Sylvia E., Matthew Krauter, Sailaja Nallacheruvu, Ksenia Polyarskaya, and Ashank Tomar
Vol. IV, No. 2
PART I
I. Introduction & Background
The massive proliferation and integration of new technologies have made technology an integral and necessary part of modern-day society’s infrastructure. Though these technologies have benefits such as increasing economic efficiency, improving health outcomes, and reducing errors in day-to-day tasks, they often come with harms that have been difficult for policymakers to regulate.
In the following sections, we overview four areas of technological regulation in the United States. In the first two subsections, we will discuss the advent of cryptocurrency and artificial intelligence to illustrate the difficulties associated with regulating new, or “breakthrough” technologies. In the third and fourth subsections, we will discuss the difficulties posed by Big Tech in data privacy and antitrust regulation.
II. A Look into Crypto: an Analysis of Tech Regulations
A. Background
Cryptocurrency, NFTs, and Decentralized Financial Assets have been the talk of popular media for the past decade. From the emergence of Bitcoin in 2008 to the second-wave revolutionary technology Ethereum, the crypto-world has been on the rise, especially in blockchain markets. As of April 6, 2022, the total market value of circulating Cryptocurrency, NFTs, and DeFi was at 1.99 trillion dollars, 10.25 billion dollars, and 150.11 billion dollars respectively. [1] As the market value continues to rise, an increasing number of investors, policymakers and government officials have tried to get in on the craze. What is it about this technology that makes it so popular, and does it need to be regulated?
B. Bitcoin
Bitcoin was developed by an anonymous entity recognized by the name of Satoshi Nakamoto. Following the distress of the Great Recession, the world experienced an unprecedented distrust in banks and their function in the financial system. In the 2008 release of the white paper titled “Bitcoin: A Peer-to-Peer Electronic Cash System”, Nakamoto urged for an “electronic payment system based on cryptographic proof instead of trust, allowing any two willing parties to transact directly with each other without the need for a trusted third party.” At the time, this was a revolutionary concept because people were not able to conduct online transactions without a bank connected to a credit card or quasi-bank applications like Paypal.
As a decentralized asset, people using Bitcoin are able to complete financial transactions without the need for a middle-man with added transaction costs. As a bank “proofs” their customer’s transactions, the hand-off of Bitcoin is checked via a special authentication process. As described in the white paper, “each owner transfers the coin to the next by digitally signing a hash of the previous transactions and the public key of the next owner and adding these to the end of the coin. A payee can verify the signatures to verify the chain of ownership.” In easier terms, the transaction is verifiable via a blockchain and the bitcoin gets put into a personal “wallet” that is heavily proofed via authentication codes and other security measures.
Traditionally, a bitcoin is presented as a tangible gold coin, yet Nakamoto describes it as “a chain of digital signatures”. A person is able to buy bitcoin via credit card payment, PayPal, Apple Pay/Google Pay, face-to-face, bartering, etc. It just depends on the preference and convenience of the purchasing party. The process that maintains the bitcoin network and aids the creation of new coins is called mining. Any given person can choose to complete special cryptographic calculations that help with the creation of new “blocks”. As an incentive, they get rewarded with a particular amount of bitcoin.
What sets this particular currency apart from the rest on the market is its cap at 21 million bitcoins. There will never be more in the circulating market [2]. The first notable use of Bitcoin is thought to be recorded in May of 2010 when a user paid 10,000 bitcoin for pizza delivery. In 2022, that amount of bitcoin is approximately worth 395 million dollars. As it became widely available via trading across the globe, Bitcoin has now matured into a major financial instrument, whereas before it was considered a farce.
C. Ethereum
Although Bitcoin’s main goal is to be an equivalent of “hard money”, blockchain technology use does not end there. In 2013, a Russian programmer Vitalik Buterin came up with the idea of Ethereum, a decentralized, open-source blockchain with smart contracts functionality. What is different about Ethereum is the fact that it is programmable, meaning each person can find a use in it beyond transacting the currency. Ethereum is not only a blockchain that can verify peer-to-peer transactions, but something that smart contracts can be built on top of. These “digitize agreements by turning the terms of an agreement into computer code that automatically executes when the contract terms are met.”
D. dApps
This composable technology allowed for the conceptualization of decentralized applications/projects, also known as dApps. Examples of dApps that are popular today include DeCentralized Finance applications (DeFi), Non-Fungible Tokens (NFTs), and Decentralized Autonomous Organizations (DAOs). Decentralized applications allow for programs and applications to exist on the blockchain without a single entity’s control. For users, there are numerous benefits to this system. While using decentralized applications, the user does not need to provide real-world identity, meaning their personal information can be safeguarded. Additionally, since the data is stored on the blockchain, there is a smaller chance of forgery or any type of fraudulent transaction since the information about the transaction will be visible via a public ledger. Lastly, smart contracts and decentralized applications are typically executed in predictable ways; there is no need for a financial system that can possibly misuse one’s financial data/records and increase the chance of being hacked.
E. Know Your Customer (KYC)
Despite all of these benefits for individual users, the peer-to-peer purchase of decentralized financial assets (Bitcoin, Etherium, and other crypto coins) does not go through formal screening, widely known as Know Your Customer (KYC). [3] KYC is a “financial institution’s obligation to carry out certain identity and background checks on its clients before allowing them to use its product or platform.” [4] Instead, every holder of a cryptocurrency can be traced through the digital ledger known as the blockchain via the previously mentioned decentralized exchanges. (Blockchain can be defined as “a database that stores encrypted blocks of data then chains them together to form a chronological single-source-of truth for the data.”) The asset is decentralized, which makes it publicly available, accessible, and quasi-transparent.
If a person chooses to do a peer-to-peer transaction, they could potentially avoid detection due to the difficulty of pinpointing their identity. For example, person A can send person B 1 ETH and the proof of the transaction will be stored on the blockchain. The issue in identifying who person A is comes when there is no history of person A using a centralized exchange where they would have had to provide identification details using traditional “Know Your Customer” financial regulation requirements. Due to this, there is a common concern among governments and investors that people with ulterior motives would use cryptocurrency to avoid detection. Platforms that do not use KYC typically let their users roam free as long as they have access to their own personal wallet.
Binance, a popular centralized cryptocurrency exchange, began to require their customers to provide identification and pass facial verification in order to make deposits and transactions after the UK and Japan stopped the authorization of the exchange in their country. However, when another centralized exchange ShapeShift did the same, it lost “95% of its users as a result of the KYC measures it was forced to implement.”[4]
Some countries have taken regulation a step further by imposing bans on trading and mining cryptocurrency altogether. For example, “In June 2019, trading cryptocurrency was officially banned in China, when the People’s Bank of China stated they would be blocking access to all forms of cryptocurrency exchanges, domestic and foreign, and Initial Coin Offering websites through cryptocurrency transactions continued through foreign online exchanges.” As of 2021, mining of crypto is banned in China, despite it being one of the largest crypto hubs in the world.
As of now, the US has a pretty lenient process of cryptocurrency regulation. Despite having KYC measures in many exchange applications, there are numerous decentralized exchanges that have not seen a similar amount of regulation. The reason for that is the incompatibility with current privacy and securities laws, as well as states’ own views on crypto regulation.
In order to determine whether or not a cryptocurrency is a security, the Securities and Exchange Commission (“SEC”) tends to use a Howey test, a process developed by the United States Supreme Court in 1946 in SEC v. W.J Howey Co., 328 U.S 293 (1946). According to the test, “a transaction is an investment contract if: 1) it is an investment of money, 2) there is an expectation of profits from the investment, 3) the investment of money is in a common enterprise, and 4) any profit comes from the efforts of a promoter or third party.” Interestingly, the SEC has ruled that neither Bitcoin or Ethereum are securities under the Howey test, due to the fact that they are decentralized with no entity being in control. With that in mind, KYC cannot be done on any decentralized exchange because there is a risk of information leaks. As a decentralized entity, there is nobody stopping the KYC info being leaked into the web since there is essentially nobody to blame for non-compliance. SEC regulations and requirements as well as state-level laws regarding the use of cryptocurrency are not static, making the US a lenient but confusing regulatory system. Therefore, these rules regarding securities and cryptocurrency tend to change, and it is possible we will see more regulation on even decentralized exchanges.
F. NFTs
Another digital asset that has faced a lot of scrutiny regarding its regulation are NFTs (or non-fungible tokens). Non-fungible tokens are created on top of a blockchain and are not fungible, meaning they cannot be traded or exchanged with another without losing value. Each NFT is unique and can be absolutely anything including artwork, videos, collectibles, antiques, music, sounds, and even services. When a person purchases an NFT, they gain exclusive ownership to the “piece” and thus gain a digital token with data verification of the asset. According to Global Legal Insights, “the SEC has not yet initiated an enforcement action against the creator of an NFT or the operator of a platform that facilitates the offer and sale of NFTs.” [5] Thus, regulation of NFTs and cryptocurrency in the United States is pretty lenient, though that cannot be said about other countries.
G. Regulation
As recently as April 2022, the European Parliament agreed to start negotiations with EU countries on more explicit regulations that would allow the “tracing and identification of transfers of crypto-assets, to prevent their use in money laundering, terrorist financing, and other crimes.” In the eyes of crypto proponents, this runs counter to the logic of decentralized exchanges, which is to stay private and out of reach from third-party institutions. In a market dominated by an anti-establishment set of beliefs where trading cryptocurrency can bring one millions of dollars, it is understandable that users would want to keep their business secret. However, governments have a purpose to protect the general public and promote cryptocurrency regulation.
One of the main reasons the EU is pushing for harder restrictions is the environmental concerns that come from mining coins. For instance, “the Digiconomist’s Bitcoin Energy Consumption Index estimated that one Bitcoin transaction takes 1,544 kWh to complete, or the equivalent of approximately 53 days of power for the average US household.”[6] In an effort to abide by the Green Deal, many European countries are hoping that crackdowns on mining and other regulations would encourage a more eco-friendly lifestyle.
There are other pros to regulations as well. First, having official KYC and Anti Money Laundering security measures (AML) can encourage investors to continue bringing in money in support of different projects. If one knows that their funds can be protected by law, they are more likely to give up a good chunk of their finances. These regulations also bring more legitimacy and digestibility to the idea of cryptocurrency that has been widely regarded as a “scam”. By regulating the market, governments can monitor money laundering activities and screen foreseeable terrorist transactions.
However, there are cons as well. Decentralized markets give investors the ability to invest in projects without legal hurdles, and make a sizable amount of money before others. The whole idea of cryptocurrency is that it is inclusive, and people in the business would like to keep it that way. Decentralized assets are available to the general public as “accredited investors make up a small fraction of all investors in the US, and excluding non-accredited investors from investing in private issuances limits the issuer’s ability to raise capital while denying ‘unsophisticated’ investors the opportunity to invest in early-stage, high-potential projects.” In the eyes of the general public, that could mean that only rich people would get wealthy off of cryptocurrency due to the possible regulations on what an “accredited” investor might mean, thus increasing the wealth disparity between users.
As of now, the latest regulation of the US crypto ecosystem comes from the March 9th “Executive Order on Ensuring Responsible Development of Digital Assets”, where President Biden discusses 6 main objectives for cryptocurrency regulation. These include: consumer and investor protection, financial stability, mitigation of illicit finance and national security risks, U.S leadership in the global financial system and economic competitiveness, financial inclusion, and responsible innovation. The order attempts to employ the Financial Stability Oversight Council to identify and mitigate regulatory gaps, anticipate any type of national risks that come from using cryptocurrency on unregulated decentralized assets, and promote institutional research and a democratic access to this financial market. While the Executive Order does not necessarily have a direct impact on cryptocurrency regulation, it can serve as a step toward a more regulated crypto environment in the US. [7]
H. Concluding Thoughts
To this day, it is hard to tell whether there is a right or wrong answer to the question of whether or not cryptocurrency should face stricter regulations. There are compelling points made on both sides, and regardless of the outcome, there will be backlash. As of now, cryptocurrency is still a “wild west” world for the general public and it seems set to remain that way as EU regulations of decentralized blockchain loom. However, it is important to consider the evident loss of users. If the US were to impose significant regulations on the market, large investors could move their domicile toward countries with more open regulation. Making sure regulations are friendly toward investors is key to driving new flows of capital to nation-states of the future.
III. A Case Study on Artificial Intelligence
A. Background
In this subsection, we explore the creation of new technological regulations by doing a case study on Artificial Intelligence (AI) regulations in the United States. The majority of this subsection will focus on the definitional, ethical, and status quo issues that legislators often face while regulating novel technologies like AI.
B. Theoretical And Conceptual Considerations For AI
i) Artificial Intelligence & The “Intelligence Debate”
In general, artificial intelligence can be thought of as any non-living machine, computer, algorithm, and/or system that can think “intelligently.” We call these machines “intelligent” because computer and cognitive scientists have modeled these machines to be equally capable, or even more capable than humans of doing certain human tasks.
However, the nuances of this definition of artificial intelligence are heavily debated in general, academic, and legal spheres. A lot of this debate stems from disagreement over what “intelligence” means. Two of the most popular interpretations of “intelligent” in the artificial intelligence context are:
- Non-living entities are intelligent if they are on par with human intelligence.
- Non-living entities are intelligent if they exhibit sufficient problem-solving abilities.
Understanding the debate between (1) and (2) requires an understanding of how cognition and computation interact to create “intelligent machines.” In the following sections, we will characterize this interaction.
ii) The Interaction Between Cognition & Computation
- Cognition
The majority of AI is modeled after human models of cognition. Human models of cognition are theories that use linguistics, anthropology, philosophy, computation, neuroscience, and psychology to determine how humans receive, process, and output information in pursuit of a certain goal.
Though many models of human cognition have been proposed, scientists are still unsure what mechanism actually drives human cognition. This is in part due to the complexity and abstractness of investigating information processing in the human mind. The complexity lies in the fact that humans can only observe the output of cognitive processes – behavior. Though behavior is an incredibly useful data point, behavior is not necessarily indicative of the cognitive process. For example, humans often behave in a way that is contradictory to their attitudes, and we call this “irrational” behavior. Given our current limitations, much of our understanding of human cognition involves theories based on the observable output behavior humans produce, rather than the underlying mechanism that produces those behaviors.
2. Computation
Our current understanding of cognition poses an issue for AI scientists, who attempt to model computational models after models of human cognition. The hope is that modeling AIs off of human cognition would permit AIs to perform at and above a human level.
The issue is that current AIs are modeled based on an incomplete and potentially inaccurate understanding of cognition. Without an understanding of the underlying mechanism, AIs often struggle to perform certain tasks. Additionally, the lack of ability to compare computational models to human models of cognition makes it difficult to define whether an AI is “intelligent.”
iii) Machine Learning (ML) & Deep Learning (DL)
- Machine Learning
In spite of the theoretical difficulties surrounding the creation of an AI, AIs have become increasingly capable of doing tasks as well as, or even better than us. Unlike humans, AIs have the capacity to retain and process information in higher quantities and rates than humans can, giving it enormous potential. One of the largest fields of artificial intelligence is machine learning (ML). Machine Learning is exactly what it sounds like: a machine that can learn through experience and information without being explicitly programmed to do something.
- Deep Learning
One of the most popular forms of Machine Learning is Deep Learning (DL). As its name implies, DL permits machines to learn from inputs/data using probabilistic reasoning. Computer scientists feed the machine data about how certain x inputs lead to y outputs. The machine aggregates this data and tries to characterize the trends it sees between x and y. The machine creates probabilistic connections determining how likely it is that a certain x will lead to a certain y, and makes “decisions” based on how probable an outcome is. These connections are referred to as a “neural network.”
- Systemic Issues with Artificial Intelligence
“Black Box Problem”
Much like the issue faced by cognitive scientists trying to understand how human models of cognition process information, an issue with DL is that we often cannot see why or how a machine processed information and made a certain decision. This ambiguity is referred to as the “black box problem.” The black box problem occurs because the machine aggregates and characterizes the data, not us – scientists provide the machine with inputs and associated outputs, but how the machine aggregates the data and makes connections is out of our control. The black box problem is troublesome because a lack of algorithmic transparency can make it hard to detect algorithmic bias.
“Algorithmic Bias (Data)”
The data we feed the algorithm is as important as the algorithm itself. Algorithmic Bias describes prejudice exhibited in machine decisions, such as racial bias or gender bias. While this type of bias can also apply to how the algorithm was built, it is often the case that the data we feed it is biased, which makes the machine biased.
This is because computers do not make correlations in the same way that humans do; the correlations computers make are often arbitrary and non-causal. For example, an AI could draw the conclusion that cold weather is linked to higher spending – in reality, the holiday season coinciding with cold weather is the cause.
When we feed a machine biased data, we increase the chance that our algorithm will reach erroneous, and often harmful conclusions. For example, AIs trained for facial recognition are much better at recognizing white faces than any other race because the data the AIs have been trained on consists of mostly white faces.
What this Means for the “Intelligence Debate” and the Law
Our current understanding of human and machine computation makes it difficult for us to answer questions associated with (1). Arguments in favor of (1) typically point out that AIs have the ability to “independently” learn and develop, and eventually become “independent” of the computer scientists that created them. This means that computer scientists could not be held liable for errors the AI made. As such, AIs should be treated with some modicum of legal autonomy. This would mean that AIs who meet the “threshold” of human intelligence would be provided with rights, and held liable for errors or wrongdoings that they produce.
Naturally, many people disagree with (1). Those opposing (1) typically argue that AIs will never be able to reach human levels of intelligence. They also argue that those who create AIs should be held liable for errors the AI makes. They argue this by stating that scientists are responsible for overseeing and creating algorithms that can safely and accurately complete the tasks it has set out to do – if the algorithm deviates, then it must have been a flaw of the original design.
In either case, the issues surrounding (1) make it clear that we do not have enough information to reasonably answer whether or not non-living entities such as AI possess human intelligence. Proposition (2) is considered much more reasonable because it is measurable, meaning it can be codified and considered within the scope of the law.
Therefore, most AI regulations that are in progress, have been passed, or have been suggested seem to require a focus on either researching and understanding artificial intelligence, or regulating AIs based on their ability to complete tasks. Subsequently, our analysis will focus on the legal implications and regulations surrounding (2).
C. AI in the Status Quo
- Contemporary Uses of AI
Advances in AI technology have resulted in a corresponding proliferation of AIs in daily use. The 2021 McKinsey Global Survey on AI usage reported that 56% of all respondents report AI adoption in at least one function [8].. This 50% increase from 2020 is reflective of AI’s increasing capacity to support and affect different fields. AI is commonly used in healthcare, business, law, and virtual assistants.
i) Healthcare
AIs for healthcare serve broad-ranging purposes but are primarily used in diagnostics, patient engagement and care.
Diagnostic AIs analyze patient images to diagnose an illness, receiving an image as their input and returning a diagnosis. Another proportion of diagnostic AIs analyze patient data such as blood pressure, blood work, family history, and/or genomic data, and predict how probable it is for certain patients to develop a certain illness.
AIs for Patient Engagement and Care are relatively new and aim to solve the “last mile” problem. The “last mile” problem describes how a patient’s health outcomes are contingent on their participation in their own healthcare regimen. Though providers and hospitals often create individual plans for these patients, it is thought that AIs could be built to better accomplish this task. AIs for patient engagement and care could improve health outcomes by drawing from a larger pool of information to create plans for patients [9].
ii) Business
In business, AIs are commonly used to predict consumer behavior, market and advertise, provide customer service, and detect fraud. AIs used to predict consumer behavior collect information on consumers through purchase history, browsing history, and social media. The AI takes in this information to figure out consumer trends and tailor marketing/advertising to specific consumers.
Marketing/Advertising AIs take in information about what marketing techniques and mediums were most effective (amount of clicks, views, etc.) at attracting a consumer to purchase something. This helps them tailor advertisements to specific consumers by displaying certain stores, products, or services.
Chatbots powered by natural language processors (a language AI) and audio recognition take phone calls and chat in online chat boxes to address consumer needs. The cost of maintaining a chatbot or callbot is cheaper than human labor, reducing costs for companies.
There are also AIs that help detect financial fraud. These AIs can be used in both internal and external company affairs. Using data on what types of financial trends indicate fraud, AIs can flag internal trends that indicate fraud for human review. AIs can also be used externally for customers, helping detect fraudulent credit card usage or phishing [10].
iii) Law
The most commonly used AIs in the legal field are Technology Assisted Review (TAR) and Litigation Analytics. TAR helps lawyers quickly find relevant documents in a database by analyzing what documents might be most relevant to the legal case at hand. TAR helps cut down the time lawyers spend on manual review approximately by a factor of 50, allowing lawyers to focus on cases rather than legal research.
Litigation Analytics is a slightly more controversial legal AI. Litigation Analytics pulls data and verdicts from prior court cases to predict how probable it is that a case will receive a certain verdict. This allows legal teams to strategize and plan their case. The issue with legal analytics is that court verdicts often reflect socioeconomic biases, which introduces biases into the algorithms. [11]
iv) Virtual Assistants
The best-known use of AI is in virtual assistants such as Siri, Amazon Echo, and Google Home. Virtual assistants take in a large range of data to help users perform daily functions like sending emails, purchasing groceries, or getting directions. Virtual Assistants primarily use audial natural language processors to understand what users are saying and respond accordingly.
2. Status Quo Issues with AI
Though the majority of Americans view AI technology optimistically, many are concerned about AI governance. AI governance describes “the process of defining policies and establishing accountability to guide the creation and deployment of AI systems.” [12] Because law operates as the codification of ethics, it is important for us to understand contemporary ethical issues surrounding the deployment of AI. In the following section, we explain several status quo issues with AI.
Safety
- Data Privacy
Data privacy describes the protection of personal or private information. Many issues surrounding AI and data privacy are contingent on consent and protection. In order for the data that AI receives to be private, individuals must consent for their data to be used in an AI. Individuals must also be made aware of how their data is being used and how it’s being kept private. A lot of data privacy issues surround the fact that AIs are not necessarily transparent. In order to understand how people’s data is being used, one has to understand how the algorithm is using the data, and this is something we are not always capable of explaining.
2. Cyberattacks
The issue with cyberattacks is less about how AIs are structured, and more about how people use AI. In recent years AI has been involved in email and phishing attacks, ransomware attacks, cloud infrastructure attacks, and data leakage. [13] At the rate that AI is developing, individuals and governments are struggling to develop infrastructure and technology that helps combat offensive AI attacks. Algorithms can easily be manipulated to recognize patterns that help AI attackers avoid interference, and identify vulnerabilities in systems.
3.Physical Safety
Another less obvious issue posed by AI is threats to physical safety. Part of the safety issue surrounds AI being used to control weaponry. AI systems that are poorly designed or poorly regulated could be exploited and used to deploy dangerous weapons. In addition, the misuse or mis design of AI technology can pose threats to individual safety. For example, racial biases involved in facial recognition technology can and have led to false arrests.
Legal Liability
Legal liability is meant to address who should be held accountable for AI systems. A lot of this issue stems from whether or not the programmer or the AI itself should be held accountable. Because the programmer develops the system that the AI uses, it seems reasonable to hold the programmer accountable for any missteps that the algorithm makes. On the other hand, because AI often utilizes deep learning, the algorithm makes a lot of its own decisions. This autonomy displayed by algorithms makes it difficult to hold programmers fully responsible for decisions that the AI makes.
One of the only legal cases that addresses legal liability is a civil suit involving Tesla’s autopilot feature. Tesla’s autopilot feature enables drivers to sit in the vehicle without having to control the car. In 2019, a crash involving a Tesla car killed two occupants in another car. The person in the Tesla was charged with manslaughter. This suit marks the first time a driver in the US has been prosecuted for a felony while using semi-automatic driving. It is expected that the suit may set precedents for who is held responsible when self-driving is involved in car accidents. In this situation, the driver was held liable. [14] Part of this is because Tesla has user agreements dictating how autopilot must be used. Tesla states “autopilot and Full Self-Driving Capability are intended for use with a fully attentive driver, who has their hands on the wheel and is prepared to take over at any moment. While these features are designed to become more capable over time, the currently enabled features do not make the vehicle autonomous.” [15]
Another major concern for many is that the integration of AI will displace humans from the workforce. Since the 1980s, automation has added to displacement in the labor market and inequality. Those with jobs that can be more efficiently and inexpensively implemented with AI face a very real risk that their jobs will be replaced. Subsequently, the integration of AI will force policymakers to consider how to balance innovation and efficiency with the displacement of labor. [16]
D. The Creation and Implementation of AI Regulations
How ought these ethical considerations be addressed with regulation? There is widespread agreement among Regulatory Agencies and businesses that AI regulation is necessary. Effective AI regulations will address the systemic and status-quo issues that were covered in the earlier sections of this article.
In recent years, the United States has acknowledged that there is a need for AI regulations. The National AI Initiative was aimed at fulfilling this purpose. The initiative was passed in 2021 and provided several guiding principles for how AI regulations ought to be taught. In the following section, we overview these guiding principles.
- “Public trust in AI. The government must promote reliable, robust, and trustworthy AI applications.
- Public participation. The public should have a chance to provide feedback at all stages of the rule-making process.
- Scientific integrity and information quality. Policy decisions should be based on science.
- Risk assessment and management. Agencies should decide which risks are and aren’t acceptable.
- Benefits and costs. Agencies should weigh the societal impacts of all proposed regulations.
- Flexibility. Any approach should be able to adapt to rapid changes and updates to AI applications.
- Fairness and nondiscrimination. Agencies should make sure AI systems don’t discriminate illegally.
- Disclosure and transparency. The public will trust AI only if it knows when and how it is being used.
- Safety and security. Agencies should keep all data used by AI systems safe and secure.
- Interagency coordination. Agencies should talk to one another to be consistent and predictable in AI-related policies.”
The White House states that these principles have three main goals. These principles are meant to “ensure public engagement, limit regulatory overreach, and promote trustworthy AI that is fair, transparent, and safe.” [17] The Department of Commerce, Federal Trade Commission, the Food and Drug Administration, the National Security Commission, and the Government Accountability Office have echoed this stance on AI. [18]
Though the support for AI regulations is widespread, there has been minimal progress on implementing these principles into legislation. Part of this is due to the ambiguity of AI’s systemic issues and status quo issues. One of the core issues with creating regulations is that there is no formal legal definition of “Artificial Intelligence.” Agreement upon a definition will be necessary to create a standardized set of regulations.
However, laws that have been passed on AI aim at increasing either AI education or research, which are both promising steps in the direction of implementing substantive regulations. Certain states have also passed bills that regulate or ban certain AI technologies, which could set a precedent for legislation at the federal level.
As of 2022, the European Union is the only nation that has formally passed substantive artificial intelligence regulations. The European Union’s Artificial Intelligence Act consists of a product safety framework, with a focus on risk management. Under the risk-based framework, “high-risk” or “dangerous” algorithms are prohibited.
The effectiveness of this regulation has been disputed. The AIA has been criticized for not being prohibitive enough, not being enforced enough, and being too ambiguous about what constitutes “high-risk” or “dangerous” algorithms. Additionally, some AI use cases are so novel that risk cannot be distinctly determined, meaning the AIA cannot be enforced. For these reasons, many feel that the proposal falls short of protecting fundamental rights from potential AI harm. [19]
In general, the creation of AI regulations is characterized by an attempt to maintain a balance between creating effective and ethical regulations, while addressing the pressing rate of potential AI threats in a timely manner. A lack of information on the technology and its potential harms is the major barrier preventing nations from creating and implementing AI regulations.
E. What AI Teaches Us About the Formation of New Regulations
So what can AI teach us about the formation of new technological regulations? A commonality among new technological regulations is a pressing need to address issues caused by a technology, and a lack of information on said technology.
The regulation of AI demonstrates the importance of balancing current harms and current knowledge with the potential future harms and benefits of AI. Government funding for technological research and education are critical steps to addressing any new technology through the lens of the law, such that implementations are created effectively, ethically, and safely.
IV. Data Privacy
A. Background
In the current United States legal code, there is no existing national law regulating consumer privacy or data collection. However, as the economy becomes increasingly globalized, it becomes more and more impossible for people to get services or participate in much of online society without providing their personal information in some form or another to online service providers. According to the Legal Information Institute, the collection of personal information happens online from the traversal of commercial websites, federal governments, and essentially every part of the consumer marketplace, like financial institutions, online shopping sites, and simple web browsing. [19] Companies with websites have the ability to use internet “cookies” or forms related to personal information to collect personal data like names, email addresses, IP addresses, physical locations, and prior browsing history. Much of this data is then sold back to third party marketing companies, which then provide users with targeted online advertising based on their online profiles and other consumer data accumulated by the companies. In September 2021, that targeted online advertising industry, which began growing at this rate a little over 20 years ago, was valued at around $350 billion. Digital ads make up a significant amount of revenue for companies with both large and small online presences, whether they be companies that provide free services like Google, or online businesses that rely on digital ads to increase social awareness about their brand.
B. 3rd Party Data Collection
Limits on 3rd party data collection is a large part of the conversation around what fair and ethical consumer data usage should look like. Some of the limits surrounding this were better declared in TransUnion v. Ramirez, a class action lawsuit decided in 2021. Mainly, it changed the forms of 3rd party collection and usage that the courts and laws viewed as incorrect or illegal, and forced consumers to show a tangible loss that was a result of the sharing of their data to 3rd parties. Currently, for a class action lawsuit to have standing in court under Article III of the Constitution, they must be able to show some kind of “concrete harm.” [20] In TransUnion v. Ramirez, the plaintiffs believed that TransUnion was at fault under the Fair Credit Reporting Act because the company had collected their credit score and related information, and in some consumers’ reports, had incorrectly designated them as potential terrorists. That designation was not a fact that consumers were aware of until some of the plaintiffs, Ramirez included, found that those reports had been sent to other companies, like car dealerships, and were unable to buy anything due to the incorrect statements on their credit reports. Even though the class action had over 8000 members, the court found that only just over 1600 of them had standing under Article III of the Constitution because that portion of the plaintiffs were the ones that suffered harm when their personal information was disseminated to 3rd parties. Essentially, the court believed that just having user data collected, with or without user knowledge, would not be enough to find a company at fault. The company could only be at fault when that data was then disseminated to 3rd parties, and users were able to prove a harm that came from that.
C. The Current State of Data Privacy Regulations
i) International and American Regulations
But, consumer preferences are changing at even stricter rates than what existing legislation accounts for. As governments become more and more involved in regulating Big Tech companies and their social power, consumers and businesses are trying to come to some kind of balance between the collection of data and personal online privacy. In the United States especially, that balance has been largely self-regulated because the state of data privacy regulation varies, not only across countries, but also across states and industries. The European Union has proven to be the most advanced in terms of progressive data privacy policies, both because of their adoption and consistent modifications of the strictest and most comprehensive data privacy law to date: the EU’s General Data Protection Regulation, and their constant enforcement of the clauses in the bill. Meanwhile, in the United States, different states provide different levels and different types of consumer data protections, meaning companies have to manage different data collection systems and privacy settings based on which state a user is operating from. According to Thorin Klossowski for the New York Times, “the United States doesn’t have a singular law that covers the privacy of all types of data. Instead, it has a mix of laws that go by acronyms like HIPAA, FCRA, FERPA, GLBA, ECPA, COPPA, and VPPA, designed to target only specific types of data in special (often outdated) circumstances.” [21] More so than regulating the current nature of data collection, which occurs primarily through online sources, the existing United States federal laws regulate the collection of and access to various types of personal records, whether they be financial records, credit reports, or physical health information. Some relevant legislation meant to protect the online presence and profile of users includes the COPPA and the FTC Act. COPPA, which was enacted in 1998, limited the amount and types of data that websites could collect on users under the age of 13, whereas the FTC Act allows the FTC to investigate companies that go against their own privacy policies. While the FTC and, more recently, some federal agencies have begun using their regulatory power to lodge complaints and fines against companies that misuse data or participate in anticompetitive practices, the choices that businesses make regarding their data collection practices are largely self regulated in the United States, with the exception of the few states with comprehensive data privacy laws.
ii) State-Level Data Privacy Regulations
According to Thorin Klosowski for the New York Times in September 2021, only California, Colorado, and Virginia have signed and put into effect comprehensive consumer data privacy laws. The laws apply to all consumers of a certain site or business that live in those states regardless of where the company is headquartered. While each law is different in execution, generally, “a company operating under these regulations must tell [consumers] if it’s selling [personal] data; [users] also get a choice in whether they’re okay with that or not, and they have the right to access, delete, correct, or move their [personal] data.” [22] Outside of these general restrictions, these states have different expectations and plans for specific implementations, especially when it comes to the size of the businesses that the law can be applied to, the time that companies have to correct any practices that do not comply with the restrictions of the law (also called cure periods), and the way consumers can choose to request to opt out of data collection methods used by corporations. With these differences, legal experts believe that the California consumer data privacy law, or the CCPA, provides the strongest and most complete consumer protection regulations in the entire United States. In contrast, Kate Ruane, legislative counsel at the ACLU, believes that Virginia’s law, or the VCDPA, is the weakest of the existing laws. [23] Aspects of the CCPA that experts found to be progressive in terms of consumer privacy include the law’s allowance for private action, which are still somewhat limited, and the requirement that businesses provide global opt-out consent. Meanwhile, Ruane found that the VCPDA did not address a need for the private right to action, and that “A lot of the provisions [in the VCPDA] are business-model affirming. It essentially allows big data-gathering companies to continue doing what they have been doing.” [24] Private action is the right that allows consumers to sue a private entity for certain kinds of data breaches, and opt-out/opt-in consent allows users to decide whether their data can be collected at all. Opt-out assumes that all users will say yes, and gives them the opportunity to pick, whereas opt-in does the opposite. Of the two, opt-in consent is seen as more progressive, and provides consumers with more power to choose, although it is done at the cost of the data that corporations can collect, and the related revenue they can make from their services that involve data collection. Many believe that could force companies to have to change their business models and turn to other forms of payment models or services to make up the revenue loss.
D. Guidelines for Effective Data Privacy Legislation
Consumers and regulators, regardless of the country or state they are creating litigation for, primarily center their data privacy protections around a few general topics. Many of them are present in the principles and clauses of laws in the United States, whether or not they are more comprehensive like the CCPA, and in global laws centered around consumer data usage, like the European Union’s GDPR. Some of these topics, or general principles, include managing 3rd party data sharing rights, required opt-in consent to allow companies to collect personal data in the first place, the minimization of data collected, and non-discrimination toward users based on the data that businesses are allowed to collect. Klosowski found that experts believe each of the four guidelines above is important to creating effective data privacy legislation. While many of the national laws that are present in the United States currently do a fairly good job of protecting consumer privacy in the specific industry or state that law is meant to target, the data of users in a large portion of the country still remains unregulated. But, due to the nature of the patchwork laws that exist to protect consumers, users in those regions are still subject to the same consequences and risks of a lack of data privacy as users in states with regulation in place. As a result, creating and maintaining practices that protect the data of all consumers on a national scale currently happens regularly in one of two ways in the United States: litigation, rather than legislation, and self-regulation practiced by corporations.
E. Barriers to Creating and Implementing Tech Regulations
The fact that there are only a few bills currently in place to protect the majority of consumers and their personal data in online settings is not for a lack of trying by both federal and state governments. However, the legislative process is slow, especially in the United States, making it so that the speed at which laws are made cannot match the rate at which the market power of Big Tech companies increases, especially due to the sheer number of consumers that use those companies’ services regularly. Jason Furman, professor at the Harvard Kennedy School, found that the legislative process is slower in the United States, especially in comparison to the European Union which has already been through the process of proposing and signing into law the GDPR, which protects digital consumers from many forms of data misuse. He also recommends that the United States can learn important lessons from the EU’s practice in regulating Big Tech. [25] So, while the United States might not yet have sweeping legislation regarding consumer data privacy, regulatory agencies and states are rushing to fill the gaps at faster rates than ever before. In 2021, 38 states had proposed over 100 bills relating to data privacy or speech regulation. Of those, 27 were for online privacy alone, a number that grew from just 2 online privacy bills proposed in 2018. According to David McCabe and Cecelia Kang for the New York Times, “the push signals that states are no longer content to sit on the sidelines of setting the rules for the internet — especially as Washington has moved slowly. While Congress has ramped up hearings and reports to curtail the power of Google, Amazon, Apple and Facebook in recent years, lawmakers have passed only one bill.” [26] While the legislation is being slowly debated, individual consumers and entire federal regulatory agencies have been putting up legislation battles against and investigations into large technology corporations and their unethical, and sometimes anticompetitive practices.
F. FTC and Big Tech
The majority of the burden of these investigations into Big Tech companies falls under the purview of existing regulatory agencies. Some experts have recommended creating a regulatory body to handle issues related to Big Tech companies and online services within the economy, because there are currently none. However, the FTC has unofficially been deemed responsible for managing and curtailing the market power of Big Tech companies and imposing fines and restrictions on them when those companies do not follow existing restrictions or the principles of their own privacy policies. In line with the ideas that societies and governments are less willing to share their personal information with online companies, and that a social push against the sheer power that big tech companies have is occurring, some recent fines and lawsuits imposed against Big Tech companies have been among the largest of their kind.
One such example is the 2019 fine imposed by the FTC on Facebook for failing to meet its own privacy standards. Facebook, the firm whose role in the Cambridge Analytica scandal brought a lot of scrutiny to the door of the company, and many others, and their uses of user data, was fined $5 billion for failing to meet certain standards and violating existing regulations like the FTC Act. That $5 billion fine was the result of a settlement meant to punish the company for deceiving users about the company’s use of their data, unethically allowing 3rd parties to access user data without user consent even after promising not to, and misrepresenting the steps required to increase user privacy for its facial recognition technology. The magnitude of the fine has importance because it amounts to a quarter of the firm’s 2018 profits and, according to current FTC commissioner Noah Joshua Phillips, “is the largest privacy fine in history. It’s orders of magnitude greater than previous record privacy fines in the U.S. and Europe.” [24] Further than just imposing fines, the FTC’s lawsuit and resulting settlement with Facebook allowed the FTC to force the firm to restructure its privacy practices by creating an independent commission to monitor its activities, and creating new consumer privacy protection obligations that Facebook would have to follow. In recent years, the FTC has continued to use litigation to entice companies to change their privacy policies and practices, including a $170 million fine settled by Google for violating the Children’s Online Protection Privacy Act (COPPA), and settlements against various companies that provide online services for violating existing rules regarding consumer data privacy.
However, the FTC cannot create laws or force companies to update privacy practices in ways that are more progressive than what a nationwide law, or one in a current state, requires. Phillips believes that the settlements and fines are still important because they are allowing the market to finally see that the FTC takes data privacy protections seriously, and that companies must take their privacy promises just as seriously. [24] All the FTC can do, and has been doing, is punish companies for failing to comply with existing data protection laws like the COPPA and the FTC Act in hopes that every company will use the high magnitude of the fines to change their own practices to be completely within the legal and ethical bounds of data collection through self-regulation. And to a certain extent, these strict enforcement tactics and the public environment around data misuse and collection has been causing that.
The changes in business practices that corporations are undertaking, especially when it comes to privacy policies and their types of data collection come not only from pressure from regulatory bodies, but also from social pressure from consumers. Many consumers are more and more unwilling to work with companies that have security practices they are uncomfortable with, as shown by a 2020 survey run by McKinsey Consulting stating that “One in ten internet users around the world (and three in ten US users) deploy ad-blocking software that can prevent companies from tracking online activity. The great majority of respondents—87 percent—said they would not do business with a company if they had concerns about its security practices.” [25] That social pressure is also a large reason why corporations are changing the way they collect data and sell it to 3rd party companies with or without user consent. This form of self-regulation from corporations themselves can take many forms. For example, Apple introduced a pop-up requiring users to approve third-party tracking from apps on the phone before the app can be used. Similarly, Facebook and Google were both said to be looking for a method to display ads without tracking user data, as is the current method for digital advertising. Whatever reason companies are using for modifying their data collection practices, the effects are sweeping and expensive. For example, in order to comply with California’s CCPA, companies like Microsoft are changing their practices for users all across the country, rather than just for users in California as the law requires them to do. While setting up methods to meet the requirements of the GDPR, an estimate by the International Association of Privacy Professionals found that Fortune 500 companies had spent $7.8 billion by 2018 to prepare. [25]
G. Concluding Thoughts
All in all, settlements have been effective in encouraging change in corporate practices, but the reality is that using the authority of federal agencies to force companies to change their behavior cannot bring about change in every form of data privacy issues that exist in online society today, whether it be algorithmic bias, unfair levels of unknown surveillance, or the way online profiles are built and collected through methods that consumers are not privy to. To fix those issues at their root, legislation is required. While there is some overlap between the various issues catalyzed by Big Tech companies and their use of consumer data, experts and researchers have found that addressing different problems, whether it be antitrust, AI/algorithmic bias, or data privacy, require different solutions, even though they do usually come from the same agencies. This was best articulated by Erika M. Douglas, Assistant Professor of Law at Temple University, who found that “over the last twenty-five years, data privacy has also become a separate area of legal doctrine. In that capacity, data privacy law may clash at the margins with antitrust—much like intellectual property or consumer protection law did before it.” [27] Nevertheless, every form of consumer protection tactic has large impacts on users, and their online choices. So, any actions that should be taken in regulating the market power of Big Tech companies, in whatever form or industry, should be done for the sake of consumers, and their online and offline needs, which is a trend that various countries and states have undertaken at varying rates in recent years, as the economy becomes more globalized and digital.
PART II
Introduction:
There is a rich tradition of antitrust law in the United States dating back to the passage of the Sherman Antitrust Act in 1890. Antitrust is the field of law protecting the economy from harmful business practices. While the courts have wrestled internally over the last century to settle the goals of antitrust and the proper rule of law, the nature of industry has transformed with the arrival of the digital world. “Big Tech” companies, such as Apple, Amazon, and Google, have amassed massive shares of their industries, and with it, hefty influence. The unique structure of digital platforms has posed challenges to applying the antitrust law that was cultivated around simpler industries such as railroads and tobacco. It’s unclear whether Apple’s App Store policies and Amazon or Google’s preferences for their own products on their platforms are definitively anticompetitive. In this article, we explore the legislation and case law of antitrust, three modern issues of digital antitrust, and two major theoretical approaches to antitrust: the Consumer Welfare Standard and the Competitive Process Standard. Finally, we apply these approaches to contemporary issues in the hope that we can better illuminate the path forward for the future of digital antitrust.
Relevant Acts:
Before one can fully understand the modern disputes surrounding digital antitrust, it is important to review the foundations of antitrust law in the United States. The Sherman Antitrust Act of 1890 and the Clayton Antitrust Act of 1914 highlight a clear evolutionary history of the accepted goals of antitrust.
A. Sherman
The Sherman Antitrust Act was signed into law on July 1, 1890, by President Benjamin Harrison. [1][2] The Act was a landmark piece of legislation as it was the first attempt by Congress to outlaw trusts. Section 1 of the Act criminalized “every contract, combination in the form of trust or otherwise, or conspiracy, in restraint of trade or commerce among the several States, or with foreign nations.” [3] In a similar vein of thought, Section 2 outlawed monopolization. [4] Section 4 empowered the courts “with jurisdiction to prevent and restrain violations of this act and it shall be the duty of the several district attorneys…to institute proceedings in equity to prevent and restrain such violations.” [5] The penalties were made clear in Section 6, which states that “any property owned under any contract… mentioned in section one of this act… shall be forfeited to the United States.” [6] Additionally, the victims of trusts were made explicit in Section 7: “any person who shall be injured in his business or property by any other person or corporation by reason of anything forbidden or declared to be unlawful in this act, may sue therefore in any circuit court… and shall recover three fold the damages by him sustained.” [7]
As Judge Bork noted in his analysis, Congress was well aware of its limited powers under the Commerce Clause and was tepid to test the waters of its newfound antitrust powers, emphasizing the regulation’s application to only interstate and international trade. [8] Also, by relying on common law terminology, Congress crafted legislation coextensive with the judiciary’s understanding of commerce. The law is opaque by design, relying on the circuit courts and Attorneys General to give teeth to the specifics of enforcement.
The intent of the Sherman Act, according to Bork, was to attack cartels, horizontal mergers of monopolistic proportion, and predatory business practices. [9] Abstracting from the specific outlawed business practices, the text repeatedly scorns practices that cause the “restraint of trade or commerce” in Sections One, Two, and Three. The chambers of Congress were not divided on this issue; the Senate voted in favor of the Sherman Act 51-1, followed by a strong House vote of 242-0. [10]
The intent of Congress in founding the origins of antitrust through Sherman was shrouded in ambiguity. Preventing the restraint of trade sounds a lot like protecting competition, but as Justice Holmes would later note, the Sherman Act does not use that word. [11] Deferring precise definitions to the Court’s interpretation caused debate about whether the goal of antitrust is to protect competition, ensure benefits of trade like efficiency, or merely punish business practices deemed undesirable to conflagrate across the judiciary.
B. Post-Sherman Caselaw
The post-Sherman era of antitrust was a busy time for the courts. Though it is already best chronicled by Bork, a handful of cases are worthy of mention for their role in declaring the goals and rule of reason for antitrust. [12] The Supreme Court debated the proper rule of law for applying the Sherman Act in United States v. Trans-Missouri Freight Association. [13] The issue at hand was whether the price-fixing of railroad rates by an association of eighteen railroad carriers violated the Sherman Act. Justice Peckham penned the majority decision arguing it did. He set his sights on creating a rule of law for section one of the Sherman Act. Peckham’s opinion rejects the notion that reasonable restrictions on trade and commerce are allowed under Sherman, advancing a less subjective plain and ordinary interpretation of the statute under which all trusts conspiring to restrict trade or commerce are blanketly prohibited. For it is “the public interest… to allow free and open competition among railroads upon the subject of the rates for the transportation of persons and property.” Price fixing, therefore, became a category of per se illegality, and antitrust gained a rule of reason prohibiting restrictions on trade or commerce for the sake of the public interest in low prices.
Two years later, the Supreme Court would affirm Judge Taft’s Sixth Circuit opinion for Addyston Pipe and Steel Co. v. U.S.. [14] [15] In this case, a collective of pipemakers agreed to overbid on contracts offered by municipalities so that their designated lowest bidder member would win the contract at a higher than competitive price. The question before the court was whether an association like the pipemakers’ violated the Sherman Act. Judge Taft decided it was an illegal arrangement and proceeded to distinguish between ancillary and naked restraints. Ancillary restraints, when business partners collaborate or execute a merger, may result in less competition but are “necessary to promote the free purchase and sale of property.” These instances are not the problem, nor are they outlawed under the Sherman Act. Under naked restraints, such as the pipemakers’, “the sole object is to restrain trade in order to avoid the competition.” These are hostile to the public interest and indeed violative of the Sherman Act.
This amiability toward mergers proved to be short-lived when the Supreme Court decided Northern Securities Co. v. United States. [16] The Northern Securities Company was a newly formed railway that had merged the assets of three railway companies. In a divided 5-4 opinion, Justice Harlan ruled that the merger was violative of the Sherman Act due to the effect its reduction of competition had on restraining interstate commerce. Justice Holmes dissented, making the astute observation that the Sherman Act “says nothing about competition.” [17] The goal of protecting competition at all costs is more of a judicial invention than strict interpretation. In banning mergers by making them per se illegal like cartel arrangements, the court has chosen to improperly “require that all existing competitions shall be maintained.” Just fourteen years after the Sherman Act, tension between different philosophies regarding the goals of antitrust had emerged. Shall competition or consumer prices be protected vigorously under the law?
Justice Harlan and the majority’s concerned attitude toward the anticompetitive nature of mergers would dilute with the advent of Standard Oil Co. of New Jersey v. United States. [18] The Standard Oil Company was attempting to monopolize the oil industry through aggressive techniques that put pressure on smaller companies to accept merger offers. Justice White, who joined the dissent in Northern Securities, wrote for the majority. [19] He argued Standard Oil’s attempted monopolization violated the Sherman Act by restraining the trade of crude oil. But White went further, successfully arguing that under English common law only unreasonable restrictions on trade are illegal, a similar argument he made in his dissent in United States v. Trans-Missouri Freight Association. Standard Oil’s tactics placed “unreasonable restraints” on the rights of individuals to voluntarily exercise their trade or business and therefore were illegal under the Sherman Act. The Sherman Act’s meaning was morphed to only prohibit unreasonable restraints on trade. Justice Harlan protested this legislative intent coup, maintaining his standard of outlawing all restraints of trade in his dissent. [20] But Harlan’s ringing of the alarm bell fell on deaf ears, as he wrote on how the court “has not only upset the long-settled interpretation of the act, but has usurped the constitutional functions of the legislative branch.” This doctrine was supported by American Tobacco Co. v. United States just a year later, which once again found attempted monopolization of an industry antithetical to the language of the Sherman Act and the interests of “consumers, [to whom] the benefits would otherwise flow from free, vigorous and normal competition.” [21]
C. Clayton
The Clayton Antitrust Act was signed into law by President Woodrow Wilson on October 15, 1914. [22] [23] The Act was designed to amend the field of antitrust that was established under the Sherman Antitrust Act and refined over the following two decades of case law.
The text of the Clayton Act affirms the Sherman Act’s prohibition of trusts or monopolization. Where the Clayton Act goes further is in banning “price discrimination with intent to injure a competitor, exclusive dealing, and corporate stock acquisitions” that “may be to substantially lessen competition or tend to create a monopoly in any line of commerce.” [24] In passing the Clayton Act, Congress moved to deter trusts further so that the damages to competition would never occur as in Standard Oil. [25]
D. Post-Clayton, Brandeis
Following the Clayton Act, a new vision for the goals of antitrust was charted by Judge Brandeis writing for the majority in Chicago Board of Trade v. United States. [26] The federal government brought a suit against the Chicago Board of Trade for their call rule, which fixed the price of arriving grain daily until the next session, challenging the rule under the Sherman Act. The Court upheld the rule, explaining “the true test of legality is whether the restraint imposed is such as merely regulates, and perhaps thereby promotes competition, or whether it is such as may suppress or even destroy competition.” The protection of conditions of competition is made the explicit rule of reason for applying the Sherman Act, and whether a violation occurred depends on the context and circumstances of the nature, scope, effect, and history of the action. Concluding that the nature of the rule concerned the period of price making, the scope was restricted to arriving grain in Chicago for a portion of the day, and its effects had helped improve market conditions, the Court upheld the rule. This rule of law would lay roots for a Brandeisian school interested in preserving competition for its own sake.
Current Antitrust Issues:
A. Apple v Epic
Three of the largest contemporary “Big Tech” antitrust disputes surround three “FAANG” corporations: Apple, Amazon, and Google. Each of these disputes embodies the novel challenges “Big Tech” and the internet have brought upon the world of antitrust.
In August of 2020, Epic Games released an update to the mobile version of their immensely popular game Fortnite. Within this update, Epic Games allowed users to purchase in-game items directly from Epic Games’s website without needing to go through Apple or Google’s in-app purchasing platforms. [27] As this was a violation of Apple’s IOS App Store regulations at the time, Fortnite was subsequently removed from the App Store. In response, Epic Games brought a lawsuit against Apple alleging that the latter was violating antitrust law through their regulations on in-app purchases on IOS devices. [28] Specifically, Epic Games took issue with the 30% fee rate Apple charges for all in-app purchases taking place on their platform while restricting app providers from steering their users to other in-app purchasing platforms. [29]
In her ruling over the case, Judge Yvonne Gonzalez Rogers found that Apple had not violated any laws in forcing IOS developers to only distribute apps through the App store and only use their in-app payment processor for in-app purchases. Rogers’ decisions hinge on her application of the rule of reason test in regards to these restraints of trade. This test requires a court to analyze a restraint’s effect on competition through an assessment of “market power and market structure”. For both counts, the Court admitted that Apple’s restraints have some anticompetitive effects on the market. However, after taking Apple’s procompetitive rationales into account and assessing the flaws of less restrictive alternatives offered by Epic, the Court found that neither restriction violated the Sherman Act. The main justification given for this decision was that Epic failed to show that “the restraints are ‘patently and inexplicably stricter than is necessary.’” [30]
Judge Rogers did, however, find that Apple had violated California competition laws by prohibiting app providers from advertising alternative platforms to purchase in-app goods, such as a developers webpage. [31] In turn, the Court filed an injunction against Apple, preventing them from imposing their anti-steering provisions upon IOS developers.
B. Google and State Attorneys General
State Attorneys General have trailed their crosshairs on Google, seeking to use a public utility approach to force the ubiquitous search engine company to cease potentially unfair business practices.
Ohio Attorney General Dave Yost’s office filed a lawsuit against Google in the Delaware County Court of Common Pleas in Ohio on June 8th, 2021. [32] The AG’s office prays for a declaratory judgment ruling that Google Search is a public utility and seeks injunctive relief forcing the search engine to no longer give preference to its products over competitors in search results.
Attorney General Yost argues Google ought to be classified as a public utility. Google provides services free of charge and without a contract to all Ohioans. Over ninety percent of internet searches are through Google. The state accepts Google’s dominance in internet searching and seeks to ensure it is properly classified as a public utility since the nature of its operation is a matter of public concern, and its services are reasonably and indiscriminately made available to the public. If the court were to issue a declaratory judgment classifying Google Search as a common carrier, Google would be subject to heightened duties required of such entities under common law. In their rationale, Yost’s office cited Justice Thomas’ concurrence in Biden v. Knight First Amendment Institute at Columbia University, wherein Thomas argued digital platforms are sufficiently akin to common carriers or places of accommodation to be regulated. The Justice argued this was even clearer for digital platforms with dominant market shares, such as Google, since their widespread use yields an advantageous development of their search algorithm.
The second cause of action requesting a permanent injunction argues that as a common carrier, Google has a duty not to prioritize its own goods or services in search results over those of their competitors. To not give equal access to such mechanisms to their competitors violates their duties as public utility owners. Ohio has an interest in ensuring Google users are aware Google is a common carrier under Ohio law and that the company does not discriminate against third-party sites. A spokesperson for Google refuted the claim by Yost that they qualify as a public utility and warned regulation of this sort would be unpopular and detrimental to the consumers of Ohio. [33]
Yost isn’t the only Attorney General concerned about Google’s potentially anticompetitive practices. Texas Attorney General Ken Paxton is leading a coalition of seventeen state Attorneys General in a lawsuit against Google. [34] They claim Google successfully attempted to monopolize the digital advertising market in violation of Section 2 of the Sherman Act through anticompetitive conduct such as blocking ad publishers from sharing their user IDs with non-Google ad-buying tools. [35] Another claim argues Google violated Section 1 & 2 of the Sherman Act through a tying arrangement, a method in which two products in different markets are connected in purchases, where product A has dominant market power and thus coerces consumers into buying product B in the package, effectively translating the market power of Product A into the market power of Product B. In this case, Product A is the monopolistic AdX exchange from the previous claim and Product B is their DFP ad server. Finally, the coalition alleges Google unlawfully agreed with its competitor, Facebook, to restrain trade in violation of Section 1 of the Sherman Act. The claim states the two tech giants “agreed to allocate markets, manipulate [ad] publisher auctions, depress prices paid to publishers, and exclude rival ad networks.” Paxton alleges the violation is per se violative of Section 1 of the Sherman Act since it restrains trade and harms competition through an unlawful agreement. Several additional claims are tailored to the state antitrust laws of all 17 Attorneys General. Unlike the lawsuit by Ohio Attorney General Yost, this claim doesn’t pray for Google to be declared a common carrier. Rather, they seek injunctive relief to restore the competitive conditions to their state before the unlawful conduct and wish for Google to pay a litany of fines.
Introduction of Competing Theories:
The goals of antitrust legislation in the United States have been fiercely debated throughout the country’s history. The Sherman and Clayton antitrust acts have been subject to evolving interpretations and applications by courts and legal scholars. Since the late 1970s, the dominant mode of analysis for antitrust enforcement has been the consumer welfare standard, popularized by Judge Bork. [36] However, the emergence of “Big Tech” has motivated current efforts to reorient how the government approaches antitrust. In recent years, scholars and judges have advanced alternative doctrines, namely the competitive process and public utility standards.
On April 5th, 2021, the Supreme Court released a fairly mundane one sentence decision in the case of Biden v. Knight First Amendment Institute. The case revolved around whether former president Donald Trump’s decision to block Twitter users from his personal account constituted a violation of those users’ First Amendment rights. Presumably, since Trump was no longer president, the Supreme Court vacated the Second Circuit’s ruling and considered the case moot. [37]
But in a twelve-page concurring opinion, Justice Thomas offered a potential avenue to regulate digital platforms. Justice Thomas takes aim at the “concentrated control” digital platforms have over speech on the internet and their ability to suppress and deplatform content under the First Amendment. [38][39] Common carrier laws, according to Thomas, could be interpreted to require that digital platforms serve all content and users equally. Thomas builds his case by analogizing digital platforms to traditional common carriers. He argues that similar to traditional common carriers, digital platforms “‘carry’ information from one user to another…derive much of their value from network size,” and create high barriers to entries that preclude viable alternatives. [40] Justice Thomas ultimately concludes that if his analogy is correct, digital platforms’ “right to exclude” can be taken away with common carrier regulations. [41]
Common Carriers must meet two criteria under federal law. The carrier must “1) hold the service out as being available on standardized terms to the public… and 2) transmit signals without change in form or content.” [42] As Lister explains, the first prong divided common carriers under the Communications Act of 1934 from private carriers. The second prong, “draws the line between common carriers and providers of ‘enhanced services’ such as Internet access. The FCC has long declined to regulate enhanced service providers… as common carriers.”
Though Justice Thomas’s concurrence in Knight rightfully observes that digital platforms fulfill the first prong, he brushes over the second prong and the hesitancy of the FCC to regulate the internet under the common carrier model.
Consumer Welfare Standard:
The Consumer Welfare Standard (“CWS”) has gained traction as one of the most popular approaches to antitrust. CWS’s champion, Judge Bork, chronicled the origins and precise goals of the theory in his book, The Antitrust Paradox. In it, Bork writes, “The only legitimate goal of antitrust law is the maximization of consumer welfare. Therefore, competition is an art signifying a state of affairs in which consumer welfare can’t be increased artificially.” [43]
Bork believes the origins of the CWS trace back to the original intent of the Sherman Act itself. He goes so far as to argue the construction of the Sherman Act requires that “the law should be guided solely by the criterion of consumer welfare.” [44] Though Bork admits the Sherman Act is relatively barebone in content, he claims the early cases surrounding Sherman illustrate the legislative intent. In Standard Oil, Justice White declared not only the monopoly violative of Sherman but also the conduct by which they attempted to monopolize. The value this is based upon is fleshed out by the Clayton Act, which specifically outlaws attempts to “substantially lessen competition or tend to create a monopoly in any line of commerce.” Bork argues the two key terms are competition and monopoly. [45] Both are inherently economic concepts and we should understand them as such. The precise meaning of these terms he settles upon is as follows:
Competition…. as a term of art, designating a state of affairs in which the consumer welfare cannot be increased by moving to an alternative state of affairs through judicial decree. Conversely, ‘monopoly’ and ‘restraint of trade’ would be terms of art for situations in which consumer welfare could be so improved, and to ‘monopolize’ or engage in ‘unfair competition’ would be to use practices inimical to consumer welfare. [46]
Leaving the congressional intent to Bork, to give life to this definition it’s necessary to be explicit about what “consumer welfare” is. Bork explains that the law’s mission is to preserve the economic forces that compel “businesses to respond to consumers.” [47] This is measured by two factors, “a) the assignment or allocation of the available productive forces and materials among the various lines of industry, and b) the effective coordination of the various means of production in each industry into such groupings as will produce the greatest result.” Bork understands these two factors as allocative efficiency and productive efficiency. These comprise the level of society’s wealth, which he knows as consumer welfare. Emphasizing the paramount importance of consumer welfare again, Bork writes, “The whole task of antitrust can be summed up as the effort to improve allocative efficiency without impairing productive efficiency so greatly as to produce either no gain or a net loss in consumer welfare.” Though economics may be an uncomfortable field for some legal minds, Bork adheres to this standard because it is meant to be a neutral principle free of bias. Consumer welfare is a quantifiable metric under economic models and therefore courts can be more certain whether behavior truly is adverse to consumer welfare when applying the Sherman and Clayton Acts.
Having established consumer welfare as allocative and productive efficiency, let us turn to the goals by which the CWS aspires to reach its desired state of affairs. [48] First, monopolies are antithetical to productive and allocative efficiency and must be prohibited. Horizontal agreements, i.e., those between competitors in the same market, must also be excluded even though they tend not to harm consumer welfare as potently as monopolies. Horizontal mergers that create large market shares to decrease competition are harmful in the same way. Finally, deliberate market predation to drive rivals from a market, preventing market entry by rivals, or taking disciplinary actions against rivals is unacceptable as it prevents better efficiency under consumer welfare.
There are some business practices that the CWS would permit, as they aid consumer welfare. This includes price agreements, which, though they may appear anticompetitive, effectively increase allocative efficiency and supply consumers with more product than they could otherwise afford. Territories, ancillary suppression of rivals from markets, and internal growth leading to large market shares are supported by Bork under his CWS for similar reasoning.
The question of whether the CWS can be applied to modern digital issues remains, particularly when potential offenders such as Microsoft and Facebook provide free products that escape the consumer welfare metrics of allocative and productive efficiency that so neatly captured traditional markets.
New Brandeis movement:
The consumer welfare standard has faced increasing criticism in recent years. The most vocal critics make up the New Brandeis movement. This movement, named after US Supreme Court Justice Louis Brandeis, pushes back at many of the founding assumptions of the consumer welfare standard. One of the movement’s most ardent champions is Lina Khan, the current commissioner of the FTC. In her Yale Law Journal note, aptly titled Amazon’s Antitrust Paradox, Khan outlines her critiques of the consumer welfare standard and how the courts have applied it. Primarily, Khan rejects how the consumer welfare standard generally requires observed price increases or output restrictions to prove antitrust injury. [49] She views the abstraction of consumer welfare to these two criteria as myopic, failing to account for many practices that harm competition.
Instead, Khan advocates for a structuralist approach to antitrust. Working from the general principle that “concentrated market structures promote anticompetitive forms of conduct,” Khan wants antitrust law to examine how business and market structures affect competition. [50] Her structuralist approach would call on antitrust enforcers to prevent corporate actions that threaten competitive environments within markets. In her note, Khan primarily discusses two specific areas of regulation: predatory pricing and vertical mergers. She takes issue with the consumer welfare standard’s degradation of enforcement in these areas and beckons a return to past times when courts readily prevented such corporate behavior. While much of Khan’s note already lays out the historical basis for her view, several cases and statutes must be discussed to better understand the New Brandeisian’s structuralist views and how they apply to predatory pricing and vertical mergers.
Two particular cases exemplify the structuralist analysis Khan advocates for and how it relates to vertical mergers. In 1962, the Supreme Court considered the effects vertical mergers can have on a market in Brown Shoe Co., Inc. v. United States, 370 U.S. 294 (1962). [51] In this case, the Brown Shoe Company, a leading manufacturer of shoes, had acquired Kinney Shoes, a shoe retailer. The court was tasked with determining whether a district court’s decision that the merger violated Section 7 of Clayton was correct. Chief Justice Warren authored the unanimous decision arguing that the acquisition violated Clayton. Under the Sherman Act, it was still possible for a large company to make many small acquisitions that progressively increased the concentration of a market. Justice Warren asserts that it was for this reason that Congress amended Clayton to prevent mergers that threatened competition. Congress did not want the courts to allow such conduct and wait until a company grew into a large enough monopoly that a Sherman case could be brought against them. Instead, they intended for the courts to prevent mergers with monopolistic potential in their incipiency. Justice Warren further noted that Congress was “anxious” to preserve market structures in which numerous independent businesses existed. [52] While the court also made note of the fact that Brown had foreclosed a segment of the market through its practice of “forcing its own shoes upon its retail subsidiaries,” it was their structuralist approach that drove their decision. [53]
The court affirmed a similar analysis ten years later in Ford Motor Co. v. United States, 405 U.S. 562 (1972). [54] Ford, the automobile manufacturer, acquired Autolite, the second-largest independent supplier of spark plugs and other car parts, in 1962. In a 5-2 decision delivered by Justice Douglas, the court upheld a ruling that the merger violated the Celler-Kefauver Antimerger Act, an amendment to the Clayton Act. Douglas roots his ruling in an analysis of the structure of the spark plug and automobile markets. He argues that Ford’s prior position as an “interested firm on the outside” of the spark plug market invoked competition among current competitors. [55] Before the acquisition, since Ford was a major customer of the spark plug market, firms were forced to compete with one another for Ford’s business. Additionally, with Ford no longer being a significant spark plug purchaser, the barriers to entry into the market had increased since fewer buyers existed. Douglas acknowledged that there were arguments that the merger had led to greater efficiencies for Autolite to compete with other spark plug manufacturers. However, he dismissed these claims on the grounds that Congress intended the courts to maintain a competitive structure in the economy and prevent anticompetitive mergers of “the benign and the malignant alike.” [56] Once again, the court appealed to a structural analysis that placed little importance on price or outputs.
The roots of structural analysis can also be found much earlier, in discussion of another anti competitive action. Predatory pricing is the practice of setting prices so low that competing businesses are pushed out of the market. [57] Since the adoption of the consumer welfare standard, statutes that outlaw the practice have infrequently been enforced. [58] This owes to the fact that economic theories behind the consumer welfare standard consider predatory pricing an irrational scheme that rarely works. [59] However, before the rise of Bork, the courts took a different stance on the issue. Khan points to Standard Oil as the origin of the jurisprudence against predatory pricing. In this case, the legality of Standard Oil’s trust and monopolistic practices, such as cutting prices to push competitors out of the market, were under dispute. [60] In a decision penned by Justice White, the Supreme Court ultimately ruled that Standard Oil had violated the Sherman Antitrust Act. The decision mainly rested on the trust Standard Oil had formed. However, the court also took note of “specific acts” carried out by Standard Oil, such as predatory pricing, that were enabled by its trust. [61] The sheer size and power of Standard Oil allowed it to bleed cash in certain markets and drive out rivals while making up the losses in other markets. Thus, not only was the structure of Standard Oil itself found to be unlawful, but its structure enabled it to continue tightening its grip on the market. Khan emphasizes that this ruling was cited by courts in future antitrust cases that found predatory pricing to violate the Sherman Act. [62]
Endeavors to stamp out predatory pricing continued with the passing of the Clayton Antitrust Act. Section 2 of Clayton outlawed price-cutting practices that would either elicit the creation of a monopoly or reduce competition. [63] However, loopholes created by the wording of the statute allowed certain forms of predatory pricing to continue. [64] In response, Congress continued its quest to curb predatory pricing by passing the Robinson-Patman Act in 1936. This amendment to the Clayton Act added new restrictions to price discrimination targeting both sides of the supply chain: retailers and producers. In addition, Section 3 of the Robinson-Patman Act added criminal penalties to the types of price discrimination restricted by Clayton. [65] Khan claims that Standard Oil and the subsequent Clayton and Robinson-Patman Acts outline a rich history of predatory pricing litigation in the United States. She argues for a return to active prosecution of predatory pricing in both current and future cases.
Having covered part of the extensive history Khan appeals to when advocating for Brandeisian structuralist views, let us summarize the main principles of the theory. The New Brandeis movement prioritizes promoting competitive markets rather than explicit consumer welfare. In light of this, the movement also advocates for more proactive government enforcement of antitrust laws. Through the eyes of the Brandeisians, current antitrust intervention only comes after a company begins exercising market power it has already acquired. Instead, they seek to prevent the creation of market structures that allow for a firm to become so dominant in the first place. This is why threats to competitive markets that may still promote efficiencies, such as certain vertical mergers, are not allowed under Brandeisian standards.
Application of Theories to Current Issues:
Though there’s no dispute that the history of antitrust legislation, precedent, and theory is rich, the path forward for antitrust in the digital age is neither well beaten nor a yellow brick road. The remainder of this article will assess the strengths and weaknesses of potential approaches to these quandaries before the enduring footprint of precedent is stamped.
A. State Attorneys General and Google
i) Public Utility as a Bad Fit (Yost)
The first string of cases that need some soul-searching are those involving state Attorneys General and Google. The first approach, which seems likely to yield undesirable results, is Ohio Attorney General Yost’s push to declare Google as a public utility. The tempered antitrust suit led by Texas Attorney General Paxton has a strong claim based on its ability to pin the Defendant with anticompetitive behavior and the desire to revert back to the status quo.
Yost’s quest to declare Google a public utility of Ohio is misguided. As already explained, though Google may indeed meet the first prong of the common carrier test by offering standardized services available to all Ohioans, in no way does Google fulfill the second prong of merely transmitting information like a cellular company. Google is transparent about the three step process their search engine uses. [66] First, the software continually does a search called “crawling” that involves finding new websites to be added to the search engine’s list of known pages. Second, a process of “indexing” analyzes the content of the websites and categorizes them while excluding redundant duplicates. Finally, the “serving” of search results provides users with websites matching the consumer’s search terms from the index. This three step process is transformative, as it sorts through an incomprehensible amount of communications and delivers the consumers what it deems to be the highest quality results in a tailored and readable format. This is a far more complex process with an editorial role uncharacteristic of a cellular company transmitting two parties’ conversation without modification. [67]
Justice Thomas’ concurrence raises strong points, namely that the size of Google’s platform increases its value and creates high barriers to entry for alternatives. [68] The crucial index that Google search relies upon is improved and refined by every user and search. While competitors such as Yahoo, Bing, and DuckDuckGo fight for a share of the internet search engine market, the universal use and addition of the verb “Google” to the dictionary communicate the monopolistic dominance of the product. But is this necessarily a bad thing? The Sherman Act explicitly bans horizontal mergers of monopolistic proportions and the Clayton Act bans attempts to do so. Yost alleges neither of Google; his problem is rather that their internal growth into a monopolist is problematic, something Justice Thomas also seems to imply.
The monopolistic effect is something Yost accepts “as a fact (be it good or bad).” [69] His concern is that Google’s other horizontal businesses such as advertising or services like Google Maps benefit from preferential placement at the top of Google search results. Though Yost does not explicitly invoke it by name, this sounds similar to an illegal tying arrangement. But rather than bring causes of action under Section 1 of the Sherman Act and Section 5 of the Clayton Act, Yost decides the best way to rectify this potentially anticompetitive behavior is to transform the company into a common carrier. This would require Google to “carry search result information reliably, neutrally, and without unreasonable discrimination.” [70] Effectively, Google would no longer be able to leverage their horizontal power.
Putting aside the questionable classification of Google as a common carrier, the rival Consumer Welfare Standard Borkians and Neo-Brandeisians would have conflicting views about Yost’s claims. The Borkians would support Google in this suit. Though they would acknowledge the precedent sides in favor of Yost, tying arrangements may be “used to achieve economics of scale, nondiscriminatory measurement of use, and efficient technological interdependence [that] are valuable not merely to the firm but to consumers.” [71] Again, though it may be an exercise of monopolistic power, the question is whether the effects are adverse for consumer welfare. In this case, it is likely the search engine and map search engine are tied because they become more efficient in the interest of the consumer. For instance, when a consumer searches for “pizza near me,” the search engine might present pizzeria restaurants while the map search engine refines the results to those physically near the consumer and factors in data on traffic to show those the shortest time away. Though Google’s two products in this case are in slightly different markets, their tying arrangement is justified by a level of efficiency that other map search products may not provide.
The second Google suit, led by Attorney General Paxton, is a better approach to antitrust as it stays within the confines of the Sherman and Clayton Acts. [72] Google is first accused of successfully attempting to monopolize the digital ad market through anticompetitive conduct in violation of Section 2 of the Sherman Act. Second, Google is accused of violating Section 1 and 2 of the Sherman Act through the tying arrangement between their AdX exchange and the DFP ad server. Third, the company is accused of violating Section 1 of the Sherman Act by restraining trade though an unlawful agreement with Facebook.
This case is reminiscent of U.S. v. Microsoft Corp.. [73] In this case, Microsoft was accused of violating Sections 1 & 2 of the Sherman Act by a) maintaining their monopoly for PC operating systems through anticompetitive conduct and b) tying their dominant PC operating system product to their internet browser Internet Explorer in an attempt to monopolize that market. The D.C. Circuit decided that Microsoft employed anticompetitive behavior in maintaining its operating system monopoly and that they anti-competitively tied the internet browser market. In their reasoning, after finding that Microsoft qualified as a monopoly with a 95% share of the operating system market, the court next determined whether Microsoft possessed monopoly power. The court considered whether they had done so through internal growth or anticompetitive conduct. The consumer welfare standard is in use to a degree here, because the monopolistic share alone was not indicative of a Sherman Act violation. The court reasoned “a predominant market share does not itself indicate monopoly power.” Rather, the barriers to entry must be considered. Since in the software market consumers prefer operating systems with a large library of applications and developers prefer to make applications for operating systems with a large number of consumers, Microsoft could naturally continue to be a monopoly by the choice, and to the benefit, of the consumers. However, the court found Microsoft had taken action to stop software competition from Apple, Java, and others through exclusionary deals and halting interoperability of software. These predatory tactics prevent the possibility of higher consumer welfare and are thus violative of the Sherman Act. The same was found for tying Microsoft’s web browser to its operating system with no option to uninstall, exclusionary conduct in the browser market not in the interest of consumer welfare.
Under the precedent of Microsoft, it is reasonable that Paxton may prevail in his case against Google. If Paxton can successfully argue that Google’s supposed 90 percent share of the digital advertising market qualifies as monopolistic, the question becomes whether the monopoly is violative of Sherman. Under the Borkian approach to the Microsoft precedent, internal growth to a monopoly level is acceptable. The problem arises when the conduct is unduly coercive of others’ rights to conduct business or results in poor consumer welfare, as the Court decided in Standard Oil and American Tobacco Co. Paxton alleges Facebook and Google made an internal agreement, dividing up who would win auctions for header advertisements when they bet against one another. This would be anticompetitive conduct under the Court’s reasoning in Addyston Pipe & Steel Co., qualifying as a naked restraint in violation of the Sherman Act. The allegation that their tying agreement is improper would likely also prevail for the same reasons as the Yost case.
The principle that competitive market structures ought to be maintained is key to understanding the Brandeisian perspective here. Brandeisians would have fought tooth and nail to prevent several of Google’s acquisitions outlined by AG Paxton. Drawing on the Court’s ruling in Ford Motor Co. v. United States, Brandeisians would likely reason that the acquisitions of DoubleClick and Invite Media reduced competition by increasing concentration in the ad server and demand side platform markets. Moreover, Google’s continued acquisitions combined with their in-house growth, such as Ad Exchange, demonstrated a growing trend in the concentration of the digital advertising market. In Brown Shoe Co., Inc. v. United States, a cornerstone of Brandeisian theory, the Court ruled that “tendencies toward concentration in industry are to be curbed in their incipiency.” This precedent similarly applies to the case with Google. Moving on to Paxton’s actual charges, namely the allegations of tying and collusion, Brandeisians would view these as manifestations of Google’s dominance over the market. Similar to the Borkians, the Brandeisians would consider these acts anticompetitive. However, it should be noted that, as outlined above, the Brandeisians would have attempted to prevent this situation in the first place. Mergers and acquisitions that increase a firm’s dominance over a market given its existing business structures are clear anticompetitive acts under Brandeisian theory.
B. Apple v. Epic
The second digital platform case that bore interesting fruit is Epic Games, Inc. v. Apple Inc.. [74] Epic’s concern was that Apple’s control of the App Store for their iOS devices was anticompetitive, as Apple collected a 30% commission on all in-app purchases (“IAP”), with no options for developers to direct consumers to payment methods outside the Apple App Store. When the Plaintiff directed consumers of their “Fortnite” application on the Apple App Store platform to their own payment system for the “V-Bucks” in-game currency, Apple lost out on commission from over $700 million of transactions. When the Plaintiff refused to only use Apple’s IAP payment method, their application was suspended from Apple’s platform. In Counts 1, 3, 4, and 5, the Court examined whether Apple’s anti-direction policy violated Sections 1 and 2 of the Sherman Act. [75] After deciding that Apple’s near 57% market share of mobile gaming transactions failed to meet the 70% bar for monopoly power, the court decided that the defendants in fact had market power. In addition, mobile game transactions increased in output, which is uncharacteristic of monopoly-dominated markets. Though the 30% commission may result in supracompetitive pricing, the output impact is unsupported. Next, the Court examined whether Apple violated Section 1 of Sherman by restraining trade in the App Store after market or restraining the IAP market. Applying Standard Oil Co., the Court sought to decide whether the restrictions were in the interest of the consumer when assessing the Section 1 claims. Ultimately, the Court labeled both of Apple’s restraints anticompetitive, as the prohibition of alternative game stores and the mandatory 30% commission limited competition and prevented some potential savings from being passed onto consumers. Apple’s justifications for the conduct therefore had to be proved to be in the interest of the consumer to be reasonable and survive scrutiny. [76] Thus, they successfully argued their restrictions increased security, intrabrand competition, and intellectual property investment. Finally, the Court considered whether Apple could have achieved these procompetitive justifications through less restrictive means. Finding no reasonable alternative to app security verification, the Court concluded Apple’s restriction of app distribution did not violate Section 1 of the Sherman Act. [77] The restriction of the payment methods to the Apple IAP survives under Section 1 for the consumer benefit of centralization and the fairness of Apple profiting from its platform. Epic challenges both restrictions under Section 2 as well, arguing monopoly maintenance. The Court understands monopoly power as “the power to control prices or exclude competition.” [78] With respect to the app distribution part of the claim, Epic’s claim fails because they fail to prove monopolistic power in the global mobile gaming transaction market. The IAP monopoly claim fails for the same reason. Claim 6 examines whether the tying arrangement between app distribution and the IAP model violated Section 1 of the Sherman Act. They fail to show an illegal tie because the platform and the IAP are fundamentally interconnected and Epic failed to prove a consumer demand for their separation. Though Apple may survive the Sherman Act, “antitrust law does not end with the Sherman Act.” [79] Epic’s one victory comes under California’s Unfair Competition Law, which finds the anti-steering policy banning purchase methods other than Apple’s IAP anticompetitive. By preventing developers like the Plaintiff from directing consumers to potentially lower prices off the iOS platform or informing them of Apple’s 30% commission, Apple restricts information on price information against the interest of the consumers. This lack of information prohibits consumers from making informed decisions and “may create the potential for anticompetitive exploitation of consumers.” [80] Therefore, the Court finds the anti-steering policy violative of the California law.
While Epic squeaks out with a small victory due to Apple violating the California law, the Borkian and Brandeisians would have strong feelings about the application of the Sherman Act in this case.
The consumer welfare standard camp would have strong support for the Court’s siding with Apple in this decision. The manner in which the Court considered the effects of the conduct on consumers is exactly what Bork wanted. The fact that the supracompetitive pricing increased output would be seen as positive market efficiency, and what’s more, deciding the reasonableness of market restrictions based on the benefit to the consumer is the hallmark of consumer welfare advocates. The validity of the tying arrangement would be supported by Borkians for the same reasons as the previously analyzed Google cases. The decision in favor of Epic might even be looked upon fondly by consumer welfare advocates; first, preventing exploitation of consumers is inarguably in their interest, and second, an alternative to the Apple IAP offers a possibility for lower prices and therefore increased efficiency.
On the other hand, the Brandeisians would reject the Court’s analysis of the relevant market: mobile gaming transactions. The Court reasons that Apple has significant competition in this market, citing Google and Sony’s respective app stores. This variety of options is said to give consumers “a choice of devices and transaction platforms” from which they can download mobile games and buy in-game products. However, the Court incorrectly assumes that the existence of these alternative platforms equates to their accessibility. Users who only have access to an IOS platform are limited to using Apple’s App Store and its IAP system. For these users, there is no competitor readily available to them unless they have a non-Apple mobile device. This underlying mobile phone market structure is what the Brandeisians would look at to substantiate claims that Apple’s App Store restrictions do in fact restrain trade.
C. FTC v. Facebook
Finally, the ongoing battle between the government and a social media giant in Federal Trade Commission v. Facebook merits attention. In a new amended complaint, the FTC launched serious anticompetitive allegations against Facebook. [71] Keying in on the social media company’s acquisition of Instagram and WhatsApp, the FTC has brought two Section 2 violations of the Sherman Act to the table. In the first count, the FTC alleges that Facebook has built and maintained a social media monopoly through illegal acquisitions of WhatsApp and Instagram rather than competing and innovating against rivals. The complaint also claims that the integration of the two services into the Facebook ecosystem has built barriers to entry that significantly increased the difficulty of entering the photo-sharing and mobile messaging markets. The second count takes issue with Facebook’s anticompetitive restrictions over its API usage. In 2011, Facebook released policies that disallowed developers from using their API to create or promote competitors to Facebook. [82] If found doing so, their API access would be revoked, making it impossible for the developers to communicate and work with the Facebook platform. Internal communications between Facebook employees have disclosed that these policies were used against up-and-coming social media networks like Vine. [83]
Given that this lawsuit is headed by the FTC and was in part approved by Chair Khan, the complaint contains several appeals to Brandeisian theory. Facebook, WhatsApp, and Instagram all offer their primary services for free. None of this changed when Facebook acquired the latter two. The amended complaint argues that despite this, competition and therefore consumers have been harmed by Facebook’s actions. The FTC claims that Facebook’s acquisitions freed the company from having to compete with up-and-coming rival social media platforms. This in turn foreclosed consumers from any benefits such competition would bring about, such as increased innovation. Additionally, the barrier to entry into the social media market grew higher following these acquisitions, limiting the chances of new rivals entering the market and thus limiting consumers’ potential choices. This line of argumentation is a quintessential example of Brandeisian structuralist analysis. The core of the FTC’s complaints lies in the harm Facebook’s acquisitions have had on the competitive processes within the social media market. Moreover, harms to consumers are identified not through changes in prices or outputs, but rather through innovation and choice.
The Facebook lawsuit is a difficult situation for the consumer welfare standard. Shackled so intently to an economic analysis of price efficiency, the free nature of the products Facebook produces throws a wrench into Borkian analysis. However, the rising prevalence of free digital platforms requires the standard to adapt or fall into irrelevancy. The first issue is Facebook’s aggressive purchasing of smaller social media platforms and the subsequent integration of them into its ecosystem. In a sense, this could be viewed as an ancillary restraint on competition. Facebook has an interest in creating the best platform for its users. When smaller companies arise with innovative features, they share Facebook’s interests in positive consumer experiences but they also have an interest in growth. Facebook offers a pathway to such growth through its massive user base, alongside a large paycheck. Though this may lead to fewer long-term competitors the size of Facebook, it’s difficult to say consumer interests are unrepresented. They receive the same products in an easy-to-access centralized platform. Some, such as the Brandeisians, may still protest about the barriers to entry Facebook’s acquisition-centric behavior produces. Yet, over the last three years, TikTok has risen from obscurity to a social media titan billed on the same level as any other social media company. Even though Facebook did not acquire them, Facebook has mimicked TikTok’s features through Instagram Reels. The barrier to entry is not insurmountable, even if the case of TikTok only offers anecdotal evidence of this. The barrier is simply user growth, which will come with viable and competitive products. The second issue is Facebook’s ban on using their API to produce potential competitors. While Facebook’s interest in this is clear, the interest to the consumer is not. Curtailing innovation or competition is a naked restraint on competition. No further analysis is needed to determine the Borkian disapproval of such a practice. Therefore, the FTC lawsuit ought to fail on the first issue, but prevail on the second under the consumer welfare standard.
Conclusion:
In conclusion, the newfound dilemmas of digital antitrust have a firm foundation in case law and theory to expand upon. The consumer welfare standard, though primarily concerned with price to consumers, survives as a viable approach to antitrust through an examination of effects upon consumer interests. Meanwhile, the New Brandeis movement presents a promising approach to antitrust in the wake of growing concerns about the economic power of Big Tech. With Lina Khan at the head of the FTC, only time will tell how courts respond to her and the other Brandeisians’ structuralist approach to antitrust. While the consumer welfare standard is currently the default mode of analysis for antitrust, perhaps another antitrust revolution is in its incipiency.
Works Cited
Part I
[1] Highest price NFT stats – collections and individual sales listed, CoinMarketCap, <https://coinmarketcap.com/nft/> [Accessed 1 May 2022].
[2] New Scientist. 2022. What is bitcoin and how does it work?. [online] Available at: <https://www.newscientist.com/definition/bitcoin/#ixzz7RE5gfAI3> [Accessed 16 May 2022].
[3] What Is Blockchain Technology? How Does It Work? | Built In. [online] Available at: <https://builtin.com/blockchain> [Accessed 14 May 2022].
[4] CoinDesk. 2022. What Is KYC and Why Does It Matter For Crypto?. [online] Available at: <https://www.coindesk.com/learn/what-is-kyc-and-why-does-it-matter-for-crypto/#:~:text=KYC%20means%20%22know%20your%20customer,use%20to%20fight%20money%20laundering.> [Accessed 14 May 2022].
[5] Dewey, J., 2022. Blockchain & Cryptocurrency Laws and Regulations | USA | GLI. [online] GLI – Global Legal Insights – International legal business solutions. Available at: <https://www.globallegalinsights.com/practice-areas/blockchain-laws-and-regulations/usa> [Accessed 14 May 2022].
[6] Digiconomist. 2022. Bitcoin Energy Consumption Index – Digiconomist. [online] Available at: <https://digiconomist.net/bitcoin-energy-consumption> [Accessed 14 May 2022].
[7] The White House. 2022. FACT SHEET: President Biden to Sign Executive Order on Ensuring Responsible Development of Digital Assets | The White House. [online] Available at: <https://www.whitehouse.gov/briefing-room/statements-releases/2022/03/09/fact-sheet-president-biden-to-sign-executive-order-on-ensuring-responsible-innovation-in-digital-assets/> [Accessed 15 May 2022].
[8] McKinsey. 2021. The State of AI in 2021. [online] Available at: <https://www.mckinsey.com/business-functions/mckinsey-analytics/our-insights/global-survey-the-state-of-ai-in-2021> [Accessed 15 May 2022].
[9] Davenport, T. and Kalakota, R., 2022. The potential for artificial intelligence in healthcare. [online] NIH. Available at: <https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6616181/> [Accessed 16 May 2022].
[10] Professional Development | Harvard DCE. 2022. Business Applications for Artificial Intelligence: An Update for 2020 – Professional Development | Harvard DCE. [online] Available at:<https://professional.dce.harvard.edu/blog/business-applications-for-artificial-intelligence-an-update-for-2020/> [Accessed 10 May 2022].
[11] E, S., 2022. Ethically Automating Legal Practice. [online] Medium. Available at: <https://medium.com/fair-bytes/ethically-automating-legal-practice-6265beb54f14> [Accessed 12 May 2022].
[12] Hind, M., 2022. Foundations of trustworthy AI: How mature is your AI governance? – WatsonBlog.[online]WatsonBlog.Available at: <https://www.ibm.com/blogs/watson/2021/06/ai-governance-maturity/> [Accessed 14 May 2022].
[13] Baker, D., 2022. Tesla settles lawsuit over self-driving car secrets. [online] SFGATE. Available at: <https://www.sfgate.com/business/article/Tesla-settles-lawsuit-over-self-driving-car-11084936.php> [Accessed 14 May 2022].
[14] Tesla.com. 2022. Autopilot and Full Self-Driving Capability. [online] Available at: <https://www.tesla.com/support/autopilot> [Accessed 11 May 2022].
[15] Holzer, H., 2022. Understanding the impact of automation on workers, jobs, and wages. [online] Brookings. Available at: <https://www.brookings.edu/blog/up-front/2022/01/19/understanding-the-impact-of-automation-on-workers-jobs-and-wages/> [Accessed 15 May 2022].
[16] National Artificial Intelligence Initiative. 2022. The National Artificial Intelligence Initiative (NAII). [online] Available at: <https://www.ai.gov/> [Accessed 16 May 2022].
[17] Sussmen, Heather “U.S. Artificial Intelligence Regulation Takes Shape”, Orrick, November.18.2021, Accessed December 5, 2021.
[18] Engler, A., 2022. The EU and U.S. are starting to align on AI regulation. [online] Brookings. Available At: <https://www.brookings.edu/blog/techtank/2022/02/01/the-eu-and-u-s-are-starting-to-align-on-ai-regulation/> [Accessed 16 May 2022].
[19] LII Staff, Personal information Legal Information Institute (2020), https://www.law.cornell.edu/wex/personal_Information (last visited Jan 21, 2022).
[20] Sarah L. Swan & Stephen E. Sachs, Transunion v. Ramirez Harvard Law Review (2021), https://harvardlawreview.org/2021/11/transunion-v-ramirez/ (last visited Jan 21, 2022).
[21] Thorin Klosowski, The State of Consumer Data Privacy Laws in the US (and why it matters) The New York Times (2021)
[22] Jason Furman, The value and challenges of regulating Big Tech Harvard Kennedy School (2020), https://www.hks.harvard.edu/faculty-research/policy-topics/business-regulation/value-and-challenges-regulating-big-tech (last visited Jan 21, 2022).
[23]David Mccabe & Cecilia Kang, As Congress dithers, states step in to set rules for the internet The New York Times (2021), https://www.nytimes.com/2021/05/14/technology/state-privacy-internet-laws.html (last visited Apr 30, 2022).
[24] Noah Joshua Phillips, Opinion | facebook settlement vindicates consumer privacy The Wall Street Journal (2019), https://www.wsj.com/articles/facebook-settlement-vindicates-consumer-privacy-11564008019 (last visited Jan 21, 2022).
[25] Venky Anant et al., The consumer-data opportunity and the privacy imperative McKinsey & Company (2021), https://www.mckinsey.com/business-functions/risk-and-resilience/our-insights/the-consumer-data-opportunity-and-the-privacy-imperative (last visited Jan 25, 2022).
[26] Erika M Douglas, The new antitrust/Data Privacy Law Interface The Yale Law Journal (2021), https://www.yalelawjournal.org/forum/the-new-antitrustdata-privacy-law-interface (last visited Jan 21, 2022).
Part II
[1] Sherman Anti-Trust Act of 1890, 15 U.S.C. §§ 1-38
[2] National Archives, Sherman Anti-Trust Act (1890), https://www.archives.gov/milestone-documents/sherman-anti-trust-act#:~:text=The%20Sherman%20Anti%2DTrust%20Act%20authorized%20the%20federal%20government%20to,foreign%20nations%22%20was%20declared%20illegal.
[3] Sherman Anti-Trust Act of 1890, 15 U.S.C. §§ 1
[4] Sherman Anti-Trust Act of 1890, 15 U.S.C. §§ 2
[5] Sherman Anti-Trust Act of 1890, 15 U.S.C. §§ 4
[6] Sherman Anti-Trust Act of 1890, 15 U.S.C. §§ 6
[7] Sherman Anti-Trust Act of 1890, 15 U.S.C. §§ 7
[8] Robert H. Bork, The Antitrust Paradox: A Policy At War With Itself 16 (2021)
[9] Robert H. Bork, The Antitrust Paradox: A Policy At War With Itself 17 (2021)
[10] National Archives, Sherman Anti-Trust Act (1890), https://www.archives.gov/milestone-documents/sherman-anti-trust-act#:~:text=The%20Sherman%20Anti%2DTrust%20Act%20authorized%20the%20federal%20government%20to,foreign%20nations%22%20was%20declared%20illegal.
[11] Northern Securities Co. v. United States, 193 U.S. 197 (1904) (Holmes J. dissenting opinion)
[12] Robert H. Bork, The Antitrust Paradox: A Policy At War With Itself 18-42 (2021)
[13] United States v. Trans-Missouri Freight Ass’n, 166 U.S. 290 (1897)
[14] Addyston Pipe and Steel Company et al., Appts., v. United States, 175 U.S. 211 (1899)
[15] U.S. v. Addyston Pipe and Steel Co., 85 F. 271 (6th Cir. 1898)
[16] Northern Securities Co. v. United States, 193 U.S. 197 (1904)
[17] Northern Securities Co. v. United States, 193 U.S. 197 (1904) (Holmes J. dissenting opinion)
[18] Standard Oil Co. of New Jersey v. United States, 221 U.S. 1 (1910)
[19] Northern Securities Co. v. United States, 193 U.S. 197 (1904) (Holmes J. dissenting opinion)
[20] Standard Oil Co. of New Jersey v. United States, 221 U.S. 1 (1910) (Harlan J. dissenting opinion)
[21] American Tobacco Co. v. United States, 328 U.S. 781 (1946)
[22] Clayton Antitrust Act of 1914, 15 U.S.C. §§ 12-27; 29 U.S.C. §§ 52-53
[23] Historical Highlights: The Clayton Antitrust Act, History, At & Archives: United States House of Representatives, https://history.house.gov/HistoricalHighlight/Detail/15032424979#:~:text=Aside%20from%20banning%20the%20practices,law%20on%20October%2015%2C%201914.
[24] Clayton Antitrust Act of 1914, 15 U.S.C. §§ 14
[25] Robert H. Bork, The Antitrust Paradox: A Policy At War With Itself 43 (2021)
[26] Chicago Board of Trade v. United States, 246 U.S. 231 (1918)
[27] James Cook, Matthew Field & James Titcomb, Project Liberty: Inside epic’s secret war with Apple over fortnite The Sydney Morning Herald (2021), https://www.smh.com.au/technology/video-games/project-liberty-inside-epic-s-secret-war-with-apple-over-fortnite-20210409-p57hp7.html
[28] Epic Games v. Apple Inc., 493 F. Supp. 3d 817 (N.D. Cal. 2020)
[29] Kellen Browning, Apple Appeals App Store ruling in fight with epic games The New York Times (2021), https://www.nytimes.com/2021/10/08/technology/apple-epic-games-lawsuit.html
[30] Epic Games v. Apple Inc., 493 F. Supp. 3d 817 (N.D. Cal. 2020)
[31] Sarah Perez, Apple ordered to comply with Court’s decision over in-app payments in Epic games case TechCrunch (2021), https://techcrunch.com/2021/11/09/apple-ordered-to-comply-with-courts-decision-in-epic-games-case-over-in-app-payments/
[32] Office of Ohio Attorney General Dave Yost, AG Yost Files Landmark Lawsuit to Declare Google a Public Utility (2021), https://www.ohioattorneygeneral.gov/Media/News-Releases/June-2021/AG-Yost-Files-Landmark-Lawsuit-to-Declare-Google-a
[33] Rebecca Klar, Ohio files lawsuit to declare Google a public utility (2021), available at https://thehill.com/policy/technology/557366-ohio-files-lawsuit-to-declare-google-a-public-utility/
[34] Office of Texas Attorney General Ken Paxton, AG Paxton: Alaska, Florida, Montana, Nevada and Puerto Rico Join Texas-Led Bipartisan Antitrust Lawsuit Against Google (2021), https://www.texasattorneygeneral.gov/news/releases/ag-paxton-alaska-florida-montana-nevada-and-puerto-rico-join-texas-led-bipartisan-antitrust-lawsuit
[35] Office of Texas Attorney General Ken Paxton, Paxton Files Third Amendment in Antitrust Lawsuit Against Google (2021), https://www.texasattorneygeneral.gov/news/releases/paxton-files-third-amendment-antitrust-lawsuit-against-google
[36] Markham William, The consumer-welfare standard should cease to be the North Star of Antitrust Markham Markham Law (2021), https://www.markhamlawfirm.com/the-consumer-welfare-standard-should-cease-to-be-the-north-star-of-antitrust/
[37] Biden v. Knight First Amendment Inst. at Columbia Univ., 141 S. Ct. 1220 (2021)
[38] Biden v. Knight First Amendment Inst. at Columbia Univ., 141 S. Ct. 1220 (2021) (Thomas J. concurring opinion)
[39] Rebecca Van Burken & Adrian Moore, Social media companies have the right to ban users Reason Foundation (2021), https://reason.org/commentary/social-media-companies-have-the-right-to-ban-users/
[40] Biden v. Knight First Amendment Inst. at Columbia Univ., 141 S. Ct. 1220 (2021) (Thomas J. concurring opinion)
[41] Biden v. Knight First Amendment Inst. at Columbia Univ., 141 S. Ct. 1220 (2021) (Thomas J. concurring opinion)
[42] James H. Lister, The Rights of Common Carriers and the Decision Whether to Be a Common Carrier or a Non-Regulated Communications Provider, Federal Communications Law Journal: Vol. 53: ss. 1, Article 7, 9 (2000), https://www.repository.law.indiana.edu/cgi/viewcontent.cgi?article=1262&context=fclj
[43] Robert H. Bork, The Antitrust Paradox: A Policy At War With Itself 48 (2021)
[44] Robert H. Bork, The Antitrust Paradox: A Policy At War With Itself 54 (2021)
[45] Robert H. Bork, The Antitrust Paradox: A Policy At War With Itself 55 (2021)
[46] Robert H. Bork, The Antitrust Paradox: A Policy At War With Itself 58 (2021)
[47] Robert H. Bork, The Antitrust Paradox: A Policy At War With Itself 90 (2021)
[48] Robert H. Bork, The Antitrust Paradox: A Policy At War With Itself 422 (2021)
[49] Lina M. Khan, Amazon’s Antitrust Paradox, 126 YALE L.J. 710 (2017), https://www.yalelawjournal.org/note/amazons-antitrust-paradox
[50] Lina M. Khan, Amazon’s Antitrust Paradox, 126 YALE L.J. 718 (2017), https://www.yalelawjournal.org/note/amazons-antitrust-paradox
[51] Brown Shoe Co. v. United States, 370 U.S. 294, 82 S. Ct. 1502 (1962)
[52] Brown Shoe Co. v. United States, 370 U.S. 294, 82 S. Ct. 1502 (1962)
[53] Brown Shoe Co. v. United States, 370 U.S. 294, 82 S. Ct. 1502 (1962)
[54] Brown Shoe Co. v. United States, 370 U.S. 294, 82 S. Ct. 1502 (1962)
[55] Ford Motor Co. v. United States, 405 U.S. 562, 92 S. Ct. 1142 (1972)
[56] Ford Motor Co. v. United States, 405 U.S. 562, 92 S. Ct. 1142 (1972)
[57] Will Kenton, Predatory pricing Investopedia (2022), https://www.investopedia.com/terms/p/predatory-pricing.asp
[58] Antitrust Division Workload Statistics FY 1970-1979, The United States Department of Justice (2015), https://www.justice.gov/atr/antitrust-division-workload-statistics-fy-1970-1979. Antitrust Division Workload Statistics FY 1990 – 1999, The United States Department of Justice (2015), https://www.justice.gov/sites/default/files/atr/legacy/2009/06/09/246419.pdf. Antitrust Division Workload Statistics FY 2010 – 2019, The United States Department of Justice (2015), https://www.justice.gov/atr/file/788426/download
[59] Eric Helland, Book review: Are predatory commitments credible? Who should the courts believe?, by John R. Lott Jr.. The Independent Institute (2001), https://www.independent.org/publications/tir/article.aspid=201#:~:text=The%20Chicago%20School%20analysts%20showed,losses%20in%20the%20short%20run
[60] Lina M. Khan, Amazon’s Antitrust Paradox, 126 YALE L.J. 723 (2017), https://www.yalelawjournal.org/note/amazons-antitrust-paradox
[61] Standard Oil Co. v. United States, 221 U.S. 1, 31 S. Ct. 502 (1911)
[62] Lina M. Khan, Amazon’s Antitrust Paradox, 126 YALE L.J. 723 (2017), https://www.yalelawjournal.org/note/amazons-antitrust-paradox
[63] Richard Schmalensee, Robert D. Willig & Hal Varian, Price discrimination, in Handbook of Industrial Organization: Vol.: 1 643 (1989).
[64] Richard Schmalensee, Robert D. Willig & Hal Varian, Price discrimination, in Handbook of Industrial Organization: Vol.: 1 644 (1989).
[65] Dillon, C. Brien, CRIMINAL PENALTIES, SECTION 3 OF ROBINSON-PATMAN ACT—”DEAD HORSE” OR “SLEEPER”?, Section of Antitrust Law, vol. 8, 1956, pp. 112–124, http://www.jstor.org/stable/25749983
[66] Google Search Central, How Search Works for Site Owners (2022), https://developers.google.com/search/docs/advanced/guidelines/how-search-works#:~:text=Crawling%3A%20Google%20downloads%20text%2C%20images,which%20is%20a%20large%20database.
[67] James H. Lister, The Rights of Common Carriers and the Decision Whether to Be a Common Carrier or a Non-Regulated Communications Provider, Federal Communications Law Journal: Vol. 53: ss. 1, Article 7, 9 (2000), https://www.repository.law.indiana.edu/cgi/viewcontent.cgi?article=1262&context=fclj
[68] Biden v. Knight First Amendment Inst. at Columbia Univ., 141 S. Ct. 1220 (2021) (Thomas J. concurring opinion)
[69] Office of Ohio Attorney General Dave Yost, AG Yost Files Landmark Lawsuit to Declare Google a Public Utility (2021), 3, https://www.ohioattorneygeneral.gov/Media/News-Releases/June-2021/AG-Yost-Files-Landmark-Lawsuit-to-Declare-Google-a
[70] Office of Ohio Attorney General Dave Yost, AG Yost Files Landmark Lawsuit to Declare Google a Public Utility (2021), 12, https://www.ohioattorneygeneral.gov/Media/News-Releases/June-2021/AG-Yost-Files-Landmark-Lawsuit-to-Declare-Google-a
[71] Robert H. Bork, The Antitrust Paradox: A Policy At War With Itself 397 (2021)
[72] Office of Texas Attorney General Ken Paxton, Paxton Files Third Amendment in Antitrust Lawsuit Against Google (2021) https://www.texasattorneygeneral.gov/news/releases/paxton-files-third-amendment-antitrust-lawsuit-against-google
[73] U.S. v. Microsoft Corp., 253 F.3d 34 (D.C. Cir. 2001)
[74] Epic Games v. Apple Inc., 493 F. Supp. 3d 817 (N.D. Cal. 2020) (Rule 52 Order After Trial On the Merits), https://www.google.com/url?q=https://cand.uscourts.gov/wp-content/uploads/cases-of-interest/epic-games-v-apple/Epic-v.-Apple-20-cv-05640-YGR-Dkt-812-Order.pdf&sa=D&source=docs&ust=1651705919027984&usg=AOvVaw16H-IcATaxXVGPw8G_-DWh
[75] Epic Games v. Apple Inc., 493 F. Supp. 3d 817 (N.D. Cal. 2020) (Rule 52 Order After Trial On the Merits, 135)
[76] Epic Games v. Apple Inc., 493 F. Supp. 3d 817 (N.D. Cal. 2020) (Rule 52 Order After Trial On the Merits, 144)
[77] Epic Games v. Apple Inc., 493 F. Supp. 3d 817 (N.D. Cal. 2020) (Rule 52 Order After Trial On the Merits, 150)
[78] Epic Games v. Apple Inc., 493 F. Supp. 3d 817 (N.D. Cal. 2020) (Rule 52 Order After Trial On the Merits., 152)
[79] Epic Games v. Apple Inc., 493 F. Supp. 3d 817 (N.D. Cal. 2020) (Rule 52 Order After Trial On the Merits, 160)
[80] Epic Games v. Apple Inc., 493 F. Supp. 3d 817 (N.D. Cal. 2020) (Rule 52 Order After Trial On the Merits, 165)
[81] Federal Trade Commission v. Facebook, Inc., Case No.: 1:20-cv-03590-JEB (D.C. Cir. 2021) (First Amended Complaint for Injunctive and Other Equitable Belief), https://www.ftc.gov/system/files/documents/cases/ecf_75-1_ftc_v_facebook_public_redacted_fac.pdf
[82] Timothy B. Lee, ARS Technica, FRC urges courts not to dismiss Facebook antitrust case (2021),https://www.ftc.gov/system/files/documents/cases/ecf_75-1_ftc_v_facebook_public_redacted_fac.pdf
[83] Chris O’Brien, Facebook’s alleged use of apis to crush competition is a warning to other data companies VentureBeat (2020), https://venturebeat.com/2020/12/11/facebooks-alleged-use-of-apis-to-crush-competition-is-a-warning-to-other-data-companies/