be — ore Deadly Crashes Boeing Pushed — or Law that Undercut Oversight, N.Y. TIMES (Oct. 27, 2019), https://www.nytimes.com/2019/10/27/business/boeing-737-max-crashes.html [https://perma.cc/KW7Q-7M4V]; Brian Naylor, Boeing’s Not Alone in Companies that Government Agencies Have Let Sel — -Regulate, NPR (Apr. 2, 2019 5:36 PM), https://www.npr.org/2019/04/02/709203191/boeings-not-alone-in- companies-that-government-agencies-have-let-sel — -regulate [https://perma.cc/4CSF-7G2Q]. 107 See, e.g., Gary Mortimer, FlytBase and AirMap Agree to Collaborate on Accelerating UTM Support — or Drone Automation, SUAS NEWS (May 9, 2019), https://www.suasnews.com/2019/05/ — lytbase-and-airmap-agree-to-collaborate- on-accelerating-utm-support- — or-drone-automation/ [https://perma.cc/F42J- 9HWC]. 108 Developers, AIRMAP, https://www.airmap.com/developers [https://perma.cc/TH4E-X8RL]; Discover, Connect, and Now Fly Your DJI Drone with AirMap, AIRMAP, (Apr. 3, 2018), https://www.airmap.com/airmap-

or-drones-app- — ly-dji/ [https://perma.cc/AY6X-5K4U]. 109 Overview, AIRMAP, https://developers.airmap.com/docs/maps-overview [https://perma.cc/MRJ7-VR8K]. 110 Id. 111 Developers, supra note 108.

 Electronic copy available at: https://ssrn.com/abstract=3435274

developing this custom so — tware is astronomical and impossible — or small — irms to a —

ord.

“AirMap’s Airspace Advisory API takes in a geometry (i.e.

light area) and a ruleset identi — ier, returning all the relevant advisories that intersect with that — light area.”112 Regulatory issues with certain airspaces such as the type o — aircra — t allowed in a speci — ic zone, height restrictions, etc., can be monitored, — acilitating a sa — er and regulatory compliant — light path. 113 Without AirMap, mapping a sa — e and compliant — light path could be prohibitively expensive.114 Hiring someone who knows the rules and regulations — or aircra — t and di —

erentiating between di —

erent — light zones would be expensive—the so — tware clearly o —

sets that cost and makes — requent UAV — lights possible — or a wider range o —

operators.115

AirMap provides additional tools that UAV operators would otherwise have to develop and implement themselves, at a high initial — ixed cost. For example, AirMap connects with DroneLogbook to provide a simple and convenient tool — or pro — essional and private drone pilots around the world to log and save drone activities data such as — lights, drones used, and places traveled, in addition to equipment and maintenance data. 116 AirMap’s

Tra ---

ic Alert API allows developers to provide their pilots with alerts about nearby tra —

ic—including commercial airplanes, general aviation airplanes, helicopters, and some unmanned aircra — t. The tra —

ic — eed is a combination o —

several highly reliable sources used by the airline industry.
The Tra ---

ic Alert API automatically — ilters to alert the operator o — only tra —

ic that is low-altitude and near the current — light path.117

Airspace Advisory, AIRMAP, https://developers.airmap.com/docs/airspace-advisory [https://perma.cc/N2TF- S32R].
See, e.g., id.
           See         AirMap            Plat --- orm,            AIRMAP, https://www.airmap.com/plat --- orm/#industry-participation [https://perma.cc/C36A-UUST].
See id.
Solution Providers, AIRMAP, https://www.airmap.com/solution-providers [https://perma.cc/6X4G-USJR];              Features,            DRONELOGBOOK, https://www.dronelogbook.com/hp/1/ --- eatures.html          [https://perma.cc/UZ44- 2M79].
          Tra ---

ic Alerts Overview, AIRMAP, https://developers.airmap.com/docs/tra —

ic-alerts-overview [https://perma.cc/X97D-J24M].

 Electronic copy available at: https://ssrn.com/abstract=3435274

The AirMap UTM Dashboard makes it easy — or airspace managers to provide drone operators with sa — e and secure access to the airspace in three additional ways: (1) identi — ication—knowing who is — lying within managed airspace, with contact details and identity veri — ication capabilities; (2) geo — encing—creating digital boundaries with rules-based access requirements, instantly publishing geo — ences to thousands o — drone operators; and (3) communication—talking directly with drone operators via individual SMS text messages or broadcast noti — ications to AirMap-powered applications.118 All these products help small operations compete and even collaborate with large ones. AirMap’s products make it easy and relatively inexpensive to comply with regulations, even in this — ast-changing industry.

AirMap is thereby involved in RD. The costs involved in adhering to FAA regulations are high. Pilots and — irms involved in — lying need to hire employees who know the regulations, monitor those regulations as they change, and then still train those employees to properly use — light planning and airspace so — tware. The technological cost in conjunction with employment costs are high; AirMap mitigates some o — those costs by bringing it all together. 119

 118   See Authorities, AIRMAP, https://www.airmap.com/authorities/ [https://perma.cc/458B-EQSU].
 119 Airmap speci --- ically shows how RD helps young companies compete with old ones. Over time, regulated companies develop relationships with their regulators. These regulator-regulated relationships give regulated companies more access to the regulator’s rulemaking process, help the regulated understand regulatory rules, and provide a way  --- or the regulated to exert in --- luence when the regulator is determining the meaning o ---  regulatory standards. See Norm Champ, Building E ---

ective Relationships with Regulators, HARV. L. SCH. F. ON CORP. GOVERNANCE (Oct. 22, 2015), https://corpgov.law.harvard.edu/2015/10/22/building-e —

ective-relationships- with-regulators [https://perma.cc/4JTK-8GKS]. AirMap levels that playing — ield. UAV operators have to comply with a complex web o — regulations promulgated by FAA. See, e.g., Fact Sheet – Small Unmanned Aircra — t Regulations (Part 107), FED. AVIATION ADMIN. (July 23, 2018), https://www. — aa.gov/news/ — act_sheets/news_story.c — m?newsId=22615 [https://perma.cc/89KP-KP36]. These FAA UAV regulations change — requently. See, e.g., FAA Highlights Changes — or Recreational Drones, FED. AVIATION ADMIN. (May 16, 2019, 12:04 PM), https://www. — aa.gov/news/updates/?newsId=93769 [https://perma.cc/H4MG- 5C3D]. Although large corporations, such as Boeing, have in — luenced FAA rulemaking historically, today small startups like AirMap are encouraging “collaborate acceleration.” Through collaborate acceleration, startup aviation companies can engage collectively in lobbying the FAA and encouraging the agency to take a more innovative approach. See Airmap Supporting NASA and FAA UTM Research Projects, AIRMAP (Apr. 29, 2019), https://www.airmap.com/airmap-supporting-nasa-utm- — aa-upp-ipp-maap-

 Electronic copy available at: https://ssrn.com/abstract=3435274

3. Avalara

Avalara is a startup that o

ers a number o — regulatory democratization solutions — or startups and small businesses, all o —

whom need to

ile taxes. 120 Unlike TurboTax and H&R Block, which help natural persons — ile their individual tax returns, Avalara

ocuses on tax returns — or businesses.121 The startup thereby helps its customers comply with Internal Revenue Service (IRS) tax regulations.

The Avalara AvaTax product uses algorithmic devices to help e- commerce startups and small businesses determine the taxes they owe, especially regarding commerce across state lines.122

The Avalara Returns product allows businesses to “prepare[] and

ile [their] sales and use tax returns with a higher degree o —

accuracy than doing it” on their own. 123 In other words, the product is designed to reduce error costs and penalties — or — ailure to comply with regulations. Moreover, Avalara expressly acknowledges that their product meets the criteria — or a regulatory democratization solution: “Avalara pricing is volume-based and designed to scale with your business.”124 This quali — ies Avalara as a — irm o —

ering an RD solution.

4. Cognigo

Cognigo,

ounded in 2016125 and acquired by NetApp in 2019,is another business that describes itsel — as a startup o —

ering a RD solution. Cognigo uses arti — icial intelligence (AI) and machine learning to help businesses protect their data and stay in

research-projects [https://perma.cc/Y75Q-X43X]. This

acilitates the development o — rules that permit young startup companies to innovate in the UAV market. 120 See We Live and Breathe Tax Compliance So You Don’t Have To, AVALARA, https://www.avalara.com/us/en/about/trust.html [https://perma.cc/YWS4-FA6S]. 121 See Avalara Products, AVALARA, https://www.avalara.com/us/en/products.html [https://perma.cc/3G95-KMDU]. 122 See The Power o — Orange, AVALARA, https://www.avalara.com/us/en/about.html [https://perma.cc/7CP5-8SYF]. 123 Returns Preparation and Filing, AVALARA, https://www.avalara.com/us/en/products/sales-and-use-tax/returns.html [https://perma.cc/PYE4-HXFK]. 124 AvaTax Pricing, AVALARA, https://www.avalara.com/us/en/products/sales-and-use-tax/avatax/avatax- pricing.html [https://perma.cc/YLB4-7XNR]. 125 See Cognigo, MEDICI GLOBAL, INC., https://gomedici.com/companies/cognigo [https://perma.cc/XHV4-BHT7].

 Electronic copy available at: https://ssrn.com/abstract=3435274

compliance with regulations involving data privacy, 126 such as the General Data Protection Regulation (GDPR).127 Cognigo

promises that it can help businesses protect their critical
data assets and prevent personally identi --- iable in --- ormation

rom leaking outside o — the company’s network. And it says it can do so without the kind o — hands-on management that’s o — ten required in setting up these kinds o — systems and managing them over time.128

A

ter the initial setup, Cognigo’s solution is entirely “human-

ree.”129 That means Cognigo has the capability o — scaling massively. The company achieves data security and compliance with GDPR in a — raction o — the time: cognitive computing allows a startup to become compliant in days as opposed to months.130 Cognigo manages and secures critical data assets through their DataSense program by using advanced machine learning algorithms.131 The DataSense program “has been trained to detect common categories like payslips, patents, NDAs [non-disclosure agreements], and contracts. Organizations can also provide their own data samples to — urther train the model and customize it — or their own needs.”132

 126 Frederic Lardinois, Cognigo Raises $8.5M  --- or its AI-Driven Data Protection Plat --- orm, TECHCRUNCH (Nov. 13, 2018), https://techcrunch.com/2018/11/13/cognigo-raises-8-5m- --- or-its-ai-driven-data- protection-plat --- orm/ [https://perma.cc/JA9B-ZJSS]; see also Derrick L. Maultsby, Jr. & Jason L. Ott, The Future o ---  Data Privacy: Corporate Compliance in a Post-GDPR Global Market, CPO MAGAZINE (Nov. 15, 2018), https://www.cpomagazine.com/data-protection/the- --- uture-o --- -data-privacy- corporate-compliance-in-a-post-gdpr-global-market [https://perma.cc/XB3Z- T38U] (“That lack o ---  legal authority concerning ownership o ---  personal in --- ormation has changed signi --- icantly with the recent enactment o ---  the General Data Protection Regulation (the ‘GDPR’) in the European Union (the ‘EU’). In a broad sense, the GDPR has given EU residents power over their personal in --- ormation. The GDPR bill, which was passed in 2016 and took e ---

ect in May o — 2018, grants EU residents substantial rights with regard to their personal in — ormation[]” including “the right to be — orgotten[,] the right to access” the in — ormation and “the right to data portability.”). 127 Regulation 2016/679 o — the European Parliament and o — the Council o —

27 April 2016 on the Protection o

Natural Persons with Regard to the Processing o — Personal Data and on the Free Movement o — Such Data and Repealing Directive 95/46/EC, 2016 O.J. (L119/1) [hereina — ter “GDPR”]. 128 See Lardinois, supra note 126. 129 Id. 130 See Cognigo, CRUNCHBASE, https://www.crunchbase.com/organization/cognigo [https://perma.cc/RZ4A- 3TLC]. 131 See Joseph F. Kovar, NetApp Acquires Israeli Data Security Developer Cognigo, CRN (May 30, 2019, 4:26 PM), https://www.crn.com/news/storage/netapp-acquires-israeli-data-security- developer-cognigo [https://perma.cc/T8T9-KAXX]. 132 Lardinois, supra note 126.

 Electronic copy available at: https://ssrn.com/abstract=3435274

The speci — ics regarding Cognigo’s AI technology and examples o —

its use in the real world will require

urther study. Although

ounded recently in 2016, the company has already been acquired by NetApp in an e —

ort to “provide NetApp’s Cloud Volume Ontap with AI-driven compliance. That will be increasingly important given that over 80 percent o — organizations will — ail to implement a comprehensive data governance scheme by 2021 . . . .”133 It is clear that Cognigo is involved in RD. Moreover, Cognigo shows how AI technology can enable more RD solutions.

5. Metomic

Metomic is a new startup that was

ounded in June o — 2018134 to help startups and small businesses comply with a new regulation, the GDPR,135 which was implemented on May 25, 2018.136 In

act, Metomic was — ounded to be a GDPR solution — or the multitude o — small — irms that could not otherwise a —

ord to comply with the new regulation.137

There are a number o

irms o —

ering GDPR solutions, but most o —

er these solutions in the — orm o — a traditional consulting arrangement, where a consultant o —

ers customized compliance advice.138 Much like the high — ixed cost o — implementing customized HIPAA compliant email services in a medical o —

ice,139 implementing customized GDPR compliant solutions requires — irms to expend signi — icant resources up- — ront.

Metomic is di

erent in that it o —

ers tools that — irms can use on a variable basis. Its application program inter — aces (APIs) can be licensed on a per-domain or per-user basis.140 For small — irms who

Kovar, supra note 131.
              See               Metomic,                 CRUNCHBASE, https://www.crunchbase.com/organization/metomic#section-overview [https://perma.cc/V9WC-V2CZ].
METOMIC, https://metomic.io [https://perma.cc/XZ8B-JAKH].
See Linda Hsu & Jena Valdetero, Data Breach Litigation Preparation: How Do European Union Breach Noti --- ication Requirements Di ---

er — rom the U.S.?, JD SUPRA (Feb. 6, 2020), https://www.jdsupra.com/legalnews/data- breach-litigation-preparation-how-31454 [https://perma.cc/5BWC-789F]. 137 See METOMIC, supra note 135. 138 See Amy Cross, Best GDPR So — tware Tools and Solutions, NGDATA (Feb. 1, 2018), https://www.ngdata.com/gdpr-so — tware-tools-and-solutions [https://perma.cc/AZJ7-2VXN] (listing — i — ty-one services o —

ering GDPR compliance solutions). For example, IBM Security, Connexica, LogicManager, OneTrust, BWise, Forcepoint, Proo — point, ZenGRC, etc. Id. 139 See supra notes 79–82, 91–95 and accompanying text. 140 See Metomic, GOLDEN, https://golden.com/wiki/Metomic [https://perma.cc/QHS5-LVBX]; Jonathan Freeman, What Is an API? Application Programming Inter — aces Explained, INFOWORLD (Aug. 8, 2019, 3:00 AM),

 Electronic copy available at: https://ssrn.com/abstract=3435274

have low site usage, Metomic is totally — ree. 141 As the business grows, Metomic o —

ers various packages that range up to $250 per month per domain.142

The scalable, variable cost nature o

Metomic’s GDPR solution distinguishes it — rom traditional privacy consulting services, which have high up- — ront — ixed costs. As such, Metomic is an RD — irm, whose business model is to help startups and small businesses o —

all sizes—especially the smallest ones—and to provide web solutions that comply with GDPR.

6. Amazon Web Services (AWS)

While Amazon Web Services (AWS) is not a startup—Amazon is one o — the largest companies in the world 143—many entrepreneurs use AWS — or compliance.144 From the interviews conducted in The Startup Survey, researchers learned that AWS prominently

actors into how startups are able to compete with incumbents. AWS is a paradigm o — RD, and it makes a huge impact on the ability o — startups to enter regulated markets.

AWS launched in 2006 to provide pay-as-you-go access to computing power.145 Prior to AWS, data processing generally required the construction o — expensive server — arms,146 a project that only large and well- — unded — irms could undertake.147 AWS

https://www.in

oworld.com/article/3269878/what-is-an-api-application- programming-inter — aces-explained.html [https://perma.cc/PBE7-V92M]. 141 Metomic, supra note 140. 142 See Choose the Plan That’s Right — or You, METOMIC, https://metomic.madetogether.co/pricing [https://perma.cc/6NLL-56ZB]. 143 See Alexis C. Madrigal, When Amazon Went — rom Big to Unbelievably Big, ATLANTIC (Feb. 7, 2019), https://www.theatlantic.com/technology/archive/2019/02/when-amazon-went-

rom-big-to-unbelievably-big [https://perma.cc/B69X-D2GT]. 144 See About AWS, AMAZON, https://aws.amazon.com/about-aws [https://perma.cc/C7VV-YXLQ]. 145 See id. 146 See Six Advantages o — Cloud Computing, AMAZON, https://docs.aws.amazon.com/whitepapers/latest/aws-overview/six-advantages- o — -cloud-computing.html [https://perma.cc/N98K-UA7W] (“Cloud computing lets you — ocus on your own customers, rather than on the heavy li — ting o —

racking, stacking, and powering servers.”). 147 See Ke-Wei Huang & Mengqi Wang, Firm-Level Productivity Analysis

or So — tware as a Service Companies, 2009 PROC. INT’L CONF. ON INFO. SYSTEMS 1, 5, https://aisel.aisnet.org/cgi/viewcontent.cgi?article=1184&context=icis2009 [https://perma.cc/768W-CT8N] (“In an industry with economies o — scale, — irms will enjoy lower average costs when the — irm size is larger, rewarding larger

irms and leading to a monopoly or oligopoly in the end.”).

 Electronic copy available at: https://ssrn.com/abstract=3435274

o —

ered “data democratization” and provided a scalable in — rastructure that startups and small businesses could use.148

Moreover, AWS is certi

ied as compliant in over — i — ty programs worldwide.149 In particular, AWS is compliant with GDPR.150 AWS users do not need to take any additional action to get the bene — it o — AWS’s GDPR compliance. This represents a huge savings o —

ixed regulatory in — ormation and compliance costs to startups.151

AWS provides many services, but their central product is Amazon Elastic Compute Cloud (EC2). 152 As the name implies, EC2 makes data processing more — lexible by allowing users to rent virtual computers on which to run their own computer applications. Users pay by the second to use servers that AWS owns.153

This proved to be a win-win

or Amazon and — or the marketplace; over 150,000 developers,154 17,500 nonpro — its,155 2,400

AWS Partner Story: Edmunds, AMAZON, https://aws.amazon.com/partners/success/edmunds-tableau [https://perma.cc/8RPN-X7PQ].
     See       AWS        Compliance        Programs,        AMAZON, https://aws.amazon.com/compliance/programs [https://perma.cc/J4TZ-UQDJ].
General Data Protection Regulation (GDPR) Center, AMAZON, https://aws.amazon.com/compliance/gdpr-center           [https://perma.cc/HH75- H8ZW].
C --- . Oliver Smith, The GDPR Racket: Who’s Making Money From This $9bn Business Shakedown?, FORBES (May 2, 2018, 2:30 AM), https://www. --- orbes.com/sites/oliversmith/2018/05/02/the-gdpr-racket-whos- making-money- --- rom-this-9bn-business-shakedown/#70e8a78234a2 [https://perma.cc/882T-PZYB]. Forbes reported that the world’s Fortune 500 companies spend a total o ---  $7.8 billion on GDPR compliance, an average o ---  $16 million per  --- irm. See id. Startups would likely spend less per  --- irm, but this still paints a picture o ---  the huge compliance cost  --- or GDPR.
  See    Amazon EC2,         AMAZON, https://aws.amazon.com/ec2/ [https://perma.cc/YCE6-PTX8].
See Amazon EC2 Pricing, AMAZON, https://aws.amazon.com/ec2/pricing/ [https://perma.cc/T7TB-MMFJ] (“With On-Demand instances, you pay  --- or compute capacity by the hour or the second depending on which instances you run. . . . With per-second billing, you pay  --- or only what you use.”).
See Savia Lobo, Why AWS Is the Pre --- erred Cloud Plat --- orm  --- or Developers Working with Big Data, PACKT (June 7, 2018), https://hub.packtpub.com/why- aws-is-the-pre --- ered-cloud-plat --- orm- --- or-developers-working-with-big-data [https://perma.cc/G7EV-67GD].
See Announcing AWS Cloud Credits  --- or Nonpro --- its with TechSoup Global, AMAZON                        (Mar.                  4,                    2016), https://aws.amazon.com/blogs/publicsector/announcing-aws-cloud-credits- --- or- nonpro --- its-with-techsoup-global/ [https://perma.cc/9GA5-L46Y].




 Electronic copy available at: https://ssrn.com/abstract=3435274

educational institutions,156 and 5,000 government agencies 157 have used AWS, while AWS generated over $25.6 billion in sales in 2018.158

In economic terms, AWS changed the cost curve

or data processing in a way that especially bene — itted startups and small businesses. Prior to AWS, data processing required a huge up — ront investment, or a “ — ixed cost,” in that the expense o — the server — arm must be incurred be — ore any production or sales o — data occurred.159 Pre-sales startups may not have income to pay — or these initial — ixed costs, whereas incumbent — irms with multiple revenue streams can cover this investment relatively easily. Additionally, some o — these up — ront costs are “sunk” in that they can never be recovered i — the project — ails. 160 Startups and small businesses cannot manage the risk o — losing sunk costs to the degree that incumbents can. 161

AWS thus exhibits a strong

orm o — RD that can have a major impact on startup entry and post-entry growth and innovation in markets regulated by GDPR and other privacy laws. Incumbents may pre — er industrial organizations that demand large sunk costs so they can — end o —

startups. 162 AWS helps startups compete against

            See           AWS               Educate,            AMAZON, https://aws.amazon.com/education/awseducate [https://perma.cc/X5LS-FX8R].
   See     The   Trusted     Cloud       --- or    Government,      AMAZON, https://aws.amazon.com/government-education/government [https://perma.cc/A3S2-5QVM].
Alison Griswold, Amazon Web Services Brought in More Money than McDonald’s in 2018, QUARTZ (Feb. 1 , 2018), https://qz.com/1539546/amazon- web-services-brought-in-more-money-than-mcdonalds-in-2018 [https://perma.cc/S3D2-ADEC]; How Much is Amazon Web Services Worth on a Standalone Basis, FORBES (Feb. 28, 2019, 3:45 PM), https://www. --- orbes.com/sites/greatspeculations/2019/02/28/how-much-is- amazon-web-services-worth-on-a-standalone-basis/#41b7be --- 8bbb7 [https://perma.cc/E4D9-DL2M].
See STEVEN A. GREENLAW & DAVID SHAPIRO, PRINCIPLES OF MICROECONOMICS 167 (2d ed. 2018) (“At zero production, the  --- ixed costs . . . are still present.”).
See JEFFREY M. PERLOFF, MICROECONOMICS 185 (Donna Battista et al. eds., 5th ed. 2009) (“Because the equipment has no alternative use, the historical cost o ---  buying that capital is a sunk cost: an expenditure that cannot be recovered.”).
See Robert S. Pindyck, Sunk Costs and Risk-Based Barriers to Entry 1 (Nat’l Bureau o ---  Econ. Research, Working Paper No. 14755, 2009), https://www.nber.org/papers/w14755.pd ---  [https://perma.cc/X5PV-FYCN] (“Thus large sunk costs are clearly an entry barrier; by creating scale economies, they lead to an industry equilibrium with relatively  --- ew  --- irms.”).
See Richard Schmalensee, Sunk Costs and Antitrust Barriers to Entry 2 (Mass. Inst. o ---  Tech. Sloan Sch. o ---  Mgmt., Working Paper No. 4457-04, 2004), https://papers.ssrn.com/sol3/papers.c --- m?abstract_id=486944 [https://perma.cc/KN4W-J4HG] (“But unless entry involves sunk costs—which cannot be recovered or, i ---  amortized and treated as a  --- low, cannot be avoided i ---





 Electronic copy available at: https://ssrn.com/abstract=3435274

incumbents by trans — orming sunk costs into variable costs.163 AWS converted the up- — ront — ixed costs o — building a GDPR- compliant server — arm into the variable cost o — server access, which scales in step with output or sales. 164 This data democratization helps startups process data and comply with privacy regulations at a per unit cost similar to their incumbent competitors. Thanks to AWS’s RD solution, startups can pay as they grow, instead o —

making big investments up

ront. Otherwise, the large — ixed costs o — creating data servers that are compliant with the new, complex, and dynamic web o — privacy regulations would bene — it high- volume incumbents over pre-sales startups.

As discussed above,

ixed costs, and especially sunk costs, have a disparately negative impact on startups. 165 It is well-documented that converting — ixed costs into variable costs can enhance competition.166 By de — inition, — ixed costs are a type o — entry barrier, as they are costs that must be borne by a startup trying to enter the market, but not by an incumbent who is already in the market.167 In antitrust law, it is axiomatic that markets with lower

ixed costs are more competitive. 168 Moreover, this competition results in lower prices and higher quality products. 169 Indeed, one reason why many startups attempted to enter the so — tware industry

exit occurs—contestability theory teaches that scale economies do not su

ice to permit established — irms to hold price above cost without attracting (hit-and-run) entry.”); see also JOE S. BAIN, INDUSTRIAL ORGANIZATION 252 (2d ed. 1968) (“[T]he condition[s] o — entry [are] the extent to which, in the long run, established — irms can elevate their selling prices above the minimal average costs o — production and distribution . . . without inducing potential entrants to enter the industry.”). 163 For a discussion o — the advantages o — variable costs, see Schmalansee, supra note 162, at 5–6. 164 See PERLOFF, supra note 160 at 186 (“A — irm’s variable cost (VC) is the production expense that changes with the quantity o — output produced.”). 165 See Pindyck, supra note 161, at 1. 166 See, e.g., Yunchuan Liu & Rajeev K. Tyagi, Outsourcing to Convert Fixed Costs into Variable Costs: A Competitive Analysis, 34 INT’L J. RES. MARKETING 252, 252 (2017) (“[Outsourcing] allows the outsourcing — irm to reduce its — ixed cost[s] such as expenditures on equipment, in — ormation technology, — ixed salaries o — employees, etc., and convert those into a variable cost in the — orm o — the purchase price that the outsourcing — irm then pays the outside industry.”). 167 See William J. Baumol & Robert D. Willig, Fixed Costs, Sunk Costs, Entry Barriers, and Sustainability o — Monopoly, 96 Q.J. ECON. 405, 416 (1981) (“[F]ixed costs o — su —

icient magnitude can e —

ectively prevent entry, which suggests that they might be classed as a type o — entry barrier”). 168 See Nirmalya Kumar, Strategies to Fight Low-Cost Rivals, HARV. BUS. REV. (Dec. 2006), https://hbr.org/2006/12/strategies-to- — ight-low-cost-rivals [https://perma.cc/3AT8-EK8X]. See generally PHILLIP E. AREEDA & HERBERT HOVENKAMP, ANTITRUST LAW: AN ANALYSIS OF ANTITRUST PRINCIPLES AND THEIR APPLICATION (3d ed. 2006). 169 See BUREAU OF COMPETITION, FED. TRADE COMM’N, COMPETITION COUNTS: HOW CONSUMERS WIN WHEN BUSINESSES COMPETE 2 (2015), https://www. — tc.gov/system/ — iles/attachments/competition-counts/pd — - 0116_competition-counts.pd — [https://perma.cc/4RNG-2RQK].

 Electronic copy available at: https://ssrn.com/abstract=3435274

in the dot.com era is that so — tware engineering — irms were able to shi — t most o — their costs to variable costs. 170

There

ore, AWS provides a vital service to startups by enabling the conversion o —

ixed and sunk costs o — regulatory compliance to variable costs, which also equalizes the incidence o — these costs on startups vis-à-vis incumbents.

       IV. ANALYSIS OF REGULATORY DEMOCRATIZATIONS

What is common about the regulations that RD startups have addressed? Answering this question may provide insights toward designing e —

icient business regulations. Startups have developed RD compliance solutions — or the Health Insurance Portability and Accountability Act (HIPAA), Federal Aviation Administration (FAA) airspace regulations o — unmanned aerial vehicles (UAV), the Internal Revenue Service (IRS) tax code, and the General Data Protection Regulation (GDPR). What is common about these regulations? Section A will answer that question. Section B will apply that answer to an analysis o — when RD is likely to arise and

oster competition in regulated industries.

                   A. The Nature o ---  Regulation

HIPAA, FAA UAS, GDPR, and the IRS tax code are all complex, rule-based systems o — regulation. These are distinguishable — rom simple rules and standards. Figure 2 illustrates how complex rules

it into the taxonomy o — regulations.

 170 See J.J. Dolado, On the Problem o ---  the So --- tware Cost Function, 43 INFO. & SOFTWARE TECH. 61, 69 (2001) (“In most cases, the marginal return is so low that asserting that economies or diseconomies o ---  scale exist is very questionable.”).

 Electronic copy available at: https://ssrn.com/abstract=3435274

COMPLEX COMPLEX RULE STANDARD

                            SIMPLE                       SIMPLE
                             RULE                      STANDARD

Figure 2. Taxonomy o

Regulations.

1. Rules Versus Standards

As shown in Figure 3, there are

our categories o — regulations across two dimensions: simple/rule, complex/rule, simple/standard, and complex/standard.171 RD seems to arise only in response to complex rules. Unpacking the meaning o — this — inding requires a brie — discussion on rules versus standards.

The economics o

rules versus standards usually begins by distinguishing ex ante versus ex post decision-making.172 Ex ante costs are incurred be — ore an action is taken. 173 Ex post costs are

See Louis Kaplow, Rules Versus Standards: An Economic Analysis, 42 DUKE L.J. 557, 566 (1992).
See DAVID D. FRIEDMAN, LAW’S ORDER: WHAT ECONOMICS HAS TO DO WITH LAW AND WHY IT MATTERS 74 (2000). Ex ante decisions are those which are made with the in --- ormation the actor possesses at the time. Id. Some ex ante punishments are designed to mitigate behavior that would increase the probability o ---  an undesirable outcome, such as punishing drunk driving. Id. However, ex post laws punish behavior based on the undesirable outcome observed and there --- ore exploit an actor’s personal, private in --- ormation by having the actor use his knowledge to con --- orm to the law, such as imposing tort liability  --- or damage done. Id. at 74–75.
See id.

 Electronic copy available at: https://ssrn.com/abstract=3435274

incurred a — terward.174 These concepts map neatly on to a simpli — ied version o — rules and standards. 175

Rules are determined ex ante, in advance.176 For example, “Speed Limit 70 MPH” is a rule. 177 The speed limit is determined in advance. This requires regulators to pay an up — ront rulemaking cost,178 such as by surveying roadways, posting signs, updating equipment, and monitoring noncompliance. The regulated must also bear an up- — ront cost to become in — ormed about the speed limit and monitor — or changes. 179 Although rules also have ex post costs including detection and en — orcement, 180 breaches o — rules are relatively easy to adjudicate, so more o — the costs o — rules are incurred up — ront.

In contrast, standards are determined ex post, a

ter the — act. On highways in Montana, with no posted speed limits, drivers must still be “reasonable.”181 What does that mean? In this initial state, a standard is an abstraction, and it does not have a concrete meaning. Only a — ter someone is pulled over — or driving 120 MPH through rural Montana and is adjudicated to have been driving unreasonably does this standard have a speci — ic context. Over time, however, as cases involving unreasonable speed in Montana are adjudicated, the collective impact o — this precedential case law is to have predictive value:182 i — 99 o — 100 people who were pulled over — or driving 120 MPH were adjudicated to be unreasonable, we can now determine with su —

icient con — idence ex ante that 120 MPH is unreasonably — ast.

Thus, standards have the advantage o

generating rule-making costs only as needed.183 Unlike rules, where every scenario must be accounted — or in advance, a standard is only de — ined when its boundaries are tested. Moreover, it gains de — inition — rom the real world, and thus is most likely to comport with reasonable expectations.

174 See id. 175 See Kaplow, supra note 171, at 559–60. 176 See id. at 589. 177 See id. at 565. 178 See id. at 568–69. 179 See id. at 569. 180 See id. at 570. 181 MONT. CODE ANN. § 61-8-303(3) (2019). 182 See Kaplow, supra note 171, at 577–79. 183 See id. at 610 & n.148.

Electronic copy available at: https://ssrn.com/abstract=3435274

2. Simplicity Versus Complexity

Whether a rule or a standard, a regulation can also be simple or complex.184 The complexity o — a regulation is related to the number or amount o —

acts that must be considered in order to apply the rule.185 In the case o — rules, complexity is relatively easy to illustrate. For example, “Car Parking $1 Per Hour” is a simple rule in that it only requires one — act, namely, how long the car was parked. Parking rules can be more complex o — course, such as: “Car Parking $1 Per Hour, no parking 8:30 am – 9:30 am on weekdays, — ree parking 6 pm – 6 am and on Sundays.”

Complex rules generally cost more to develop than simple rules. Complexity with regards to standards is a bit harder to quanti — y— which is unsurprising given the inherently qualitative nature o —

standards—but here again an illustration helps demonstrate the distinction. One o — the simplest standards may be the constitutional ban on cruel and unusual punishments: 186 “It does not take most people very long to decide whether a punishment is cruel.”187

A more complex standard may be

ound in the doctrine o —

promissory estoppel. This doctrine, which is also known as detrimental reliance, was established in the late nineteenth century by judges who — elt compelled to en — orce promises that did not rise to the level o — contracts at law.188 A common articulation o — this standard is as — ollows: “A promise which the promisor should reasonably expect to induce action or — orbearance on the part o — the promisee or a third person and which does induce such action or

orbearance is binding i — injustice can be avoided only by en — orcement o — the promise.”189 This is a complex standard because reasonable people o — ten disagree when “injustice can be avoided only by en — orcement o — the promise.” 190 Making such a determination usually requires a great deal o —

act- — inding and a

See id. at 566.
See, e.g., id. at 565.
See U.S. CONST. amend. VIII.
John Harrison, Richard Epstein's Big Picture, 63 U. CHI. L. REV. 837, 862 (1996).
See Kevin M. Teeven, Origins o ---  Promissory Estoppel: Justi --- iable Reliance and Commercial Uncertainty Be --- ore Williston’s Restatement, 34 U. MEM. L. REV. 499, 500 (2004); Eric Mills Holmes, Restatement o ---  Promissory Estoppel, 32 WILLAMETTE L. REV. 263, 275–76, 275 n.23 (1996).
RESTATEMENT (SECOND) OF CONTRACTS § 90(1) (AM. LAW INST. 1981).
See id.; Charles L. Knapp, Rescuing Reliance: The Perils o ---  Promissory Estoppel, 49 HASTINGS L. J. 1191, 1218–19 (1998).




 Electronic copy available at: https://ssrn.com/abstract=3435274

ull-blown bench trial.191 This makes complex standards quite expensive to administer in comparison to simple standards.

3. Evolution o ---  Standards into Rules

However, the discussion above about simplicity versus complexity was oversimpli — ied in that it did not take into account the impact o —

precedent over time. Any cost-bene

it analysis o — rules versus standards is incomplete i — it assumes that laws are static, and that each case is a matter o —

irst impression. This is o — course not how our common law legal system works. Rather, over time, precedent, whether binding or simply persuasive, in — orms the application o — regulations.192

Recall the distinction o

two speed limits, a “70 MPH” rule and a “reasonable and prudent speed” standard. 193 In a matter o —

irst impression, a judge would need to conduct extensive — act- — inding to determine whether a driver violated the standard-based speed limit. But, over time, as similar cases are decided, this vague standard develops characteristics o — a complex rule. For example, i — a judge determines that a driver operating a high-end sports car at 120 MPH on a clear day with excellent driving conditions was driving unreasonably, this e —

ectively creates a rule that sets an upper bound to what could be considered a reasonable speed. Later, another judge determines that a driver was imprudent when driving 80 MPH in heavy snow, but another driver was prudent when driving 90 MPH to take his — riend to the hospital.

A

ter years o — this standard being on the books, something resembling a complex rule evolves. For instance, this hypothetical speed limit standard begins to have rule-like — eatures such as an absolute 120 MPH speed limit in all cases, a lower 80 MPH speed limit in daytime snow, and an exception — or emergency services.

   B. Rules, Standards, and Regulatory Democratization

The case studies o

RD described in Part II share a common thread. Although they address di —

erent regulations—HIPAA, FAA UAV regulations, the IRS tax code, and GDPR—they all address the same type o — regulation: complex, rule-based regulations. This

See generally Knapp, supra note 190 (providing an overview o ---  the Second Restatement and analyzing multiple promissory estoppel claims that went to trial).
See Kaplow, supra note 171, at 577–79.
See supra notes 173–181 and accompanying text.

 Electronic copy available at: https://ssrn.com/abstract=3435274

Section B explains why complex, rule-based regulations are most likely to encourage the development o — RD technology.

In short, computer programs are good at processing rules but incompetent at exercising discretionary standards. Now that computer technology has scaled to provide regulatory compliance

or large numbers o — small — irms, it is worth revisiting the economics behind the rules versus standards debate to see whether technological change in in — ormation and en — orcement costs have impacted the cost-bene — it analysis.

1. Simple Rules and RD

The Startup Study did not

ind evidence o — large-scale RD in response to simple rules. However, that does not mean that simple rules are in — erior to complex ones. Rather, it tends to show that simple rules may promote e —

icient compliance even without technological solutions. For example, numeric speed limits are simple rules that have been reasonably e —

icient despite a lack o —

technological advancement. 194 The non-existence o

RD in response to simple rules may simply show these rules tend to be e —

icient without — urther intervention.

Moreover, when rules are relatively simple, regulators o

ten create their own compliance technologies.195 Although this phenomenon is itsel — a subject — or another paper, this Article has alluded to the emergence o — regulatory technology that lowers in — ormation and en — orcement costs.

While complex regulations may

oster RD, they may inhibit other desirable business activities. Complexity is not a good unto itsel — . As stated earlier, holding everything else constant, a simpler regulation will have less o — a disparate impact on startups than a complex regulation.196 Complex regulations generally have higher in — ormation costs, and startups are less able to distribute those

See generally Tom Sohrweide, The Truth About Speed Limits, Explained by       an      Engineer,      SHORT      ELLIOTT      HENDRICKSON         INC., http://www.sehinc.com/news/truth-about-speed-limits-explained-engineer [https://perma.cc/3EA4-DN63] (explaining that speed limits do not dictate speeds but rather re --- lect the way drivers behave on the road).
See Richard Williams & Mark Adams, Regulatory Overload, 103 MERCATUS               ON           POL’Y          1,        3           (2012), https://www.mercatus.org/system/ --- iles/Regulatory_Overload_WilliamsAdams_ MOP103.pd ---  [https://perma.cc/H8T7-GLRF].
See supra notes 20–23, 165 and accompanying text.

 Electronic copy available at: https://ssrn.com/abstract=3435274

in — ormation costs than incumbents. 197 By lowering in — ormation costs, the startups’ disadvantage is likewise decreased.198

However, not all regulations can easily be made simple. Li

e is complex.199 There are two approaches to regulating a complex situation: granting a great deal o — discretion to the regulator to

igure out the application o — a rule in each speci — ic scenario or dra — ting complex regulations that account — or many di —

erent scenarios in advance. 200 The latter is the lesser o — two evils, at least inso — ar as RD is concerned, as the more complex but inherently knowable regulation can be resolved to some level o —

certainty through the process o

RD.

2. RD’s Incompatibility with Standards

Problematically

or RD, “reasonable” standards cannot usually be determined algorithmically in advance by technology. What speed should a sel — -driving car operate on a new vacant highway in central Montana? How should the car know what is a “reasonable” speed? And who should be liable i — that speed turns out a — ter the

act to be unreasonable?

It is the providence o

lawyers and consultants to opine on the best response to uncertainty, whereas computers cannot replace human judgment.201 Standards, in other words, present a computational problem.

Regulatory discretion—that is, the degree to which regulators can make decisions on a case-by-case basis—inhibits RD. The introduction o — human discretion into the equation makes it very di —

icult or impossible to have generic solutions to regulatory problems. Technology is still no match — or the whim o — the bureaucrat. Even AI technology has di —

iculty in predicting human behavior.202

See supra notes 74–76 and accompanying text.
See supra notes 74–76 and accompanying text.
But see OSCAR WILDE, EPIGRAMS OF OSCAR WILDE 66 (2007) (“Li --- e is not complex. We are complex. Li --- e is simple and the simple thing is the right thing.”).
See Williams & Adams, supra note 195, at 1, 3.
See in --- ra notes 202–206 and accompanying text.
See, e.g., Jim Mainprice, Ra --- i Hayne & Dmitry Berenson, Goal Set Inverse Optimal Control and Iterative Re-planning  --- or Predicting Human Reaching Motions in Shared Workspaces, 32 INST. ELECTRICAL & ELECTRONICS ENGINEERS TRANSACTIONS ON ROBOTICS 897, 897 (2016). But see, e.g., Alex Pentland & Andrew Liu, Modeling and Prediction o ---  Human Behavior, 11 NEURAL COMPUTATION 229, 229 (1999) (demonstrating that a model o ---  human behavior can improve human-machine systems).




 Electronic copy available at: https://ssrn.com/abstract=3435274

Discretion is o — ten cloaked in vagueness. Vague terms are hard to understand and impossible to code, which drives up the cost o —

inexperience while simultaneously prohibiting RD. Discretion is likewise problematic because incumbents generally have longer relationships and more signi — icant working histories with regulators. Incumbents may also have more political in — luence, possibly due to their role in creating local jobs or donating to political campaigns.203

Moreover, creating a scalable solution

or a discretionary regulation would be a di —

icult task — or an enterprising RD startup. Discretionary regulations include language such as “reasonable” and thereby require a judge or some arbiter to determine what reasonable means in a certain case. 204 Such regulations do not have general rules that can be easily extrapolated and applied in many di —

erent contexts, and so, these discretionary rule problems would be harder to solve — or a wider range o —

irms who operate in more disparate — ields than a single large — irm.

Rules, however, are within the domain o

the computational sciences. Modern computers can make short work even o — very complex rules. When regulations are rule-based, RD technology can solve them with lower in — ormation and error costs.

Simple rules may be better at producing compliance and low cost, but some rules cannot be so simpli — ied. 205 Simple standards are more — easible to implement, but their discretionary nature is suboptimal — or an algorithmic approach. Moreover, their ambiguity may encourage a di —

erent process: regulatory entrepreneurship.206 The Startup Study revealed that complex regulatory standards o — ten result in rational entrepreneurs’ will — ul ignorance and noncompliance. See Figure 3 — or an illustration o —

regulatory predictors o

RD emergence.

 203 See Jonathan A. Knee, Review: Why Start-Ups Need a Regulatory Strategy to        Succeed,         N.Y.        TIMES        (Sept.       11,       2018), https://www.nytimes.com/2018/09/11/business/dealbook/ --- ixer-regulatory- hacking-review.html [https://perma.cc/F5YP-GBBA]; see also Eyal-Cohen, supra note 9, at 863–64 (“Whereas old-timers may be able to spread regulatory costs over their output or longevity, newcomers are more likely limited in their ability to mitigate these costs due to structural obstacles. . . . The e ---

ects o —

regulatory asymmetries on newcomers are palpable.”). 204 See supra Section IV.A.3. 205 See, e.g., Kaplow, supra note 171, at 589–90. 206 See Barry & Pollman, supra note 17, at 383, 385.

 Electronic copy available at: https://ssrn.com/abstract=3435274

Regulatory Non-Compliance Democratization

                                                                 Regulatory
                   Compliance
                                                              Entrepreneurship

Figure 3. Predictors o

regulatory democratization.

3. Complex Rules and RD

In contrast with regulatory discretion, which is likely to inhibit RD, regulatory complexity—the number o — words, lines, provisions, etc. in a rule—does not appear to hinder RD. In — act, there are theoretical reasons to believe that regulatory complexity promotes RD.207 The Startup Study’s qualitative — indings also support this.208

The Startup Study shows evidence o

RD in the — ace o — complex regulations. Consider the three regulations addressed by the startups in our case study: GDPR, HIPAA, and FAA UAS. They are all highly complex regulations. Just the o —

icial title o — the General Data Protection Regulation (GDPR)209 contains more words and characters than portions o — the Sherman Act that makes conspiracy illegal.210 The GDPR spans eighty-eight pages and is translated into twenty- — our languages.211 A common re — rain in the legal news is that “GDPR compliance can be complex.”212

HIPAA is also very complex. The U.S. Department o

Health & Human Services (HHS) Summary o — the HIPAA Privacy Rule is

See supra Section IV.B.1.
See supra Part II.
See GDPR, supra note 127.
Compare id., with Sherman Act, 15 U.S.C. § 1 (2018).
See GDPR, supra note 127; Jeroen Terstegge, GDPR: Lost in Translation?, INT’L ASS’N PRIVACY PROFESSIONALS (May 1, 2018), https://iapp.org/news/a/gdpr-lost-in-translation [https://perma.cc/MP86-BKY8].
Top Five Concerns with GDPR Compliance, THOMSON REUTERS, https://legal.thomsonreuters.com/en/insights/articles/top- --- ive-concerns-gdpr- compliance [https://perma.cc/3NJ8-E4LM].




 Electronic copy available at: https://ssrn.com/abstract=3435274

eighteen pages long, not including endnotes.213 The rule itsel — is so complex that HHS promulgated a HIPAA Administrative Simpli — ication.214 However, that simpli — ication is itsel — so complex that it is scattered in three di —

erent places throughout the Code o —

Federal Regulations.215 The combined text is 105 pages.216

Regulation o

drones is even more complex because it involves a network o —

ederal, state, and local regulations on unmanned aircra — t.217 While the — ederal government has “exclusive sovereignty o — airspace” over the United States, 218 there are actually six classes o — airspace, depending on the space’s height above sea level, height above ground level, and other — actors, plus special use airspace, and other airspace areas.219 These classi — ications are not static but may vary on a day-to-day or even hour-by-hour basis.220 Moreover, their use may be coordinated by a controller and conditioned on weather and other external

actors.221 In addition to this complex regulation o — public airspace, drone operators must also consider private airspace. 222 In short, we see regulatory democratization in highly complex regulatory environments.

 213 See U.S. DEP’T OF HEALTH AND HUM. SERVS., SUMMARY OF THE HIPAA PRIVACY RULE (2003), https://www.hhs.gov/sites/de --- ault/ --- iles/privacysummary.pd ---

[https://perma.cc/J5QJ-3WPU]. 214 See The HIPAA Privacy Rule, U.S. DEP’T OF HEALTH AND HUM. SERVICES, https://www.hhs.gov/hipaa/ — or-pro — essionals/privacy/index.html [https://perma.cc/E3HT-U6WL]. 215 See id.; see also 45 C.F.R. §§ 160, 164(A), (E) (2013). 216 See HIPAA Administrative Simpli — ication, 45 C.F.R. §§ 160–164 (2013). 217 See OFFICE OF CHIEF COUNSEL, FED. AVIATION ADMIN., STATE AND LOCAL REGULATION OF UNMANNED AIRCRAFT SYSTEMS (UAS) 1–3 (2015), https://www. — aa.gov/uas/resources/policy_library/media/UAS_Fact_Sheet_Final .pd — [https://perma.cc/Y283-9AVG] (providing an overview o — in — ormation about the — ederal regulatory — ramework — or UAVs). 218 49 U.S.C. § 40103(a)(1) (2018). 219 FLIGHT STANDARDS SERV., FED. AVIATION ADMIN., PILOTS HANDBOOK OF AERONAUTICAL KNOWLEDGE, at 15-2 (2016), https://www. — aa.gov/regulations_policies/handbooks_manuals/aviation/phak/me dia/17_phak_ch15.pd — [https://perma.cc/KVC2-HG9J]. 220 Matthew Johnston, Di —

erences Between Airspace Classi — ications, CAL. AERONAUTICAL U.: CALAERO BLOG (Nov. 8, 2018), https://calaero.edu/di —

erences-between-airspace-classi — ications/ [https://perma.cc/9N2C-B4J9]. 221 See id. 222 See United States v. Causby, 328 U.S. 256, 264 (1946). In United States v. Causby, the Supreme Court held that landowners have “exclusive control o —

the immediate reaches o

the enveloping atmosphere” and “[t]he landowner owns at least as much o — the space above the ground as he can occupy or use in connection with the land.” Id. This is a notable deviation — rom Roman law, which held: Cuius est solum, eius est usque ad coelum et ad in — eros (“Whoever owns the land it is theirs up to the heavens and down to hell.”). A DICTIONARY OF LAW 166–67 (Jonathan Law ed., 8th ed. 2015).

 Electronic copy available at: https://ssrn.com/abstract=3435274

This qualitative evidence supports the hypothesis that entrepreneurs may see a greater market opportunity in tailoring their service to complex regulations rather than simple ones. 223 There may generally be a trade-o —

between discretion and complexity. While prior scholarship seems to assume that startups would generally bene — it — rom simpler regulations,224 this does not account — or the discretion/complexity trade-o —

. In — act, RD has been observed only in industries whose regulations are highly complex. Holding everything else—including discretion— constant, a simpler regulation is more RD — riendly than a complex one, as the RD can o —

er a greater contribution by lowering regulatory in — ormation costs. Entrepreneurs may see a greater market opportunity in providing solutions to complex regulations rather than simple ones.225 As mentioned above, so long as the regulation is ultimately comprehensible, its complexity will not necessarily sti — le RD.

There are two countervailing issues that counsel against the use o

complex regulations to

oster RD. The — irst is an endogeneity problem, namely, that discretion and complexity generally correlate inversely.226 A regulation that gives regulators a great deal o — discretion can o — ten be much simpler than a regulation accounting — or most cases—although regulatory simplicity may have no correlation with regulatory burden. 227

                      V. IMPLICATIONS FOR POLICY

This Article is not the

irst to recognize that regulations have a disparate impact upon startups.228 Others have proposed various

Suggestions  --- or testing this hypothesis are  --- ound in the Conclusions, in --- ra notes 289⁠–294.
See supra Section IV.B.1.
Suggestions  --- or testing this hypothesis are  --- ound in the Conclusions, in --- ra notes 289⁠–294.
See supra Section IV.B.2.
For example, our  --- undamental antitrust regulation is incredibly simple. The Sherman Antitrust Act prohibits restraints o ---  trade in just one sentence: “Every contract, combination in the  --- orm o ---  trust or otherwise, or conspiracy, in restraint o ---  trade or commerce among the several States, or with  --- oreign nations, is hereby declared to be illegal.” 15 U.S.C. § 1 (2018). Antitrust law holds that a contract in unreasonable restraint o ---  trade is illegal. Standard Oil Co. o ---  N.J. v. United States, 221 U.S. 1, 87 (1911) (Harlan, J., concurring and dissenting). However, there is still much dispute over what restraints are “reasonable” in various contexts, each subject to a  --- ull-blown jury trial with  --- act-intensive analysis. See Mark C. Anderson, Sel --- -Regulation and League Rules Under the Sherman Act, 30 CAP. U. L. REV. 125, 128–30 (2002). In  --- act, despite antitrust law being  --- ounded on one o ---  the simplest statutes in the books, antitrust cases are some o ---  the most expensive to litigate. Jonathan M. Jacobson, From the Section Chair: Tackling the Time and Cost o ---  Antitrust Litigation, 32 ANTITRUST 3, 3–4 (2017).
See, e.g., Barry & Pollman, supra note 16.




 Electronic copy available at: https://ssrn.com/abstract=3435274

solutions to this problem.229 This new — inding contributes to the conversation on how to make regulations better — or entrepreneurs and encourage small business innovation. In particular, this Part discusses how regulatory democratization has an impact on three proposed solutions: regulatory sandboxes, in — ormation-sharing networks, and tax credits. Deregulation is also addressed.

                      A. Regulatory Sandboxes

Regulatory sandboxes are touted as a promising way

or regulators to partner with startups in experimenting with more e —

icient regulations.230 In this scheme, a regulator grants a speci — ic startup a variance on a regulation or order to permit that startup to experiment with a new technology in a live environment — or a limited time.231 Theoretically, this should provide sa — e spaces — or innovation.232

To date, however, reality has not matched this promise. Indeed, the — inding o — RD suggests that regulatory sandboxes are theoretically unlikely to improve the disparate impact o — regulation on startups and especially small businesses.233

To the extent that the regulatory sandbox approach has been employed at all, its use has generally been limited to — inancial technologies.234 The regulatory sandbox process is most predominately employed by the Financial Conduct Authority (FCA),235 a — inancial regulator in the United Kingdom who operates independently o — the UK government, 236 much like the Financial Industry Regulatory Authority (FINRA) does in the

See, e.g., id. at 430–47.
The Role o ---  Regulatory Sandboxes in Fintech Innovation, supra note 30.
Mike Faden, Regulatory Sandboxes Provide “Sa --- e Spaces”  --- or Fintech Payment              Services          Innovation,           AM.          EXPRESS, https://www.americanexpress.com/us/ --- oreign-exchange/articles/regulatory- sandboxes- --- or-innovative-payment-solutions          [https://perma.cc/P2SQ-LDFD] (citing FIN. CONDUCT AUTHORITY, REGULATORY SANDBOX 9 (2015), https://www. --- ca.org.uk/publication/research/regulatory-sandbox.pd ---

[https://perma.cc/3HPU-EWFR]). 232 See id. 233 See supra Part IV. 234 See, e.g., Allen, supra note 30, at 579; Faden, supra note 231. Note that legal scholarship actually de — ines regulatory sandbox narrowly to pertain to

inancial technology. E.g., Allen, supra note 30, at 579 (2019) (“A regulatory sandbox allows — intech startups to conduct a limited test o — their products with

ewer regulatory constraints, less risk o — regulatory en — orcement action, and ongoing guidance — rom regulators. . . .”). 235 See generally FIN. CONDUCT AUTHORITY, supra note 231. 236 About the FCA, FIN. CONDUCT AUTHORITY (Apr. 21, 2016) https://www. — ca.org.uk/print/about/the- — ca [https://perma.cc/924V-2UVF].

 Electronic copy available at: https://ssrn.com/abstract=3435274

United States.237 The FCA’s process allows innovative — irms to approach the FCA via its Innovation Hub and request a no en — orcement action letter (NAL),238 obtain individual guidance,239 or seek a waiver240 to o —

er novel — inancial products or services.

Congress showed some interest in implementing a regulatory sandbox in the United States.241 Also known as the McHenry Bill

or its sponsor, Rep. Patrick McHenry (R-NC), the Financial Services Innovation Act would create a new Financial Services Innovation O —

ice (FSIO) in each o — the — ederal — inancial agencies.242 Each FSIO would have a process whereby startups could propose an alternative compliance strategy — or a — inancial innovation that serves the public interest, improves access to

inancial products and services, and does not present systemic risk while promoting consumer protection. 243 But the bill, introduced

About FINRA, FIN. INDUSTRY REG. AUTHORITY, https://www. --- inra.org/about [https://perma.cc/TN3K-X8YJ] (“FINRA is not part o ---  the government. We’re a not- --- or-pro --- it organization authorized by Congress to protect America’s investors by making sure the broker-dealer industry operates  --- airly and honestly.”).
FIN. CONDUCT AUTHORITY, supra note 231, at 9 (“We could issue a NAL stating that no FCA en --- orcement action will be taken against testing activities where we are reasonably satis --- ied that the activities do not breach our requirements or harm our objectives. We think it would be appropriate  --- or the FCA to reserve the right to close the trial. The FCA’s commitment not to take en --- orcement action applies to the period  --- rom the issue o ---  the NAL until the testing is completed or closed by the FCA. The US Consumer Financial Protection Bureau (CFPB) is implementing a similar policy.”).
Id. (“In addition to NALs, the FCA can issue individual guidance to a

irm on the interpretation o — applicable rules in respect o — testing activities the

irm may be carrying out. I — the — irm acts in accordance with this guidance, it will give them certainty that the FCA would not take action against them.”). 240 Id. (“Where it is clear that testing activities do not meet our rules but the

irm can meet the waiver test and the rules are within the FCA’s power to waive, the FCA can waive or modi — y particular rules — or sandbox — irms. A waiver or modi — ication would allow what would otherwise be a temporary breach o — our rules. The FCA is limited in what it can waive by EU legislative requirements. This is not an option — or — irms not regulated under FSMA (e.g. payment institutions).”). 241 See Financial Services Innovation Act o — 2016, H.R. 6118, 114th Cong. (2016). 242 See id. The — ederal — inancial industries contemplated by this bill are: CFPB, CFTC, FCA, FDIC, FHFA, FRB, FTC, HUD, OCC, NCUA, Treasury, and SEC. AM. LAND TITLE ASS’N, FINANCIAL SERVICES INNOVATION ACT 6 (2016), https://www.alta.org/ — ile.c — m?name=McHenry-Bill-Explainer [https://perma.cc/4M7Q-M4EP]. 243 Id. at 14–15.

 Electronic copy available at: https://ssrn.com/abstract=3435274

in 2016, has not yet seen any action in the House o —

Representatives.244

Moreover, such a bill is unlikely to have a substantial impact on the disparate impact o — regulation on startups. First o — all, the McHenry Bill is limited to FinTech.245 Second, based on the length o — time that the McHenry Bill has already sat in Congress, and the time-consuming process that would be required — or an agency to process a proposal — or an alternative compliance strategy, it is hard to believe that this cumbersome process will be

ast enough to support startup innovation.

But most troublesome is the ambiguous nature o

the standards — or having a regulatory sandbox proposal approved. The McHenry Bill and other proposals — or regulatory sandboxes give a broad range o — discretion to regulatory agencies. 246 Discretion is the

unctional opposite o — certainty. Startups will have a hard time determining in advance i — or when a request will be granted. Moreover, incumbents are more able to take advantage o — this process. Large established — irms, that already have relationships with regulators, will be in a better position to get their requests approved as opposed to startups who are not — amiliar with the process. Even scholars who are proponents o — the sandbox approach recognize that incumbents could be able to game the system.247

RD teaches us that vague standards and regulatory discretion are anathema to entrepreneurial innovation. It is the large and old companies who have established relationships with regulators who are most likely to bene — it — rom sandboxes. For most startups and small businesses, however, these bogeys are regulatory sand traps.

 244 JD Alois, Financial Services Innovation Act Seeks to Mandate Innovation O ---

ices in Federal Agencies that Impact Fintech, CROWD FUND INSIDER (Oct. 21, 2019), https://www.crowd — undinsider.com/2019/10/153145-

inancial-services-innovation-act-seeks-to-mandate-innovation-o —

ices-in-

ederal-agencies-the-impact- — intech/ [https://perma.cc/RBQ2-N77J]. 245 See McHenry, Meeks Introduce Fintech Bill to Encourage Greater Financial Inclusion, U.S. CONGRESSMAN PATRICK MCHENRY (July 19, 2017), https://mchenry.house.gov/news/documentsingle.aspx?DocumentID=398631 [https://perma.cc/5TBK-7YMW]. 246 See, e.g., Allen, supra note 234, at 582. 247 Eyal-Cohen, supra note 9, at 910 (“[T]he government will need to place sa — eguards to prevent certain old-timers — rom gaming and using such regulatory sandboxes to their bene — it.”).

 Electronic copy available at: https://ssrn.com/abstract=3435274

B. Regulatory In — ormation Sharing Regimes

Several scholars have theorized that regulatory costs—especially in — ormation costs—could be mitigated by either mandatory or voluntary networks that share in — ormation about regulations. For example, “connected contracts” could lead to collaborative economic activity.248 One scholar even described such in — ormation-sharing networks in the real world. 249

But there are

undamental theoretical problems that would preclude regulatory in — ormation sharing networking — rom being a stable solution to the problem o — the disparate impact o — regulation on startups.

First and

oremost, this would necessitate sharing o — in — ormation. However, startups are in competition with each other. Knowledge about regulations would give one startup a competitive advantage over others.250 So, why would one startup voluntarily share in — ormation that was costly — or it to obtain with another competing startup? Generally, competitors do not give away their competitive advantage to rivals — or nothing. 251 Indeed, doing so could amount to corporate waste and a violation o —

iduciary duties.252

ormation i — there is some expectation o — quid pro quo. Even i — this is the case, such in — ormation sharing relationships are unlikely to be stable. Once groups o — startups reach any reasonable size, they are prone to — ree-riding problems, coordination problems, and collective action problems.253 These

G. Mitu Gulati, William A. Klein & Eric M. Zolt, Connected Contracts, 47 UCLA L. REV. 887, 894–95 (2000) (“This metaphor, called ‘connected contracts,’ emphasizes the complex interactions among all o ---  the participants in an economic venture. . . . Connected contracts broadens the scope o ---  analysis to invite attention to the cooperation, con --- lict, competition, and compromise among equity investors, lenders, managers, workers, suppliers, customers, and all others who contribute to an economic endeavor . . . .”).
See Darian M. Ibrahim, Financing the Next Silicon Valley, 87 WASH. U. L. REV. 717, 727 (2010) (“Silicon Valley is home to unique sociological networks and an open and sharing entrepreneurial culture, even among high-tech competitors.”).
See Eyal-Cohen, supra note 9, at 904 (“[R]egulatory in --- ormation is a competitive advantage that many newcomers would pre --- er not to share once obtained.”).
See, e.g., id. But see Ibrahim, supra note 249, at 727 (noting an open source trend in Silicon Valley).
ALAN PALMITER, FRANK PARTNOY, & ELIZABETH POLLMAN, BUSINESS ORGANIZATIONS: A CONTEMPORARY APPROACH 84–85 (3d ed. 2019).
See George J. Stigler, Free Riders and Collective Action: An Appendix to Theories o ---  Economic Regulation, 5 BELL J. ECON. & MGMT. SCI. 359, 359, 364–65 (1974).




 Electronic copy available at: https://ssrn.com/abstract=3435274

groups are similar to conspiracies, oligopolies, and cabals, which are notoriously unstable and prone to cheating.254

Moreover, such in

ormation sharing groups may not even be legal. Competitors sharing in — ormation may violate antitrust laws.255 I —

any competitor is le

t out o — the group, it may claim that its omission constitutes an illegal group boycott, a violation o — the Sherman Act.256 This is not mere theorizing. Such a case came be — ore the Supreme Court in 1985, when Paci — ic Stationary was ostracized — rom a group o — o —

ice suppliers called Northwest Wholesale Stationers. 257 In that case, the Supreme Court held that an in — ormation sharing group may violate antitrust law where it possesses “market power or unique access to a business element necessary — or e —

ective competition.”258 Such in — ormation about regulations could be considered a business element necessary — or e —

ective competition.259

In sum, regulatory in

ormation sharing networks are rarely observed in the world, theoretically unstable, and potentially illegal as a violation o — antitrust law i — all — irms do not have equal access to the in — ormation. There — ore, they are not a viable solution to the problem o — the disparate impact o — regulations on startups.

                    C. Tax Credits  --- or Startups

Regulation and taxation are two sides o

the same coin. They both represent a cost whose incidence may be borne by either the supplier or the consumer, depending on the ratio o — the elasticities o — supply and demand.260 Accordingly, some have advocated — or

See Stiglitz, supra note 38, at 19 & n.6.
Barak D. Richman, The Antitrust o ---  Reputation Mechanisms: Institutional Economics and Concerted Re --- usals to Deal, 95 VA. L. REV. 325, 352 (2009) (“A common  --- acilitating practice that has been  --- ound to violate the Sherman Act is an agreement between competitors to exchange in --- ormation on prices or output. Such coordination draws scrutiny because it enables illegal collusion even in the absence o ---  an explicit agreement to collude.”).
See id. at 347–48.
Nw. Wholesale Stationers, Inc. v. Pac. Stationery & Printing Co., 472 U.S. 284, 285–88 (1985).
Eyal-Cohen, supra note 9, at 903 (citing Nw. Wholesale Stationers, Inc., 472 U.S. at 290).
Id.
Who pays  --- or a tax, more  --- ormally known as the “incidence o ---  cost,” was

irst presented in 1924 by economist T.N. Carver, who asked the question, “Who bears the burden o — extra costs?” T.N. Carver, The Incidence o — Costs, 34 ECON. J. 576, 576, 588 (1924). When does the producer, the consumer, or some ratio o —

both bear the extra costs? “The general answer is, that it depends upon the commodity on which the new charge is laid, or, more speci — ically, upon the ratio

 Electronic copy available at: https://ssrn.com/abstract=3435274

either additional taxes on incumbents, 261 or a tax credit toward startups.262 This is not an article on tax policy per se, so it will provide only a brie — discussion o — why both o — these tax-based solutions are unlikely to work.

Tax burdens on incumbents generally have the impact o

increasing prices and lowering quality

or consumers.263 However, this negative externality can be countered with a positive impact, i —

such taxation simultaneously increases competition. Whether or not this is likely to occur is a very technical question that is beyond the scope o — this Article. On the one hand, taxing larger — irms could give smaller — irms a cost advantage. On the other hand, larger — irms tend to have more market power, and market power allows — irms to pass on the tax burden to consumers.264 Whether or not a tax on incumbents will increase social wel — are in highly regulated industries is a question that must be answered empirically and speci — ically in each di —

erent market.

There are practical problems with any tax regime that is designed to encourage entrepreneurial innovation. How does one determine which entities shall be taxed? Any line-drawing between the taxed and the untaxed invites — irms to change their behavior to avoid being taxed.265 This is ine —

icient in that it may require — irms to produce a suboptimal amount o — output or to engage in other gamesmanship.266

o

the elasticity o — the supply o — the commodity to that o — the demand — or it.” Id. at 578. 261 See JOSEPH W. ROSENBERG & DONALD B. MARRON, TAX POLICY CTR., TAX POLICY AND INVESTMENT BY STARTUPS AND INNOVATIVE FIRMS 23–25 (2015), https://papers.ssrn.com/sol3/papers.c — m?abstract_id=2573259 [https://perma.cc/GM8A-LL7Y]. 262 Id. at 3 (“The R&E tax credit, — or example, is intended to — oster innovative investment, and the capital gains tax exemption — or long-term investments in startups is intended to encourage equity investment in new ventures.”). 263 See, e.g., Douglas J. Young & Agnieszka Bielinska-Kwapisz, Alcohol Taxes and Beverage Prices, 55 NAT’L TAX J. 57, 58 (2002) (“The evidence — rom this study indicates that alcohol taxes are over-shi — ted to consumers. For example, the Federal excise tax on beer increased by $9 per barrel in 1991. It is estimated to have increased retail prices by $15 to $17.”). 264 See Clemens Fuest, Dominik Schober & Oliver Woll, Does Market Power Allow Firms to Pass on More o — the Tax Burden to Consumers? Evidence — rom Gasoline Tax Re — orms in Austria, 108 NAT’L TAX ANN’S ANN. CONF. ON TAXATION PROC. 1, 1 (2005), https://ntanet.org/wp- content/uploads/proceedings/2015/139- — uest-schober-woll-market-power-allow-

irms-pass.pd — [https://perma.cc/G8HK-24T7]. 265 C — . Zoë Prebble & John Prebble, The Morality o — Tax Avoidance, 43 CREIGHTON L. REV. 693, 702 (2010). 266 See, e.g., id. at 707.

 Electronic copy available at: https://ssrn.com/abstract=3435274

Moreover, the utility o — tax credits is limited — or startups whose business model does not contemplate being pro — itable — or several years.267 Tax credits — or startups are only use — ul when a startup is incurring tax liability.268 Tax liability arises — rom pro — its, but startups are o — ten unpro — itable in their — ormative years.269 Some “startups” remain unpro — itable — or over a decade and remain unpro — itable even when they have grown so large as to be better seen as incumbents. 270 Indeed, some have opined that the hallmark o — a high-growth startup is its ability to channel all potential pro — its into accelerating growth.271

Accordingly, whether or not compensatory tax instruments will encourage entrepreneurial innovation remains an open question. On the one hand, it is theoretically possible to use taxes to discourage monopoly and encourage competition. 272 On the other hand, calculating the appropriate tax rate is both an art and a science. We might be skeptical that regulators or lawmakers who impose these tax policies design e —

icient equations, especially where they may be in — luenced—or even captured273—by the incumbents in that regulated industry. 274

See Susan C. Morse & Eric J. Allen, Innovation and Taxation at Start-up Firms, 69 TAX L. REV. 357, 357–59 (2016).
See id.
See id.
See, e.g., Seth Fiegerman, Twitter Records Its First Annual Pro --- it, but It Is Losing Millions o ---  Users, CNN BUS.: MKTS NOW (Feb. 7, 2019, 1:20 PM), http://cnn.com/2019/02/07/tech/twitter-earnings-q4/index.html [https://perma.cc/24H2-WPKQ]. For example, Twitter turned its  --- irst annual pro --- it in 2018, eleven years a --- ter it was incorporated on April 19, 2007. See id.; Restated Certi --- icate o ---  Incorporation o ---  Twitter, Inc., U.S. SEC. & EXCHANGE COMMISSION, https://www.sec.gov/Archives/edgar/data/1418091/000119312513390321/d5640 01dex31.htm [https://perma.cc/2YCM-E4MQ].
See Alison Coleman, Pro --- itability Versus Growth: A Balancing Act, FORBES (Mar. 4, 2018, 1:20 PM) https:// --- orbes.com/sites/alisoncoleman/2018/03/04/pro --- itability-versus-growth- a-balancing-act- --- or-startups/#6 --- 794a1d3e3e [https://perma.cc/Z5UL-HA4G].
But see Casey Mulligan, Monopolies Are Unhealthy but High Taxes Make the      Disease      Worse,     HILL       (Mar.     30,     2018,      2:00     PM), https://thehill.com/opinion/ --- inance/381001-monopolies-are-unhealthy-but-taxes- make-the-disease-worse [https://perma.cc/K2YG-SVUG].
Regulatory capture is a state o ---  a ---

airs where a lesislature or administrative agency regulates in an industry’s interest, as opposed to regulating in the public interest. See Alan Schwartz & Robert E. Scott, The Political Economy o — Private Legislatures, 143 U. PA. L. REV. 595, 644 (1995) (A lawmaker is captured when it chooses “a policy which would not be rati — ied by an in — ormed polity — ree o —

organization costs.”), citing Michael Levine & Jenni

er Forrence, Regulatory Capture, Public Interest, and the Public Agenda: Toward a Synthesis, 6 J.L. ECON. & ORGANIZATION 167, 178 (1990) 274 See, e.g., supra notes 105–107 and accompanying text.

 Electronic copy available at: https://ssrn.com/abstract=3435274

D. Deregulation?

Deregulation is an obvious solution to the problems o

regulation. O — course, regulations exist — or theoretical reasons such as

ostering public interests and en — orcing market discipline, although these theories have also been questioned in the literature.275 It is beyond the scope o — this Article to discuss whether regulation or deregulation is generally pre — erable. However, this Article can shed light on the debate, as its — inding o — the emergence o — RD impacts analysis o — the cost o — regulation.

There are two main schools o

thought when it comes to deregulation. The more extreme school advocates — or total deregulation.276 The more moderate school advocates — or deregulating small — irms. This moderate regulatory philosophy was expressed in legislation through the Small Business Regulatory En — orcement Fairness Act.277 Government regulators o — ten exclude or exempt small businesses — rom a number o —

regulations.278 These exemptions include securities registration

 275 See, e.g., Andrei Shlei --- er, Understanding Regulation, 11 EUR. FIN. MGMT. 439, 443 (2005). Regulation versus deregulation o ---  businesses is a broad debate that connects in some ways to the even larger issue o ---  the trade-o ---

between dictatorship and disorder. Whether regulations increase or decrease social wel — are may be a — unction o — the nature o — the activity being regulated and the characteristics o — the society in which the regulation takes place. See id. Moreover, there is scholarly disagreement about whether e —

iciency is even the right metric to use when evaluating whether more or less regulation is desirable. Other metrics may include justice, stability, or perceptions o —

airness. See id. 276 See, e.g., Richard A. Epstein & David A. Hyman, Fixing Obamacare: The Virtues o — Choice, Competition, and Deregulation, 68 N.Y.U. ANN. SURV. AM. L. 493, 516 (2012). 277 See 5 U.S.C. § 601 (2018). Some scholars have written speci — ically on the misguided nature o — that act. See Richard J. Pierce, Jr., Small Is Not Beauti — ul: The Case Against Special Regulatory Treatment o — Small Firms, 50 ADMIN. L. REV. 537, 538 (1998). 278 See Mirit Eyal-Cohen, Down-Sizing the “Little Guy” Myth in Legal De — initions, 98 IOWA L. REV. 1041, 1065–86 (2013) (outlining small business deregulation in various areas o — the law).

 Electronic copy available at: https://ssrn.com/abstract=3435274

and reporting requirements, 279 labor and employment rules,280 as well as health and sa — ety guidelines. 281

But such small business exemptions do not solve the problem. For one thing, small businesses must expend costs to determine whether the exclusionary rule applies to them. 282 As these exclusionary rules change, small — irms must continue to expend resources to stay abreast o — those changes. Third parties must also expend resources to determine whether a — irm they are dealing with is regulated or exempt.283

Perhaps more problematically, small

irms may change their behavior in otherwise ine —

icient ways in order to — it into the exemption.284 For example, a small business may choose not to hire an additional worker in order to stay under the limit — or mandatory provision o — health insurance. 285 Likewise, many — irms hire part-time employees to work thirty hours per week or less to

See, e.g., Emerging Growth Companies, U.S. SEC. & EXCHANGE COMMISSION, (July 24, 2019), https://sec.gov/smallbusiness/goingpublic/EGC [https://perma.cc/9S5W-MYUC] (“I ---  your company quali --- ies as an ‘emerging growth company’ as de --- ined in Section 2(a)(19) o ---  the Securities Act, it may choose to  --- ollow disclosure requirements that are scaled  --- or newly public companies.”).
See, e.g., Family Help, INTERNAL REVENUE SERV. (Jan. 16, 2020), https://irs.gov/businesses/small-businesses-sel --- -employed/ --- amily-help [https://perma.cc/PPU2-3CYN (describing Federal Unemployment Tax Act (FUTA) exemptions  --- or small  --- amily-run businesses).
See, e.g., 29 C.F.R. § 1904.1 (2019) (describing OSHA rule exemptions

or employers with less than ten employees). 282 C — ., e.g., Allen Smith, Weigh Risks in Using FFCRA’s Small-Business Exemption, Soc’y — or Hum. Resource Mgmt. (May 1, 2020), https://www.shrm.org/resourcesandtools/legal-and-compliance/employment- law/pages/coronavirus- —

cra-small-business-exemption.aspx [https://perma.cc/7239-GGX6]. 283 See C. Steven Brad — ord, Does Size Matter? An Economic Analysis o —

Small Business Exemptions

rom Regulation, 8 J. SMALL & EMERGING BUS. L. 1, 25 (2004) (“[E]xemptions sometimes increase the in — ormation costs o —

unregulated third parties who may need to distinguish between regulated and unregulated — irms.”). 284 See id. (“Exemptions also encourage waste — ul strategic behavior by — irms seeking to avoid regulation.”). 285 See William E. Even & David A. Macpherson, The A —

ordable Care Act and the Growth o — Involuntary Part-Time Employment, 72 INDUS. & LAB. REL. REV. 955, 955–56 (2019) (“At its passage, the 2010 A —

ordable Care Act (ACA) required that — irms with 50 or more employees provide health insurance — or their

ull-time workers or be subjected to penalties beginning in 2014. Many analysts argued that the law created incentives — or large — irms to shi — t — rom — ull-time to part-time workers to escape the penalties and the cost o — providing health insurance.”).

 Electronic copy available at: https://ssrn.com/abstract=3435274

avoid paying — or employee bene — its. 286 Some have even argued that this arti — icial bi — urcation between regulated and exempt — irms is an immoral double standard. 287

Others have argued that the seemingly moderate position o

deregulating small

irms is — oolish; rather, they argue that the better approach is a sweeping reduction o — regulations — or all — irms, which can best be summed up in the phrase “Deregulate Now!” 288 It is simply beyond the scope o — this Article to address in any meaning — ul way the arguments — or and against sweeping deregulation. O — course, deregulation would eliminate many o — the problems with regulations. However, many would argue that regulations exist — or a reason. Others would counter-argue that those reasons are primarily based on power and politics.

For the instant purposes o

this Article, it is not necessary to resolve the regulation versus deregulation debate writ large. Rather, the Article’s contribution is that not all regulation is created equal. Granting broad discretion to regulators through vague standards may have more negative consequences — or

 286 Kara E Shae, Risks o ---  ACA Avoidance Strategies  --- or Employers, HR DAILY ADVISOR (Nov. 1, 2013), https://hrdailyadvisor.blr.com/2013/11/01/risks- o --- -aca-avoidance [https://perma.cc/EG8D-4FXE] (“Already, several ACA avoidance strategies, including layo ---

s, downsizing . . . moving employees — rom

ull- to part-time status, and replacing employees with a contract work — orce, have made headlines.”). Under ObamaCare, the hourly threshold to be considered “ — ull-time” is thirty hours. See Cynthia J. Borrelli, A —

ordable Care Act Compliance — rom the Employer’s Perspective, N.J. LAW., Apr. 2016, at 35, 35, (“A — ull-time employee is an individual employed at least 30 hours per week, on average.”). 287 See, e.g., Ruben H. Arredondo, Di —

erent Strokes — or Di —

erent Folks: Balancing the Treatment o — Employers and Employees in Employment Discrimination Cases in Courts within the Tenth Circuit Court o — Appeals, 16 BYU J. PUB. L. 261, 285 (2002) (“As a practical matter, state and — ederal legislation have di —

erent protections — or workers employed by small businesses and those employed by larger employers, which is essentially a double standard. As a policy matter, it sends a message that though elimination o — workplace discrimination is important — or some o — the work — orce, it is not important enough to cover the entire work — orce.”); Seth C. Oranburg, Unbundling Employment: Flexible Bene — its — or the Gig Economy, 11 DREXEL L. REV. 1, 3–4 (2018). 288 See Epstein & Hyman, supra note 276, at 495 (The “[Patient Protection and A —

ordable Care Act]’s [(PPACA)] — undamental design de — ect was to superimpose additional layers o — regulation and subsidies on a system that was already top-heavy with both. These preexisting regulations and subsidies have already misaligned the incentives within the health care system. The next generation o — rules will only compound the errors. In our view, the right approach to these problems is to promptly initiate a program o — systematic deregulation that will introduce the choice and competition to which PPACA gives, at best, lip service.”); Richard A. Epstein, Deregulate Now, HOOVER DIGEST (Apr. 21, 2010), https://www.hoover.org/research/deregulate-now [https://perma.cc/5N3M-9KQE].

 Electronic copy available at: https://ssrn.com/abstract=3435274

entrepreneurial innovation than instituting rules that apply equally and straight — orwardly to all — irms. According to RD, deregulatory e —

orts should — ocus on identi — ying regulations where regulators have too much discretion.

                             CONCLUSIONS

Most scholars agree that regulation is at odds with innovation. 289 Some have even described innovation as a square peg that does not

it in the round hole o — regulation. 290 But this Article hoped to show that not all regulations are created equal. A more nuanced view o — regulatory regimes, using the axes o — rules versus standards and simplicity versus complexity, can shed light onto the design o — optimal regulations.

This Article has shown how startups are creating RD technologies that help other startups and small businesses comply with regulations. This changes the regulatory landscape by leveling the playing — ield — or entrepreneurial innovation. Future studies o — the impact o — regulation should consider how RD may change the incidence o — regulatory cost on startups and small businesses.

Moreover, this positive observation leads to normative policy suggestions. In particular, regulators should consider the nature o —

the regulation in addition to the amount o

regulation. Regulations that have a vague, standard-like nature are more likely to have a disparate impact on small businesses, whereas regulations that can be addressed by computers may not depress entrepreneurial innovation as much.

Entrepreneurs

ace many barriers to entry, but regulators should be particularly concerned about regulatory barriers to entry, as this is something ostensibly in their direct control. I — business regulations can be designed more e —

iciently, should they not be? This Article hoped to show how regulatory rules might be better than regulatory standards inso — ar as rules are more compatible with technological solutions, such that rule makers should try to cra — t rules that can be addressed by RD solutions.

 289 See, e.g., Wul ---  A. Kaal & Robert N. Farris, Innovation and Legislation: The Changing Relationship—Evidence  --- rom 1984 to 2015, 58 JURIMETRICS J.L., SCI., & TECH. 303, 304 (2017).
 290 See Mark Lavender, Regulating Innovative Medicine: Fitting Square Pegs in Round Holes, 2005 DUKE L. & TECH REV. 1, 2, https://dltr.law.duke.edu/2005/01/13/regulating-innovative-medicine- --- itting- square-pegs-in-round-holes/ [https://perma.cc/6KPJ-3DKU].

 Electronic copy available at: https://ssrn.com/abstract=3435274

I termed this process regulatory democratization because democratization is the action o — making something accessible to everyone.291 Likewise, RD — irms make regulatory compliance accessible — or a wider range o — startups and small businesses. Lowering the cost o — regulatory compliance makes it easier — or more businesses to enter regulated markets. RD — irms lower the cost o — regulatory compliance by creating new technological products and services that solve regulatory problems — or a large number o — other startups and small businesses. Theoretically, the result o — this process is a more competitive and there — ore e —

icient market, despite a high degree o — regulation.

This

inding counsels against the idea that regulatory ine —

iciency should be remedied with additional regulatory apparatuses.292 A priori, the notion that more regulations will solve the problems caused by regulations does not make sense.293 Moreover, this Article has shown that increasing regulatory discretion—via regulatory sandboxes—is likely to — urther depress entrepreneurial innovation — rom small and young — irms.

Indeed, it may be the case that the best regulatory solution is no solution. RD occurs where the — ree market, powered by technology, solves regulatory problems on its own.

This theory should be tested empirically. Fortunately, on the horizon are new tools — or measuring the impact o — regulation. 294 This Article theorized that rules and standards may have a di —

erent impact on regulation, especially as a — unction o —

irm size and age, and these variables should be considered in empirical analysis o —

regulation.

There is much more to learn and understand about the relationship between regulation and entrepreneurial innovation. The — inding o —

RD shows that regulations have complex impacts on startup

See Democratization, OXFORD ENGL. DICTIONARY (3d ed. 2014).
See Eyal-Cohen, supra note 9, at 886–87 (“Our economy and society as a whole lose when innovations are placed on hold or barred  --- rom entering the marketplace. It re --- lects poorly on the government when its action might be one o ---  the reasons  --- or this outcome.”).
See id. at 865.
See Michael Simkovic & Miao Ben Zhang, Measuring Regulation 1 (Univ. o ---  S. Cal. Law Sch., Legal Studies Working Paper No. 298, 2019), https://law.bepress.com/cgi/viewcontent.cgi?article=1433&context=usclwps-lss [https://perma.cc/B2VX-C9KQ]. For example, Michael Simkovic and Miao Ben Zhang developed a new method to measure the intensity o ---  regulation. See id. The availability o ---  their Regulation Index and others’ e ---

orts to make regulatory impact quanti — iable may help researchers conduct empirical analysis on the impact o — regulation on entrepreneurial innovation. See id.

 Electronic copy available at: https://ssrn.com/abstract=3435274

decision-making, but what are the individual, environmental, and industry conditions that will impact an entrepreneur’s decision to create a RD solution? How does RD impact entry, — orm, and per — ormance in regulated markets? What is the impact o —

are? The Author hopes this Article inspires other scholars to challenge the ideas stated herein and move us closer to a theory o — optimal regulation.

                            ***

Electronic copy available at: https://ssrn.com/abstract=3435274