Shadow Agreements: How Project Nessie Evades Consumer Law Through Code

In September 2023, the Federal Trade Commission (FTC), 18 states, and Puerto Rico sued Amazon for using anticompetitive and unfair retail technology [1]. The technology is Amazon's pricing algorithm, Project Nessie, which was used from 2014 through 2019 to predict where Amazon could raise prices and have its competitors raise theirs as well. By allowing Amazon to predict its competitors' behaviors, this technology allows Amazon to raise its prices without fear of being undercut. The FTC argued that Project Nessie illegally inflated prices, violating the FTC and the Sherman Act, along with several state laws [2].

Amazon has motioned to dismiss the case on nearly every ground possible. District Judge John H. Chun let two things stick: on March 20, 2025, he dismissed consumer protection claims brought by New Jersey and Pennsylvania, ruling that Project Nessie did not violate either state’s laws on unfair commercial practices or deception.[3]

New Jersey argued that Nessie violated the New Jersey Consumer Fraud Act (NJCFA) on two counts: unconscionable commercial practice and a knowing concealment of material facts. The NJCFA defines unconscionable commercial practice as commercial conduct that is fraudulent, deceptive, or involves the knowing suppression of material facts with the intent to mislead [4]. Judicial interpretation has refined the concept. In Fenwick v. Kay Am. Jeep, Inc., New Jersey state court defined unconscionable conduct as behavior so unfair that it “shocks the conscience” and is conducted with the intent to mislead consumers [5]. This judicial standard requires unconscionability to show an intentional and egregious departure from acceptable competitive practices. New Jersey argued that Amazon’s use of Nessie to coordinate price increases across competitors deceived consumers into paying inflated prices. Judge Chun, however, rejected this argument, finding that Amazon’s Nessie constituted market-driven and legal competitive behavior rather than intentionally deceptive conduct [6].  

Regarding the concealment claim, both New Jersey and Pennsylvania argued that Amazon had a duty to disclose Nessie’s existence and function. Judge Chun rejected each state's claim, reasoning that because Nessie was based on market forces and consumers could shop elsewhere, Amazon had no duty to disclose under NJCFA and Pennsylvania’s Unfair Trade Practices and Consumer Protection Law (PUTPCPL) [7].

Judge Chun relied on legal precedent–including Yingst v. Novartis AG [8] and BCR Carpentry LLC v. FCA [9]–to grant Amazon’s motion to dismiss New Jersey and Pennsylvania’s claims under the NJCFA and PUTPCPL. In Yingst, Novartis, a Swiss pharmaceutical company, upcharged for identical products, while in BCR Carpentry LLC, Chrysler imposed inflated transportation fees. Both cases served as examples of market-driven decision-making on a much smaller scale, where market forces tightly constrained the companies involved. In contrast, Project Nessie is not just a standard market price tweak, it’s a self‑learning algorithm that observes competitors in real time [10]. Amazon's substantial market share raises questions about whether its algorithm merely responds to market forces or actively shapes the market, and "market-driven" cannot serve as a legal justification for monopolistic power when a company like Amazon controls the steering wheel. The dismissal of the New Jersey and Pennsylvania claims in the Project Nessie case exposes a clear disconnect between traditional consumer protection laws and the realities of modern, technologically advanced market practices. Moreover, the broad exemptions in the NJCFA and PUTPCPL, when applied to monopolies like Amazon, fail to protect consumers.

The NJCFA falls short because it only focuses on direct deception, missing how advanced market tactics harm consumers. In his reasoning, Judge Chun dismissed New Jersey's charges because consumers still had “[the] ability to negotiate in a meaningful fashion”[11]. Amazon, by timing price hikes when competitors follow, leaves consumers with no real option to shop around for a better deal. NJCFA's reliance on overt deception, along with exceptions for market forces, overlooks how monopolistic market power and sophisticated pricing strategies can undermine a truly free market.

Like New Jersey’s NJCFA, PUTPCPL allows for “market-driven” exemptions for powerful companies that shape its market, and relies on traditional definitions of collusion, requiring an expressed or inferred agreement [12]. Judge Chun classified Nessie as a “predictive pricing tool” rather than an agreement, which exempted Amazon from PUTPCPL's “deceptive conduct” provisions [13]. These traditional definitions ignore how Nessie acts as both a predictive pricing tool and an agreement. 

Nessie’s algorithmic price adjustments functionally invite competitors to match hikes, creating the effects of collusion without any official agreement. In the future, as the market develops even more sophisticated algorithms, this silent coordination will only become more dangerous. Once multiple algorithms can accurately predict each other’s behavior, they may independently determine that raising prices benefits them all, creating an environment similar to intentional collusion without formal agreements or human intent. 

New Jersey's NJCFA and Pennsylvania’s PUTPCPL were landmark pieces of legislation for their time, 1960 and 1968, respectively, but they are ill-equipped to protect consumers in the modern market. Collusion no longer requires backdoor in-person meetings; in fact, these meetings are far less efficient than predictive models. Project Nessie doesn't need explicit agreements with competitors because they can achieve the same result through automated decision-making. New Jersey and Pennsylvania state laws rely on human-led deception and collusion, allowing companies like Amazon to sidestep the question completely through its algorithm, which is no less harmful to consumers. For state-level protections to keep up with increasingly sophisticated pricing algorithms, consumer protection laws must recognize and regulate algorithmic collusion. Additionally, the market-driven justifications provided by these laws misunderstand how monopolistic firms like Amazon shape the market itself.

Sources:

1. Complaint at 1–2, FTC v. Amazon.com, Inc., No. 2:23-cv-01495-JHC (W.D. Wash. Nov. 2, 2023).
2.  Id. at 5.
3. FTC v. Amazon.com, Inc., No. 2:23-cv-01495-JHC, 2025 BL 92339, at 1 (W.D. Wash. Mar. 20, 2025).
4. FTC v. Amazon.com, Inc., 2025 BL 92339.
5. Fenwick v. Kay Am. Jeep, Inc., 72 N.J. 372, 371 A.2d 13 (1977).
6. FTC v. Amazon.com, Inc., 2025 BL 92339.
7. FTC v. Amazon.com, Inc., No. 2:23-cv-01495-JHC (W.D. Wash. Nov. 2, 2023).
8. Yingst v. Novartis AG, No. 2:13-cv-07919 (D.N.J. Nov. 24, 2014).
9. BCR Carpentry LLC v. FCA US LLC, No. 21-19364 (D.N.J. Oct. 23, 2024).
10. Amazon’s $1.4B price-raising ‘Project Nessie’ algorithm exposed in FTC antitrust fight, The Register (Nov. 2, 2023), https://www.theregister.com/2023/11/02/amazon_ftc_project_nessie/.
11. FTC v. Amazon.com, Inc., 2025 BL 92339.
12. 73 P.S. § 201‑3(a).
13. FTC v. Amazon.com, Inc., 2025 BL 92339.