NOTE-1026: Perzanowski and Schultz — The End of Ownership

Contribution

It brings together in one book the authors’ law-review work on “digital exhaustion” and the “Buy Now” survey (Acknowledgments). It tells one story across media, software, devices and patents: rights that buyers held by default as owners of copies are moving to rights holders through licences, DRM and the courts. Its question is put on p. 20: “do we own our digital goods?” It answers that, for the most part, we do not, and that we should. It adds a legal test for when a buyer is an owner (pp. 76–77), a test for the scope of digital exhaustion (pp. 179–183), and survey evidence that buyers are misled (ch. 5).

Key insight

Copyright protects buyers through rules about copies. Exhaustion gives the owner of a lawful copy the rights to resell, lend, display and, for software, run and back it up (pp. 25–26, 63–64). It rests on a sharp line between the work and the copy (pp. 36–38). Digital distribution breaks the line in two ways. A transfer makes a new copy, so the first sale doctrine does not cover it (ReDigi, pp. 40–41). A cloud or streamed purchase may leave no copy the buyer holds (pp. 43–48). Retailers then fill the gap with licences that deny a sale and with DRM that enforces the licence. The authors’ fix is to stop counting copies: “Copies today are cheap, disposable things.” (p. 46). What should matter is “who has the right to use and transfer it” (p. 188).

Assumptions

• US law, early 2016. The cases and statutes are American: the Copyright Act §§ 109, 117 and 1201, the Lanham and FTC Acts, and the patent cases to Lexmark v. Impression Products (Fed. Cir., February 2016). The EU appears only as a foil (pp. 175–176, 184–185).
• The RAM-copy rule as live risk, not settled law. “Some courts” treat a copy in memory as a copy (p. 46, p. 63). The authors rely on the risk rather than the rule, and cite Perzanowski’s own article against it (ch. 4 n. 26). Cablevision (1.2 seconds is not a copy) is set against MAI v. Peak (p. 47).
• Property is made by law. Property rights are “a product of law” (p. 192), defined by who wins “a contest for control over the behavior of others” (p. 23). The authors favour the bundle-of-rights view but want a short, standard menu of transactions for consumer goods (pp. 21–22).
• Information costs justify standard forms. The case against bespoke licences rests on the numerus clausus argument (Merrill and Smith) that idiosyncratic rights impose costs on third parties (pp. 8–9, 17). It is argued, not measured.
• Consumer expectations are evidence, not a ground of right. The survey shows what buyers believe. The authors say plainly that this does not settle what rights they should have: “Rights—property, constitutional, or human—need a firmer foundation.” (p. 98).

Key results

The results are arguments from cases, statutes and one survey. In the book’s order:

• The stakes (ch. 1). Amazon’s remote deletion of 1984, Linn Nygaard’s wiped Kindle and Scholastic’s Storia show purchases revoked (pp. 1–3). Licensing costs buyers resale, lending, permanence and privacy (pp. 6–7). It also imposes costs on others: information costs, lost preservation, less user innovation and higher switching costs (pp. 8–10). The authors write that “the most fundamental value at stake in the choice between ownership and licensing is autonomy” (p. 10). They argue that ownership is also the carrot that makes lawful copies worth buying (p. 12).
• Property and exhaustion (ch. 2). Personal property admits no servitudes. Restrictions do not “run with the tuxedo” (pp. 16–17). Rowling’s goblins, for whom the maker remains the owner, are the imagined alternative (p. 17). Exhaustion runs from nineteenth-century cases (Clemens v. Estes on Huckleberry Finn, bound fire-damaged pages, Kipling’s loose sheets) through Bobbs-Merrill v. Straus (1908) to the 1909 Act and Kirtsaeng (2013) (pp. 26–29). Rights holders have resisted it in every medium. GameStop says 70 percent of trade-in value goes to new games (p. 31). The authors favour exhaustion “not because property is a label with talismanic properties, but because it is smart policy” (p. 33).
• Copies, clouds and streams (ch. 3). The work/copy distinction runs from Pope v. Curl (1741) to § 202 of the 1976 Act (p. 37). ReDigi held that a forward-and-delete transfer still reproduces the work (p. 41). Section 117 extended exhaustion to software copies, but no equivalent exists for other media (p. 42). A cloud copy might be like a cinema’s film reel, a library book or a safe-deposit heirloom, and only the first two fit copyright’s vocabulary (p. 44). White-Smith v. Apollo (1908), where piano rolls were not copies, shows the law failing to find a copy before (p. 47). On ownership and identity, buying an album “identifies you, both to others and to yourself” as a fan, and the endowment effect holds for digital goods (p. 53).
• The fine print (ch. 4). Licences are “privately made law” (p. 58). The iTunes terms run past 19,000 words (p. 59). Perhaps one software shopper in a thousand looks at them (p. 60), so licences are a market for lemons. Record labels call iTunes downloads sales when royalties are due (p. 62). Since ProCD, courts enforce terms buyers see only after paying (pp. 67–71). On the permission view the GPL takes, a licence cannot take back what the Copyright Act gives owners (pp. 71–74). UMG v. Augusto looked at the facts of distribution and Vernor v. Autodesk at the licence’s words, on near-identical terms (pp. 75–76). The authors’ test for a sale is duration, one-time payment and the public characterization of the deal (pp. 76–77). Price discrimination, which licences enable, aims to take the whole consumer surplus (p. 80).
• The “Buy Now” survey (ch. 5). It sampled nearly 1,300 likely buyers on a mock store, MediaShop (p. 90). Only 14 of the 956 shown a digital product opened the terms (p. 92). Nearly nine in ten thought they could keep the purchase as long as they liked. More than 40 percent thought they could lend or give it away, and nearly 30 percent that they could leave it in a will (pp. 93–94). On resale the figures were 12 percent for books and 17 for music and movies (p. 95). This was the only right not well above the legal threshold, though the authors argue it still suffices. Preferences for these rights were stable across digital and physical formats. Respondents would pay on average $3.82moretolendand$3.24 to resell (pp. 96–97). On a seven-point accuracy score, the short notice averaged 3.0, “Buy Now” 2.45, “License Now” 2.27 and physical goods 4.7 (pp. 100–101).
• Libraries (ch. 6). Publishers impose “artificial friction”: HarperCollins’s 26-loan cap, delays, caps and per-loan pricing (pp. 106–109). OverDrive serves over 90 percent of the library ebook market (p. 108). A Dudamel recording sold only on iTunes could not be lent at all (p. 109). Ownership underwrote preservation, patron privacy and the HathiTrust fair-use win (pp. 111–118). It also underwrote Friends of the Library book sales (pp. 118–119).
• DRM (ch. 7). DRM enforces terms no legislature wrote: “We’ve replaced courts and due process with code and license terms.” (p. 123). It fails by Godwin’s “smart cow” problem (p. 125). After Sony, Vault v. Quaid and Sega v. Accolade (pp. 126–129), the DMCA’s § 1201 banned circumvention and circumvention tools (pp. 130–131). In Corley, “Breaking DRM was unlawful, regardless of the reason.” (p. 132). DRM locked labels into iTunes until they dropped it (pp. 133–134). The Sony BMG rootkit compromised about two million computers (pp. 135–137). Chamberlain (Federal Circuit) and Lexmark (Sixth Circuit) refused to extend § 1201 to garage door openers and printer cartridges (pp. 137–138).
• The Internet of Things (ch. 8). Nest bricked the Revolv hub despite a “lifetime subscription” (pp. 140–141). Apple told the Copyright Office that iPhone users are licensees of iOS (p. 143). John Deere said farmers have “an implied license for the life of the vehicle to operate the vehicle” (p. 145). Further cases are GM, Mercedes, Ford’s speed limiter, Ferrari’s resale option, Keurig 2.0 and Hello Barbie, whose recordings ToyTalk may use (pp. 146–151). Pacemaker and insulin-pump owners may not read their own devices (pp. 152–154).
• Patents (ch. 9). Patent exhaustion runs from Bloomer v. McQuewan (1852, “the ordinary pursuits of life”) through Adams v. Burke, Keeler, Straus v. Victor and Motion Picture Patents (1917) (pp. 156–159). Edison’s agents built the first electric chairs from second-hand Westinghouse dynamos because the patents were exhausted (pp. 157–158). The Federal Circuit’s Mallinckrodt (1992) allowed conditional sales, and Jazz Photo (2001) national exhaustion. Quanta (2008) and Kirtsaeng undercut both, but the en banc Lexmark v. Impression Products (2016) reaffirmed them (pp. 160–168).
• Reforms (ch. 10). The “sharing economy” is mostly for-profit temporary access, and real sharing presupposes ownership: “You usually can’t lend something that you don’t own.” (p. 170). The proposals are (pp. 174–185):
  1. An FTC inquiry into “Buy Now” and short notices.
  2. Unconscionability, the EU Unfair Terms Directive and copyright preemption against form contracts.
  3. Owners exempted from § 1201, which could still enforce rentals and subscriptions.
  4. A return to patent exhaustion.
  5. A digital exhaustion test. A one-time payment for open-ended access makes the buyer an owner. The use is then weighed by whether the owner gave up access, whether it denies the rights holder a fair return, and whether the owner altered the expression. A perfectly efficient lending pool shows why the second factor is needed (p. 182). UsedSoft (CJEU) and the Dutch Tom Kabinet rulings show resale working in Europe (pp. 184–185). Mulligan’s proposal of “taking the copy out of copyright altogether” is called promising but too broad (p. 185). Forward-and-delete schemes and the Amazon and Apple resale patents are rejected as private, platform-bound and DRM-dependent (pp. 185–188). A blockchain ledger is offered as a title registry for “bitproperty” (pp. 188–191).

Claims

id	claim	strength	support
C1	Copyright’s exhaustion rules are tied to ownership of a copy, so digital transfers that make new copies, and purchases that leave no copy with the buyer, fall outside them	strong (as US law in 2016)	ReDigi, Cablevision, MAI, Vernor; §§ 109, 117 (chs. 3–4)
C2	A licence cannot turn what is in substance a sale into something less; ownership should follow duration, payment structure and public characterization	moderate (normative)	Augusto, Krause, DSC; the permission theory of licences (Newman, Moglen); the UCC on reservation of title (ch. 4 n. 44)
C3	”Buy Now” leads a legally significant share of buyers to believe they get ownership rights, and the belief is material	moderate	the authors’ own online survey of ~1,300 on a mock store; legal thresholds from case law (pp. 89–90)
C4	Bespoke licence terms impose information costs on all buyers, an externality that consumer choice cannot correct	weak to moderate	economic argument (Coase, Pigou, Merrill and Smith); no measurement
C5	Anti-circumvention law has failed at its stated goals and harmed security, competition and repair	moderate	cases of cracked DRM, the Sony rootkit, iTunes lock-in, Keurig (ch. 7); one study on sales (ch. 10 n. 16)
C6	Exhaustion and resale markets help rights holders on balance (used games feed new sales; home video outgrew the box office)	weak	GameStop’s own figure (p. 31), industry history; asserted against rights holders’ own claims
C7	Library ebook licensing reduces access, preservation and patron privacy relative to owned print	moderate	HarperCollins, OverDrive, Dudamel and Adobe examples; the legal analysis of the third-party doctrine (ch. 6)
C8	A multifactor digital exhaustion test, applied by courts like fair use, can protect owners without undue harm to incentives	weak	proposal by analogy to § 107; “undue harm” is left undefined (p. 182)

Concepts

• Exhaustion. The rule that a rights holder gives up control of an item once it is sold or given away by its authority. In copyright the main form is the first sale doctrine (§ 109); others are display (§ 109(c)) and software copies (§ 117).
• Copy/work distinction. The work is the author’s intangible expression; the copy is the material object that fixes it. Copyright’s exhaustion rules give buyers the copy and leave the work with the author (pp. 36–38).
• Licence as contract / licence as permission. On the contract view, a licence binds by assent, which courts now find in clicks and inaction. On the permission view, a licence is a property owner’s grant, needs no assent, and works only within the licensor’s existing rights (pp. 64–74).
• Essential step copy. A copy needed to run software, allowed to the owner of a copy by § 117(a)(1) (p. 63).
• Artificial friction. Limits imposed on ebook lending to mimic the costs of physical books (pp. 106–107).
• Smart cow problem. Mike Godwin’s name for the fact that one person who breaks a lock frees everyone else (p. 125).
• Short notice. A plain, iconographic statement of the rights a buyer gets and does not get (pp. 99–100).
• Temporary-access model. Rentals, subscriptions and most of the “sharing economy”. The authors accept these when they are clearly labelled (pp. 169–173).
• Bitproperty. Fairfield’s term for property interests decoupled from any tangible object (p. 191).

Connections

It draws on the authors’ law-review articles (“Digital Exhaustion”, “Copyright Exhaustion and the Personal Use Dilemma”, “Legislating Digital Exhaustion”, “Reconciling Personal and Intellectual Property”) and on Perzanowski and Hoofnagle’s “What We Buy When We Buy Now” (Acknowledgments). For property theory it uses Merrill and Smith on the numerus clausus, Van Houweling and Mulligan on servitudes, and Coase and Pigou on externalities. For contracts it uses Radin’s Boilerplate, Kim’s Wrap Contracts and Lemley. For DRM and lock-in it quotes Cory Doctorow’s Information Doesn’t Want to Be Free three times (ch. 7 nn. 4, 19, 27). It uses von Hippel’s Democratizing Innovation for user innovation and Fairfield for bitproperty. It does not cite Lessig, Benkler or Jenkins. Its “sharing economy” (ch. 10) is Uber and Airbnb, not Benkler’s or Lessig’s.

Bearing on the record

• Lessig’s copy trigger (LIT-1298, NOTE-1023). This is the closest contact, and the two books agree on the mechanism. Lessig: the 1909 Act made copyright regulate “copies”, every digital use makes one, so the law now reaches ordinary use (pp. 97–103 of Remix). Perzanowski and Schultz show the same mechanism from the buyer’s side. The rules that protect owners are rules about copies. A digital transfer is a new copy (ReDigi), and use may be one (the RAM-copy cases). So the buyer’s property shrinks just as the rights holder’s reach grows. The remedies converge too. Lessig’s fourth reform is to regulate uses, not the copy. These authors want courts to track “who has the right to use and transfer” a work rather than count copies (p. 188), and they note Mulligan’s proposal to drop the reproduction right (p. 185). They differ on what is at stake. Lessig’s concern is read/write culture, making new work from old. Theirs is what Lessig would call read-only use: keeping, lending, reselling and repairing. They exclude remix from exhaustion in so many words. Exhaustion “isn’t the right framework to analyze the legality of remixing” (p. 183). Read together, the two books show the copy trigger reaching both ends of the reader’s relation to a work: the right to remake it and the right to keep it. The record has no claim yet that states the copy trigger as its own. If one is written, this book supplies the case law Lessig’s sketch lacks.
• Doctorow on platforms (LIT-731, LIT-033; THEORY-150, THEORY-470). The connection is real but partial. THEORY-150 lists lock-in through DRM among what sustains platform decay, and THEORY-470 names DMCA § 1201 as what disabled the interoperability constraint. This book supports both steps with cases. DRM raised the cost of leaving iTunes until the labels abandoned it, quoting Doctorow’s “roach motel” (pp. 133–134). § 1201 makes circumvention unlawful whether or not a use infringes (Corley, p. 132). Keurig and Lexmark used locks to tie consumables, as in Doctorow’s HP ink example (NOTE-041). The book also qualifies Doctorow. Courts have sometimes required a nexus to infringement before § 1201 bites (Chamberlain, Federal Circuit; Lexmark, Sixth Circuit; pp. 137–138), so the law is less uniformly hostile to interoperability than NOTE-041’s talk implies. The book says nothing about the three-stage cycle of THEORY-150, or about competition, regulation and worker power as THEORY-470’s other constraints. It supports the lock-in mechanism, not the decay schema.
• CLAIM-332 and QUESTION-119 (distributed uptake). The book has incidental cases of a loose population changing a firm’s conduct. Microsoft dropped its Xbox One resale limits after “persistent and overwhelmingly negative feedback from customers” (p. 32). Keurig partly retreated after Facebook posts, Amazon reviews and a 10 percent stock drop (p. 150). A petition of over 100,000 led to the unlocking law (p. 143). They fit CLAIM-332’s pattern of separately motivated acts without a collective judgment. But they are chosen anecdotes, and the book itself says market pressure “is often only partly effective in protecting consumer interests” (p. 150). That is QUESTION-119’s caution. They are not evidence for the claim.
• CLAIM-002 (artifact identity). The book’s copy/work distinction is the law’s version of CLAIM-002’s first relation, identity of the artifact. The book’s point is that in digital media the artifact no longer has a stable identity: copies “flit into and out of existence” (p. 46). Even “the definition of the work” becomes hard for copyright (p. 181). That bears on how the record uses artifact identity for digital texts. It says nothing about retelling or translation.
• No bearing on CASE-100 or CLAIM-034. The book is about consumption and transfer, not making new work, and excludes remix. It has no bearing on QUESTION-002, QUESTION-023, CLAIM-210 or CLAIM-220 either. Its remarks that owned things “help define who we are” (p. 53) and on Hegel (p. 22) are passing, and bear on nothing in the personhood line.
• No instruction for machine-learning practice, and not held by the anthology.

Limitations

• Advocacy by participants. The authors argue one side. One of them represented one of the groups that found the Sony rootkit (p. 136), and one represented the bnetd developers (ch. 7 n. 24). One author consulted for the plaintiffs in the iPod antitrust case (ch. 7 n. 18). Rights holders’ arguments are reported, but mostly to be answered.
• The survey is the authors’ own and limited. It used a mock store and stated preferences and willingness to pay. Women were overrepresented (ch. 5 n. 49). Its full report was forthcoming at the time (ch. 5 n. 34). That stated beliefs show deception is a legal argument no court has tested. The book concedes that no consumer is likely to bring the case (p. 174).
• The economic claims are argued, not measured. The externality of information costs, the benefit of resale to rights holders, and the harm of price discrimination rest on theory and industry figures. The “undue harm” factor of the proposed exhaustion test is left open (p. 182).
• Dated by events. These are from general knowledge and not checked in this session. The Supreme Court reversed the Federal Circuit in Impression Products v. Lexmark (2017), applying exhaustion to post-sale restrictions and to foreign sales, as ch. 9 hoped. The Second Circuit affirmed ReDigi (2018). The CJEU held in Tom Kabinet (2019) that ebook resale is not covered by exhaustion. That cuts against the book’s European example (pp. 184–185). The blockchain chapter (pp. 188–191) predates the NFT episode.
• Small errors. “Capital Records” for Capitol (p. 41). “digital copies are more than occasionally more expensive than their digital counterparts” (p. 134) should read analog. Smaller slips occur in the notes.
• The digital bundle is left uncalibrated. The last page of the Conclusion (p. 193) grants that the ideal digital property bundle will differ from the analog one and that calibrating it will not be easy, and asks only for meaningful choice among rentals, subscriptions and ownership. Which rights the bundle holds is not specified.

Open questions

• Does resale or lending of digital goods reduce rights holders’ revenue enough to matter, and under what lending rules? The book’s second exhaustion factor needs an answer it does not give.
• Would short notices change behaviour at purchase, or only answers in a survey?
• If copyright moved from counting copies to tracking rights of use, as both this book and Lessig propose, what would count as a use? Would private reading or running then fall inside copyright, as Mulligan’s display-and-performance proposal suggests (p. 185)?
• For the record: does the loss of a stable artifact in digital media change what CLAIM-002’s first relation can do for digital texts?