Harvard Law Review Harvard Law Review Harvard Law Review

Harvard Law Review Forum

One-Legged Contracting

Rabbi Hillel was once challenged by a gentile to teach “the whole Torah” while the challenger stood on one leg. 1×1. Philologos, The Rest of “The Rest Is Commentary,” Forward (Sept. 24, 2008), https://forward.com/culture/14250/the-rest-of-the-rest-is-commentary-02564 [https://perma.cc/F9K4-MFBP]. The sage responded with his version of the Golden Rule: “That which is hateful to you, do not unto another: This is the whole Torah. The rest is commentary.”2×2. Id. One of us implemented a version of this one-legged constraint by trying to summarize in a video everything one needs to know about retail investing while standing on one leg. Ian Ayres, Concise Advice for Investing While Standing on One Leg, Forbes (Jan. 30, 2016), https://www.forbes.com/sites/whynot/2016/01/30/hillel-investment-advice/#2e0e477a6b55 [https://perma.cc/N98U-VUCM]. The story exemplifies both how succinctly a sagacious interlocutor can summarize a vast tract, and the extreme impatience of many listeners.

Professors Robin Kar and Margaret Radin, in their thought-provoking article Pseudo-Contract and Shared Meaning Analysis,3×3. Robin Bradley Kar & Margaret Jane Radin, Pseudo-Contract and Shared Meaning Analysis, 132 Harv. L. Rev. 1135 (2019). have proposed that courts ignore vast portions of contractual writings: the unread standard terms ubiquitous in modern consumer and other transactions.4×4. Id. at 1139–40 (citing Yannis Bakos, Florencia Marotta-Wurgler & David R. Trossen, Does Anyone Read the Fine Print? Consumer Attention to Standard-Form Contracts, 43 J. Legal Stud. 1 (2014)). Before enforcing such boilerplate text, a court should “imagine that all of the written and digital text exchanged during contract formation is converted into oral form and takes place in a face-to-face conversation between the relevant parties.”5×5. Id. at 1167. It should then ask: “Could this boilerplate text have plausibly contributed to an oral conversation that contributes terms to a contract consistent with the presupposition that both parties were observing the cooperative norms that govern language use to form a contract?”6×6. Id. If the answer is “No,” the boilerplate “should not be enforced.”7×7. Id. Kar and Radin’s standard places particular emphasis on succinctness. Plausibly contributing to an oral conversation where the drafter is observing the cooperative conversational norms requires the drafter “to say neither too much nor too little” given the shared purpose of the conversation.8×8. Id. at 1151.

Kar and Radin are asking important questions, and there is much to admire in their analysis.9×9. See, e.g., id. at 1182–92 (criticizing Alan Schwartz and Robert Scott’s argument for textualism and identifying examples in which context evidence might improve the accuracy of interpretation in contracts between sophisticated parties); id. at 1196–203 (collecting doctrinal tools courts use to prevent adhesive contracts from overriding the parties’ actual communications, and emphasizing the potential power of Restatement (Second) of Contracts § 211(3)). The past century has seen radical changes in the technologies for entering into contracts, including consumer contracts. If in the early twentieth century many worried about preprinted forms businesspeople never read,10×10. See, e.g., Nathan Isaacs, The Standardizing of Contracts, 27 Yale L.J. 34 (1917); Karl N. Llewellyn, Book Review, 52 Harv. L. Rev. 700 (1939). today we have shrinkwrap, clickwrap, and browsewrap. There is a good case that courts, legislatures, and regulators have not kept up — that the old rules of contract formation, construction, and enforcement are ill suited to these new technologies.

But we worry that Kar and Radin’s proposed solution would, in many contexts, lead to a kind of one-legged contracting. We do not understand Rabbi Hillel to be saying that the commentary is unimportant or might be dispensed with. Indeed, the full quotation ends with an imperative: “The rest is commentary — [and now] go study.”11×11. Philologos, supra note 1. But as modern buyers, we are often like the impatient gentile demanding to learn everything in just a few moments. Whether we are standing at a rental car counter, enrolling to buy 99-cent songs from iTunes, or even closing on a mortgage, the vast majority of us want to contract quickly. Because nondrafters’ conversational tolerance for details is extremely limited, in many contexts Kar and Radin’s enforcement standard would severely limit the length of contracts without regard to the substance of their terms. Kar and Radin appear to view this as a good thing. We are not so sure.

Part I of this Response criticizes as arbitrary and essentializing Kar and Radin’s insistence of shared meaning as the core of contracting. Part II argues that even if shared meaning were the sine qua non of contracting, their proposal fails to achieve it because it does not assure that the terms would be “cooperatively communicated.” Part III argues that the proposed enforcement standard would, in practice, severely limit freedom of contract and likely reduce consumer welfare.

I. Essentializing Shared Meaning

Kar and Radin criticize those who seek “to assimilate all boilerplate text to ‘contract’ so long as it is delivered with actual or merely constructive ‘notice’ to a party who agrees to a more basic transaction.”12×12. Kar & Radin, supra note 3, at 1139. The authors appropriately single out Judge Frank Easterbrook’s opinion in Hill v. Gateway 200013×13. 105 F.3d 1147 (7th Cir. 1997). as a classic statement of this “assimilationist” approach: “A contract need not be read to be effective; people who accept take the risk that the unread terms may in retrospect prove unwelcome.”14×14. Kar & Radin, supra note 3, at 1143 n.19 (quoting Hill, 105 F.3d at 1148).

Kar and Radin maintain that “assimilationist approaches have not yet offered a fully workable or coherent alternative to centering contract interpretation on the common meaning of the parties.”15×15. Id. at 1156; see also id. at 1143 (“Nor have assimilationists advanced any adequate normative theory for why contract formation by means of assumption of risk by noncognizant parties ought to replace traditional contract, with its basis in actual agreement with shared meaning.”). Their core criticism is that assimilationists fail to limit enforcement to terms cooperatively communicated to the buyers.16×16. Id. at 1139–40 (“Assimilationists assume that all boilerplate text serves the same essentially contractual function, and they do not recognize the critical difference between terms that parties cooperatively communicate and agree to during contract formation and the increasingly copious boilerplate text that is merely tacked onto that agreement but never read.”). “The premises of freedom of contract — and also freedom from contract — suppose parties with equal capacities to define and enter into only those terms that both agree offer expected gains for each.”17×17. Id. at 1161 (emphasis added). The article abstract states that “actual agreement” is “required by core contract law principles.”18×18. Id. at 1137; see also id. at 1139 (“[A]n actual agreement with common meaning is central to the normative justification of contract.”). If assent to communicated terms is the bedrock of valid contractual enforcement, assimilationist approaches must fail, as they call for enforcement of predictably unread, and hence uncommunicated and non-agreed-upon, terms.

A distinctive feature of contractual obligations is that they are both voluntary and chosen. Contractual obligations are voluntary in the sense that the parties must agree to them, unlike most duties found in tort or criminal law. Contractual obligations are chosen in the sense that the parties get to decide for themselves the content of the obligations. This distinguishes contractual duties from those that attach, say, to a political office, military service, marriage, or a fiduciary — duties voluntarily acquired but difficult or impossible to alter.19×19. See Gregory Klass, What If Fiduciary Obligations Are Like Contractual Ones?, in Contract, Status, and Fiduciary Law 93, 101–08 (Paul B. Miller & Andrew S. Gold eds., 2016).

It does not follow, however, that “[t]he premises of freedom of contract” require that the parties actively choose, or even comprehend, all terms to which they agree. If a contract is the whole of the legal relationship between the parties that results from formation, then contracts regularly include terms that the parties have not communicated and to which they have not “actually agreed.” Default terms apply precisely when the parties have not reached an agreement to the contrary. Mandatory terms apply even in the face of the parties’ contrary agreement. And many rules of contract construction — contra proferentem, interpretations favoring the public interest, formalities like “F.O.B.” or “as is” — look beyond what was actually communicated or the parties’ actual agreement. Such nonchosen terms do not threaten freedom of contract. Although the parties have not agreed to them individually, they have assented to the transaction as a whole. That assent, which renders contractual obligations voluntary if not all individually chosen, can do a lot of normative work.20×20. For two very different examples, see Randy E. Barnett, Consenting to Form Contracts, 71 Fordham L. Rev. 627, 634–36 (2002) (argument based on an autonomy-based consent theory); and Lucian A. Bebchuk & Richard A. Posner, One-Sided Contracts in Competitive Consumer Markets, 104 Mich. L. Rev. 827, 833–34 (2006) (arguing that apparently one-sided unread terms can in fact enhance consumer welfare).

This is not to deny the important differences between legislatively or judicially created default and mandatory terms and standard terms that one party drafts and gives to the other on a take-it-or-leave-it basis. There are reasons to scrutinize the latter that do not apply to the former. Our point is merely that the core principles of contract law do not obviously require actual agreement to or understanding of all terms. The claim that choice of all terms is essential to contracting is not true to the phenomena.

Kar and Radin level a separate, and to our minds even less persuasive, criticism of the so-called assimilationist approach: that it is “[l]inguistically [i]ndeterminate.”21×21. Kar & Radin, supra note 3, at 1160. To explain why, they point to several examples of unread terms in the iTunes online terms and conditions, such as: “If you see content submitted to the Apple Music Service that does not comply with these Guidelines, [then] use the Report a Concern feature.”22×22. Id. at 1162. The authors treat the discovery of such provisions as strong evidence of the failure of the assimilationist approach — claiming that Judge Easterbrook’s position “start[s] to look strange” in light of them.23×23. Id. Kar and Radin’s argument is subtle, but we take the claim to be twofold. First, they emphasize that instructions like the above have the same if-then structure as sentences that add terms, such as, “If you paint my house, then I will pay you $1000.”24×24. Id. at 1163. An exclusive focus on semantic meaning — “sentence meaning” — does not therefore differentiate between sentences that add terms and those that are mere instructions.25×25. Id. In order to mark the difference, one must recur to linguistic norms of cooperation, the very norms that, according to Kar and Radin, assimilationists themselves ignore or reject. Assimilationists have therefore failed to explain “how courts are supposed to translate boilerplate if-then statements like these into ‘contract meanings.’”26×26. Id. Second, the examples belie the assimilationist claim that “all boilerplate text found under the contemporary label of ‘terms and conditions’ is being assigned a ‘contract meaning’ and treated by everyone as adding ‘terms’ to a ‘contract’ once a consumer clicks ‘I accept.’”27×27. Id. at 1165.

We find ourselves baffled by both claims. As Kar and Radin observe in a footnote, when using “a fully ‘textualist’ or ‘four corners’ approach to interpretation, courts often do and must implicitly rely on conversational implicatures to identify even the ‘plain meaning’ of text.”28×28. Id. at 1183 n.139. Even the most textualist of courts regularly intone, a contractual writing should be “read as a whole to determine its purpose and intent.”29×29. W.W.W. Assocs., Inc. v. Giancontieri, 566 N.E.2d 639, 642 (N.Y. 1990); see also, e.g., Empire Props. Corp. v. Mfrs. Trust Co., 43 N.E.2d 25, 28 (N.Y. 1942) (“The meaning of a writing may be distorted where undue force is given to single words or phrases. We read the writing as a whole. We seek to give to each clause its intended purpose in the promotion of the primary and dominant purpose of the contract.”). It strikes us as fairly straightforward to differentiate between the instructions and terms in the iTunes terms and conditions, as illustrated by Kar and Radin’s own puzzlement at the idea that obvious instructions might be treated as terms.30×30. See Kar & Radin, supra note 3, at 1164 (suggesting that trying to give contractual meaning to instructional provisions would lead to “ludicrous” and “absurd[]” results); id. at 1208 (noting with respect to provision, “You can disable an app’s access [to Apple Music] on your iOS device in Settings,” and arguing “[n]o one would think that Apple had actually breached the resulting contract if its developers were, for instance, to come up with an easier and more user-friendly way to disable an app’s access without going to Settings and could therefore eliminate the Settings bar altogether in a useful software update”). Doing so does not require applying the shared meaning analysis Kar and Radin advocate — focusing only on language that could have been communicated in an oral conversation. Meaning happens in nonconversational contexts too. Identifying it requires simply making sense of words in the context in which they appear, reading a document as a whole, and applying the principle of charity.31×31. See Donald Davidson, Radical Interpretation, in Inquiries Into Truth and Interpretation: Philosophical Essays 125 (2001).

Nor do we understand the grounds on which Kar and Radin attribute to their opponents a “blanket assumption that all boilerplate text conveyed during contract formation must seek to add terms to a contract.”32×32. Kar & Radin, supra note 3, at 1209; see also id. at 1208 (“[P]lacing noncontractual boilerplate text under a misleading label like ‘terms and conditions’ can only cause obfuscation.”). We know of no theorist who makes such a claim. Nor can we think of any reason why someone who advocates enforcing unread, or even unreadable terms, must assume that every word in a document labeled “terms and conditions” should designate a contract term.

Perhaps Kar and Radin are concerned about the label “terms and conditions.” They write: “Apple is offering a mere instruction for use even though the instruction is presented under the misleading label of ‘terms and conditions.’”33×33. Id. at 1165. We grant that Apple’s lawyers could have chosen a more descriptive title. But it is odd for Kar and Radin to complain that it is misleading. As noted above, they too find it easy to differentiate between instructions and terms. Nor do they point to any evidence of consumer confusion. And we think there are good reasons to mix terms and instructions in a single document. In a world where few nondrafters bother reading standard terms ex ante, it becomes all the more natural to include instructions to guide the subset of those who consult them ex post. We do not expect new car buyers to read all of their owner’s manual before using the car. Terms and conditions, like owner’s manuals, are often best consulted when particular problems arise. Including instructions in them strikes us as not only unproblematic, but beneficial.

II. Failing to Assure Shared Meaning

We believe the core risk of adhesive contracts, including consumer contracts, is drafter overreach. Knowing the nondrafting party is unlikely to read, the drafter is tempted to include terms to which the nondrafting party would object if they were brought to her attention — terms that are unfair or inefficient. In addition to harming the nondrafting party, the widespread use of such terms can also have negative social consequences — a case Radin has made forcefully elsewhere.34×34. See, e.g., Margaret Jane Radin, Boilerplate: The Fine Print, Vanishing Rights, and the Rule of Law 33–51 (2013). One finds in the law two nonexclusive approaches to addressing this risk. The first, procedural approach attempts to secure consumer comprehension of terms, on the theory that consumers can be trusted to recognize unfair terms if they only know that they are there. The second, substantive approach relies on scrutiny of the content of the terms. That scrutiny might employ ex ante regulatory mechanisms such as black or grey list, or it might rely on ex post review against a standard such as substantive unconscionability.

Kar and Radin themselves emphasize the risk of unfairness to nondrafting parties. To the extent that their proposed test seeks to address it, it would appear to represent a modified version of the procedural approach. Their core premise is that “parties with equal capacities [will] define and enter into only those terms that both agree offer expected gains for each.”35×35. Kar & Radin, supra note 3, at 1161. Unlike other procedural proposals, however, Kar and Radin do not advocate securing nondrafter comprehension of important boilerplate terms. They view that project as hopeless.36×36. See id. at 1171–72 (citing Omri Ben-Shahar & Carl E. Schneider, More than You Wanted to Know: The Failure of Mandated Disclosure 169 (2014)). For a trenchant critique of Ben-Shahar and Schneider’s use of evidence and mode of argument, see Richard Craswell, Static Versus Dynamic Disclosures, and How Not to Judge Their Success or Failure, 88 Wash. L. Rev. 333 (2013). Instead, they would radically cut back the effective boilerplate language to that which could have been communicated in a conversation.37×37. See Kar & Radin, supra note 3, at 1172, 1175.

But the proposed “face-to-face conversation” test does not assure that enforcement is limited to terms the nondrafting party actually understands.38×38. See id. at 1176–77. Kar and Radin recommend that with respect to disputed boilerplate text, courts ask the following question:

Could this boilerplate text have plausibly contributed to an oral conversation that contributes terms to a contract consistent with the presupposition that both parties were observing the cooperative norms that govern language use to form a contract?

Any boilerplate text that meets this test falls within the correct boundary of parties’ actual agreement for a contract, and courts can rely on their ordinary linguistic intuitions to interpret the contract meaning of the text. Otherwise the boilerplate text is mere pseudo-contract, which does not contribute to the common meaning of the parties and should not be enforced.39×39. Id. at 1167 (footnote omitted).

Although the above test uses the words “actual agreement,” it appears not to require that a term actually be communicated to the nondrafting party — there is no expectation that the nondrafting party actually read the term. All that is necessary is that it be conveyed in “a sufficiently cooperative manner” such that parties’ common meaning is created.40×40. Id. at 1155. Nor do Kar and Radin present any evidence that nondrafting parties are more likely to read and comprehend boilerplate text that passes their test.41×41. For a dramatic illustration of how difficult it can be to secure consumer comprehension, see Omri Ben-Shahar & Adam Chilton, Simplification of Privacy Disclosures: An Experimental Test, 45 J. Legal Stud. 541 (2016) (finding that warning boxes highlighting terms many consumers might find important had little effect on consumer comprehension, decision making, or understanding of their legal rights). Although Kar and Radin criticize those who would enforce boilerplate terms, “so long as [the text] is delivered with actual or merely constructive ‘notice,’”42×42. Kar & Radin, supra note 3, at 1139. their preferred rule in fact merely sets out a different, higher standard for what constitutes constructive notice. The test does not eliminate “the so-called ‘duty-to-read’” doctrine, which holds nondrafters to unread terms. It merely limits which unread terms will be enforced.43×43. Id. at 1182 (“If courts are interested in discerning the common meaning of the parties and correctly identifying the scope of their actual agreements, then the so-called ‘duty to read’ should be limited to text that was cooperatively communicated.”).

We agree that the correct test cannot be actual agreement. Limiting enforcement to terms that have been actually communicated or agreed upon (objectively considered) would wreak havoc on contract doctrine. As we noted above, many rules of contract construction — default terms, mandatory terms, contra proferentem, and so forth — generate terms in the absence of, and sometimes despite, the parties’ actual agreement to them.

But this causes us to wonder about the purposes and implications of the proposed test. Whereas actual comprehension by a critical mass of nondrafters can, at least in theory, serve to discipline drafters,44×44. For a classic statement of this claim, see Alan Schwartz & Louis L. Wilde, Intervening in Markets on the Basis of Imperfect Information: A Legal and Economic Analysis, 127 U. Pa. L. Rev. 630, 643–46 (1979). potential comprehension does not. Without an argument that nondrafters are more likely to read terms that pass their proposed test, there is little reason to think that Kar and Radin’s proposal will address the risks adhesive contracts pose.

If the test is meant to serve a purpose other than securing fair or efficient terms — if it comes from, say, a more abstract commitment to what freedom of contract requires — we wonder about its implications. What, for example, does it say about how the law establishes and communicates default terms?45×45. Kar and Radin acknowledge the existence of default rules and other noninterpretive rules of construction. See Kar & Radin, supra note 3, at 1138 n.2. They do not explain, however, why those rules should be exempt from their proposed test. Does section 2-314 of the Uniform Commercial Code sufficiently convey the implied warranty of merchantability to make it part of the shared meaning of a contract for the sale of goods?46×46. U.C.C. § 2-314 (Am. Law. Inst. & Unif. Law Comm’n 2002). And according to the proposed test, the question is not whether individual sections of Article Two might have been effectively communicated in conversation, but whether all the enforceable terms could have been. Clearly the default terms contained in Article Two, plus the raft of judicial decisions comprising the common law of contracts, fail that test. But if unread and uncooperatively conveyed provisions of the Code and relevant caselaw can become part of the parties’ shared meaning, why can substantively reasonable boilerplate terms not also become part of the parties’ shared meaning?

To repeat, we do not mean to assimilate terms drafted by one party and given to the other on a take-it-or-leave-it basis to legislatively or judicially established defaults. But the conceptual apparatus Kar and Radin use has implications they do not address. Or to put the same point a different way, their conceptual apparatus does not capture what is truly worrisome about contracts of adhesion — the risks of opportunism and unfairness that come with terms drafted by one party and given to the other on a take-it-or-leave-it basis. We would do better by focusing on those risks, rather than the form in which nonsalient terms are provided.

III. Crypto-Mandatory Rules and Consumer Welfare

Kar and Radin claim that their approach “is not a form of paternalism or market regulation.”47×47. Kar & Radin, supra note 3, at 1172. We disagree. In fact, their approach would prevent buyers from taking a reasonable risk on unread boilerplate. The authors analogously argue that “utilizing shared meaning analysis would not interfere with freedom of contract.”48×48. Id. at 1178. This claim borders on the disingenuous. Although shared meaning analysis does not impose specific, substantive mandatory rules, its procedural mandate has important substantive consequences. It is therefore best understood as a covert or crypto-mandatory rule.

First, although there are circumstances in which parties might reasonably want to take on the risk of unread terms, Kar and Radin’s proposal would deny them the ability to do so. Technological changes have radically expanded the use of standard terms. But technology has also expanded the possibility of reputational disciplining of sellers’ boilerplate. A “bug me not” consumer renting from Avis or licensing from Apple might rationally prefer a streamlined contracting process, knowing that social media would likely uncover and bring to bear public pressure on any untoward terms. Assuming the risk is all the more reasonable when a buyer takes into account the protection afforded by the doctrine of unconscionability and other doctrines that limit the enforceability of substantively unreasonable standard terms — doctrines that Kar and Radin themselves helpfully collect and summarize.49×49. See id. at 1197–202.

Second, Kar and Radin give insufficient attention to the possibility that terms that do not pass their proposed test can increase aggregate consumer welfare. Even if a subset of contractors ends up being bound by terms to which they would never have agreed, it is still possible that enforcing unread terms produces transaction cost savings that on net increase nondrafter welfare. Judge Easterbrook in Hill memorably captured just this idea:

Cashiers cannot be expected to read legal documents to customers before ringing up sales. If the staff at the other end of the phone for direct-sales operations such as Gateway’s had to read the four-page statement of terms before taking the buyer’s credit card number, the droning voice would anesthetize rather than enlighten many potential buyers. Others would hang up in a rage over the waste of their time. . . . Customers as a group are better off when vendors skip costly and ineffectual steps such as telephonic recitation, and use instead a simple approve-or-return device.50×50. Hill v. Gateway 2000, Inc., 105 F.3d 1147, 1149 (7th Cir. 1996).

Standard terms save drafters money, which in competitive markets can benefit nondrafters in the form of lower prices. A nondrafter might find it in her interest to roll the dice on being among a majority of those that receive that benefit. Again, Kar and Radin’s approach would prevent her from doing so.

Third, the proposed test could prevent terms that directly benefit consumers. In practice, the “face-to-face conversation” test would work as a complexity-limiting device. Drafters attempting to contract around default rights and duties would be severely limited in the number of provisions that could be altered — perhaps to two or three. The shared meaning approach would therefore result in a regime in which almost all default terms become mandatory. One suspects that this complexity-limiting feature is an intended consequence of Kar and Radin’s enterprise — a feature, not a bug. But the proposed test would also deny enforcement to boilerplate terms that are unexpectedly generous relative to the legal default or to the buyer’s beliefs.51×51. See Ian Ayres & Alan Schwartz, The No-Reading Problem in Consumer Contract Law, 66 Stan. L. Rev. 545, 578, 583 n.94, 598 n.140, 605 (2014) (discussing examples where a majority of consumers expected less favorable terms than those included in the actual agreement). And sometimes complexity is necessary to secure real benefits for the nondrafting party.

Consider the margin rate currently offered by Interactive Brokers, an electronic trading platform.52×52. One of us has extolled the use of Interactive Brokers. See Ian Ayres & Barry Nalebuff, Lifecycle Investing: A New, Safe, and Audacious Way to Improve the Performance of Your Retirement Portfolio (2010). Whereas at this writing Fidelity and Schwab were charging 9.32% interest to customers buying stock on margin, Interactive Brokers was charging an interest rate of just 3.90%.53×53. See Online Brokers, Interactive Brokers, https://www.interactivebrokers.com/en/index.php?f=1340 [https://perma.cc/7E38-VLSX]. One reason that its margin rate is so much lower is that the company has contractually dispensed with the tradition of “margin calls” after prespecified drops in the margined stock’s price (where brokers would telephonically call their margin customers and give them the option of adding additional funds to their account to avoid involuntary liquidation of the margined position).54×54. See Interactive Brokers, Disclosure of Risks of Margin Trading 1 (Mar. 10, 2015), https://gdcdyn.interactivebrokers.com/Universal/servlet/Registration.formSampleView?file=registration_1/margin_trading_risk_disclosure.html [https://perma.cc/B2SB-FQEL] (“You should understand that pursuant to the IB Margin Agreement, IB generally will not issue margin calls, that IB will not credit your account to meet intraday margin deficiencies, and that IB generally will liquidate positions in your account in order to satisfy margin requirements without prior notice to you and without an opportunity for you to choose the positions to be liquidated or the timing or order of liquidation.”). Under Kar and Radin’s proposal, a seller who wanted to engage in such contractual innovation would face substantial risk: a court might later conclude that the language it used to describe this term could not have plausibly contributed to an oral conversation under cooperative norms.

Kar and Radin ominously claim that “market forces have begun to interact with assimilationist legal doctrine to create powerful incentives for businesses systematically to mislead consumers.”55×55. Kar & Radin, supra note 3, at 1196. To be sure, boilerplate at times hurts nondrafters. The low-salience, “shrouded” pricing of back-end fees is a vivid example.56×56. See Xavier Gabaix & David Laibson, Shrouded Attributes, Consumer Myopia, and Information Suppression in Competitive Markets, 121 Q.J. Econ. 505, 506–08 (2006); see also Sendhil Mullainathan et al., The Market for Financial Advice: An Audit Study 11, 14, 17 (Nat’l Bureau of Econ. Research, Working Paper No. 17929, 2012), https://ssrn.com/abstract=2028263 [https://perma.cc/K7UV-X9JR]. But the implied claim that assimilationist enforcement hurts nondrafters more than it helps them is unsubstantiated. The enforcement of boilerplate that meets the ex post fairness test of unconscionability and associated doctrines, as well as the disciplinary reputational tests of social media, leaves considerable room for contractual creativity. That is not to say that the current protections are enough. Where there are systematic problems, for example, ex ante regulation of terms should perhaps step in, as the Consumer Financial Protection Bureau attempted to do with respect to class arbitration waivers.57×57. Arbitration Agreements, 81 Fed. Reg. 32,830 (May 24, 2016) (codified at 12 C.F.R. pt. 1040 (2017)), rendered ineffective by Pub. L. No. 115-74, 131 Stat. 1243 (2017). A contractual regime that imposed the further limits of shared meaning analysis, however, would be a world in which innovative, contractually structured products like Interactive Brokers, Uber, and Airbnb would be less likely to exist.

Conclusion

Almost half a century ago, Professor Arthur Leff suggested that we think of consumer contracts not as agreements, but as things that businesses market to the public, which like an automobile or other technological good, a person might purchase without fully understanding how it works.58×58. Arthur Allen Leff, Contract as Thing, 19 Am. U. L. Rev. 131 (1970). Like Kar and Radin, Leff argued that it was a category mistake to lump such consumer contracts of adhesion together with fully negotiated contracts. But Leff’s solution was very different. Rather than excising what was not part of the actual agreement, Leff recommended regulating consumer contracts like any other consumer product.59×59. “When things are too dangerous or too worthless the government does directly intervene. There is no reason why that intervention should not take place as directly with respect to contracty things.” Id. at 155. This is not to say that we agree with all of Leff’s regulatory approaches, such as the use of warning labels, id. at 153–55. Our understanding of which regulatory techniques work has grown since 1970. Viewed from this perspective, Kar and Radin’s proposal looks like a form of Luddism. Rather than regulating complex adhesive contracts for fairness, safety, and social benefit, Kar and Radin would lop off that which is too complex. The equivalent in the product safety realm would be requiring auto manufacturers to go back to the horse and buggy.

In the end, Kar and Radin’s analysis seems too divorced from real-world consequences. They choose formal abstractions — and at times moving poetry60×60. See, e.g., Kar & Radin, supra note 3, at 1208 (“The tree of contract is getting lost in an expanding forest of pseudo-contract, and the forest is being mistaken for the tree.”). — over practical policy analysis. The decisions concerning whether and when to enforce boilerplate text can impact social welfare. We too worry that buyers would not have purchased many items if they had known the details of unread terms. But we also wonder whether different enforcement rules might more effectively harness competition to produce better offers. What we want are rules that, on net, both increase consumer autonomy and enhance the gains of trade. We believe Kar and Radin’s shared-meaning approach is too indirect a route to get us there.


* William K. Townsend Professor, Yale Law School.

**Agnes N. Williams Research Professor, Professor of Law, Georgetown University Law Center. We are indebted to Joe Burson for his excellent research assistance on this Response.