By Dr. Thibault Schrepel
This short article is derived from Thibault Schrepel, Collusion by Blockchain and Smart Contracts,33 Harv. J.L. & Tech. 117 (2019), available at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3315182.
Antitrust and competition law is subject to increasing polarization, especially regarding Internet giants. Some scholars argue that the tech giants abuse their dominant positions, while others point to the increase they contribute to consumer welfare. These two camps seem irreconcilable, mainly because they are not discussing the same thing.
There is, however, a broad consensus on antitrust and competition policy in the area of collusion (Section 1 of the Sherman Act in the United States, Article 101 of the TFEU in Europe). Few economists and lawyers would defend the beneficial effects of anti-competitive collusion for consumers. In fact, cartels are regularly described as being “the supreme evil of antitrust,” a quasi-moralist assertion on which the academic community seems to have found a point of agreement. On that basis, algorithmic collusion is now the focus of much academic research. This article ambitions to explain, in the absence of fundamental challenges for antitrust law, why it actually results from a publication bias (I), and why the focus should be put instead on blockchain based-collusion (II).
I. Algorithmic collusion: Not a Fundamental Issue
The publication of Virtual Competition in 2016 was a pivotal moment that began putting emphasis on the alleged danger of algorithmic collusion. In it, antitrust and competition law is described as powerless in the face of practices organized by computers, some of which are using artificial intelligence whose impact is sometimes described as “the biggest risk that we face as a civilization.” A semblance of academic consensus has formed around the need for antitrust and competition authorities to focus their efforts on this new supreme – “supreme evil,” namely, algorithmic collusion. Nevertheless, algorithmic collusion is a fundamentally unimportant subject for antitrust and competition law, for at least two reasons.
Lack of conclusive empirical studies
Algorithmic collusion is the subject of a growing literature, yet, empirical studies documenting the frequency of the phenomenon in the real-world remain to be produced. One cannot find any quantification of algorithmic collusion in official publications coming from antitrust and competition agencies, in any of the reports given to these agencies, or in the OECD publications. When having a look at the litigation brought in the U.S. and in Europe, algorithmic collusion is virtually non-existent.
For that reason, the priority is to first quantify the phenomenon rather than to propose drastic changes to antitrust and competition law. The academic community must play its part, but for the time being, the popularity of algorithmic collusion as a topic is a publication bias–in which this article (partially) takes part.
Old wine in new bottles
Even if algorithmic collusion were already, or could become, a frequent practice, it would remain old wine in new bottles. Indeed, such collusion is simply a more elegant way of implementing the same practices known for centuries. Whether they are algorithmic or not, the nature of anti-competitive collusion remains identical. In fact, algorithmic collusion is at the origin of two non-fundamental problems: detection, and liability assignment.
In terms of detecting illegal practices, algorithms could enable faster implementation of agreements between companies, potentially for only a few seconds. In terms of liability assignment, one may note the difficulties that arise in situations where the algorithm, whose initial order was to maximize the company’s profits, has decided on its own to implement an anti-competitive practice.
In both cases, solutions are already emerging. Part of the doctrine underlines that algorithms may be used by competition authorities to detect algorithmic collusion. Robots Computer war is not far off. As for the assignment of liability, there is no doubt as to lawyers’ ability to distribute it, whether to the company that is using the algorithm, to the company that created it, to individuals or to the algorithm itself.
To sum up, algorithmic collusion is not yet quantified, and even if it was, it would not raise fundamental problems for antitrust and competition law. The law will respond to relevant issues in due course. Antitrust and competition agencies would, therefore, be well advised to focus their resources where consumer harm can already be quantified. Unfortunately, the publication bias that pushes part of the scientific community to publish on the subject of algorithmic collusion creates a headwind and leads authorities to misdirect these resources.
II. Blockchain-Based Collusion: A Fundamental Issue
Contrary to what can be said for algorithmic collusion, taking a closer look at blockchain-based collusion is urgent, precisely because it creates fundamental issues for antitrust and competition law.
Blockchain is a technology allowing for different layers to be superimposed on top of each other. The first level is generally described as the “platform layer.” It is a database with the following characteristics: decentralized, pseudonymous, immutable, and unstoppable. The second level, called the “software layer,” is then added on top of it. All kinds of applications can operate on the basis of the characteristics and data contained within the first level.
One of these applications, smart contracts, is of particular interest to anti-competitive agreements. A smart contract is a potential transaction that is recorded in a blockchain and will be automatically executed if and when several conditions are met. It can be the automatic sending of a sum of money when a plane or train is delayed for more than an hour, the unlocking of an apartment’s door rented on Airbnb when the amount is paid into the owner’s account, or, an agreement between companies whose governance follows a combination of smart contracts.
Empirical studies and capacity for action
As with algorithmic agreements, the frequency of blockchain-based agreements (whether or not they involve smart contracts) is yet to be quantified. Unlike algorithmic agreements, however, antitrust and competition law as we know it today will be mostly ineffective the day blockchain-based agreements are documented, hence the importance of studying the issue without further ado. Indeed, such agreements meet the characteristics of the blockchain’s first layer, on top of which they operate. These agreements are decentralized, the identity of its participants is unknown, and above all, they cannot be altered or stopped by any user of the blockchain. For these reasons, it is essential to address the issue of blockchain-based collusion before the phenomenon becomes apparent. The applicability of antitrust and competition law is at stake.
Fundamentally new types of collusion
Unlike algorithmic collusion, blockchain-based collusion, particularly when it involves the use of smart contracts, is fundamentally different in nature from collusion made without the support of this technology. There are two reasons for this.
The first relates to the non-cooperative nature of collusion carried out outside blockchains, while blockchain-based collusion is cooperative. When several companies decide to collude, they must consider two perspectives: one economic, one social. They need to ensure that they have more to gain economically by staying in the agreement than by leaving it. They also need to make sure they can trust each other to avoid the denunciation of the collusion by one of them.
To the extent that anti-competitive agreements are illegal, colluders cannot deploy legally binding agreements to ensure their implementation and strengthen economic or social perspectives. The strategy of other colluders is, by nature, unpredictable. They act in their own interests, making the stability of agreements highly dependent on each’s interests. For this reason, collusion is said to be non-cooperative.
Blockchain is transforming collusion into a cooperative game. Smart contracts automatically execute any agreement when triggered. They can, for instance, be used to set a collusive price, to share markets, and to punish any deviation from such agreement. Specifically, the selling of a product at a price different than the one agreed upon by colluders could be recorded automatically into the blockchain by way of smart contracts, and these smart contracts could punish the behavior upon several conditions (duration, spread with the collusive price… ). As smart contracts are immutable, no colluders can change the set governance.
One may see why a vast majority of deviant behaviors would be eliminated. Blockchain and smart contracts would, as such, strengthen the trust colluders place in each other, and therefore, economic and social stability. In a nutshell, by allowing the implementation of agreements whose constraint stems from cryptographic rules, blockchain transforms non-cooperative games (collusion) into cooperative ones. Game theorists often ignore the importance of the medium on which the games (collusion) are being played, but it is nonetheless essential.
The second reason why blockchain-based collusion is fundamentally different from others is that blockchain-based collusion is dynamic. The Oxford Dictionary defines the word “dynamic” as “a force that produces change, action or effects.” By way of illustration, algorithmic collusion is not dynamic. When an algorithm is set up, it identifies patterns (after being taught so) and runs accordingly, therefore following a linear and predictable learning curve. A physical person (a force) can only intervene to stop that algorithm and introduce a new version of it. Algorithmic collusion is, therefore, not dynamic by nature.
Blockchain-based collusion, on the contrary, can be genuinely dynamic when a smart contract uses a decentralized application (D-App) as a clause. Any software could be turned into a D-App and be implemented into a smart contract, creating an almost infinite number of evolving possibilities for antitrust and competition law infringements. It results that blockchain-based collusion is immutable by nature, which creates trust, but is also adaptable since D-Apps give colluders flexibility to bring external forces to the same contract. As a result, blockchain-based collusion is truly dynamic, as it allows third machines (D-Apps) or physical persons to change its trajectory while maintaining its initial existence. Here again, blockchain changes the nature of collusion, which should conduce to updating the law that regulates it.
III. Some conclusive thoughts
Antitrust and competition agencies must address the issue of blockchain-based collusion. In practice, it means that agencies have the duty to tackle competitive issues whose effects manifest already on the market (very often, far from digital services, but in the field of transport, energy, etc., and where companies are protected by the State), and at the same time, to invest in forward-looking issues that are fundamentally challenging antitrust and competition law. Algorithmic collusion has none of that. Blockchain-based collusion does.
Dr. Thibault Schrepel, LL.M. is an Assistant Professor of Antitrust at Utrecht University and a Faculty Associate at the Berkman Klein Center for Internet & Society at Harvard University.
This article was also posted on JOLT Digest: https://jolt.law.harvard.edu/digest/the-fundamental-unimportance-of-algorithmic-collusion-for-antitrust-law.
 Assistant Professor at Utrecht University School of Law, Faculty Associate at Harvard University’s Berkman Klein Center for Internet & Society, Associate Researcher at University of Paris 1 Panthéon-Sorbonne and Invited Professor at Sciences Po Paris. No outside funding was received or relied upon for this short article.
 See, for example, Robert D. Atkinson & Michael Lind, Big Is Beautiful: Debunking the Myth of Small Business (2018), and Tim Wu, The Curse of Bigness: Antitrust in the New Gilded Age (2018) for polarized works released just a few weeks apart.
 Verizon Commc’ns v. Law Offices of Curtis V. Trinko, 540 U.S. 398, 408 (2004).
 For a critique of the moralization of antitrust and competition law, see Thibault Schrepel, Antitrust Without Romance, 13 N.Y.U. J.L. & Liberty (forthcoming 2020).
 Google Scholar lists 141 academic articles discussing “algorithmic collusion” since 1st January 2017. See Search for Algorithmic Collusion, Google Scholar, http://scholar.google.com (search field for “algorithmic collusion”) [https://perma.cc/RLM2-SS7F].
 Ariel Ezrachi & Maurice E. Stucke, Virtual Competition: The Promise and Perils of the Algorithm-Driven Economy (2016). This book is, arguably, the best-written antitrust law essay of the past decade.
 Jamie Condliffe, Elon Musk Implored Lawmakers to Prevent People from Building AI that Could Destroy Us All, MIT Tech. Review (July 17, 2017), https://www.technologyreview.com/s/608296/elon-musk-urges-us-governors-to-regulate-ai-before-its-too-late [https://perma.cc/QF3K-9GUW].
 For even more SUPREME, see The Dictator (Paramount Pictures 2012), a movie about The Supreme Leader of Wadiya.
 See Org. for Economic Co-operation and Development [OECD], Algorithms and Collusion — Note from the European Union, DAF/COMP/WD(2017)12 (June 4, 2017), https://one.oecd.org/document/DAF/COMP/WD(2017)12/en/pdf [https://perma.cc/Q8SS-6QRJ]; Margrethe Vestager, Comm’r, Eur. Comm’n, Algorithms and Competition (Mar. 16, 2017), https://wayback.archive-it.org/12090/20191129221651/https://ec.europa.eu/commission/commissioners/2014-2019/vestager/announcements/bundeskartellamt-18th-conference-competition-berlin-16-march-2017_en [https://perma.cc/YES2-4GMJ]; Bundeskartellamt & Autorité de la Concurrence, Algorithms and Competition (Nov. 2019), https://www.autoritedelaconcurrence.fr/sites/default/files/algorithms-and-competition.pdf [https://perma.cc/CD5K-GRYK].
 See, e.g., Jacques Crémer, Yves-Alexandre de Montjoye & Heike Schweitzer, Competition Policy for the Digital Era: Final Report, European Commission (Feb. 2019) https://ec.europa.eu/competition/publications/reports/kd0419345enn.pdf [https://perma.cc/RS4G-MVTR]; UK Digital Competition Expert Panel, Unlocking Digital Competition (Mar. 2019), https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/785547/unlocking_digital_competition_furman_review_web.pdf [https://perma.cc/7GUU-6ABP].
 See, e.g., Org. for Economic Co-operation and Development [OECD], Summary of Discussion of the Roundtable on Algorithms and Collusion, DAF/COMP/M(2017)1/ANN2/FINAL (Sept. 26, 2018), https://one.oecd.org/document/DAF/COMP/M(2017)1/ANN2/FINAL/en/pdf [https://perma.cc/XX27-K7BJ].
 A search on the WestLawNext search engine, using “adv: “algorithms” AND “Sherman Act § 1″”, only brings up 13 cases, hardly any concern actual algorithmic collusion, but rather, non-algorithmic collusion in which one or two companies were using algorithms in their business model. A similar search, using “adv: “algorithmic collusion”,” brings up zero cases.
 For an emphasis on the need for such empirical work, see Thibault Schrepel, Here’s Why Algorithms Are Not (Really) a Thing, Revue Concurrentialiste (May 15, 2017), https://leconcurrentialiste.com/2017/05/15/algorithms-based-practices-antitrust/ [https://perma.cc/4WA4-R76Y]. For now, legal and economic papers dealing with the subject adopt an “experimental approach.” See, e.g., Nan Zhou, Li Zhang, Shijian Li & Zhijian Wang, Algorithmic Collusion in Cournot Duopoly Market: Evidence from Experimental Economics, Cornell University arXiv:1802.08061 (Feb. 21, 2018).
 See Sam Schechner, Why Do Gas Station Prices Constantly Change? Blame the Algorithm, Wall St. Journal (May 8, 2017, 6:41 PM), https://www.wsj.com/articles/why-do-gas-station-prices-constantly-change-blame-the-algorithm-1494262674 [https://perma.cc/BD9J-Y5R3]. Generally speaking, the main challenge posed by the digital economy to antitrust and competition agencies is the one of speed. See Richard Posner, Antitrust in the New Economy, 68 Antitrust L.J. 925, 925 (2001) (“[T]he enforcement agencies and the courts do not have adequate technical resources, and do not move fast enough, to cope effectively with a very complex business sector that changes very rapidly.”).
 See Org. for Economic Co-operation and Development [OECD], Algorithmic Collusion: Problems and Counter-Measures — Note from Ariel Ezrachi & Maurice E. Stucke, DAF/COMP/WD(2017)25, 97 (May 31, 2017), https://www.oecd.org/officialdocuments/publicdisplaydocumentpdf/?cote=DAF/COMP/WD%282017%2925&docLanguage=En [https://perma.cc/3EA9-RZR7] (explaining that algorithms may, with no human input, decide on implementing illegal practices to maximize profits).
 See Foo Yun Chee, EU Considers Using Algorithms to Detect Anti-Competitive Acts, Reuters (May 4, 2018, 7:18 AM), https://www.reuters.com/article/us-eu-antitrust-algorithm/eu-considers-using-algorithms-to-detect-anti-competitive-acts-idUSKBN1I5198 [https://perma.cc/BP3N-4USS]; see also Michal Gal, Algorithms as Illegal Agreements, 34 Berkeley Tech. L.J. 67, 115 (2019).
 See generally Thibault Schrepel, Collusion by Blockchain and Smart Contracts, 33 Harv. J.L. & Tech. 117 (2019).
 Schrepel, supra note 16, at 122.
 Id. at 119-122.
 Let’s get something out of the way. Smart contracts can certainly be described as a set of operating rules specific to a calculation, i.e., algorithms. However, this is not what the articles dealing with algorithmic collusion are aiming at. This literature focuses on the simple use of a computer, without mentioning the technology or the type of platform being used. See Ariel Ezrachi & Maurice E. Stucke, Virtual Competition: The Promise and Perils of the Algorithm-Driven Economy 42 (2016). Unfortunately, studying algorithmic agreements without taking their medium into account is equivalent to analyzing the market for smartphone apps without taking into account how operating systems work. Such an analysis is incomplete and runs the risk of being unproductive, on that. See Schrepel, supra note 16, at 126.
 For a description of smart contract combination through smart contract libraries, see Schrepel, supra note 16, at 143.
 Id. at 120.
 For an explanation of the economic perspective, see Robert C. Marshall & Leslie M. Marx, The Economics of Collusion: Cartels and Bidding Rings 108 (2012). For an explanation of the social perspective, see J. D. Jaspers, Managing Cartels: How Cartel Participants Create Stability in the Absence of Law, 23 Eur. J. on Crim. Pol’y & Res. 319, 322 (2017).
 Schrepel, supra note 16, at 124.
 Id. at 147.
 Dynamic, Oxford Advanced American Dictionary, https://www.oxfordlearnersdictionaries.com/definition/american_english/dynamic_1 [https://perma.cc/76P9-HL74].
 For a more detailed explanation of how the D-Apps function, see Chainlink: Linking Smart Contracts with Real-World Applications, State of the Dapps (Sept. 4, 2019) https://www.stateofthedapps.com/fr/spotlights/chainlink-linking-smart-contracts-with-real-world-applications [https://perma.cc/BQ5S-DT99].
 See Schrepel, supra note 16, at 122.
 See Schrepel, supra note 4.
 See Schrepel, supra note 16, at 160 (making recommendations to antitrust and competition agencies to tackle blockchain-based collusion).