Personal Property Servitudes on the Internet of Things
MULLIGAN C.M., Personal Property Servitudes on the Internet of Things, Georgia Law Review, Forthcoming Brooklyn Law School, Legal Studies Paper No. 400, 14.07.2014
|Abstract|| Software and internet connections once were confined to multi-purpose
computers housed in rectangular boxes. No longer. Now, small appliances such as thermostats, watches, jewelry, and eyewear are being made available with networking capability. These networked objects make up the growing Internet of Things — pieces of personal property that run software and connect to the global Internet. These products are typically governed by terms of service or end-user license agreements that create restrictions on how products can be used or transferred — restrictions which would be unenforceable if the inside of the product consisted of gears rather than processing chips. This Article explores the differences between digital and analog goods by returning to a nearly century-old question from property law that has remained largely unanswered: why are complex and nonpossessory personal property interests disfavored, while comparative flexibility is permitted in real property? The Article concludes that greater flexibility in property interests is most beneficial when property is distinct, valuable, and rarely encountered. In contrast, greater standardization is appropriate when property is fungible, lacks value, and is casually or frequently interacted with. By establishing a new framework for deciding when property interests should be flexible and when they should be standardized, this Article not only answers a longstanding question from property law, but renders the answer highly relevant to some of the most interesting and novel types of property we encounter today.
|Topics||Intellectual Property, Personality, Property, Technology, DRM, Transparency|
Internet of Things can be defined as "pieces of personal property that run software and connect to the global Internet" (p. 3).
The purchasing of IoT products is usually accompanied by ToS (Terms of Service) or EULAs, which restrict customer's ability to use and/or transfer the product. These licenses would be forbidden - under common law - if the product didn't run copyrighted software: in fact, "personal property traditionally cannot be subjected to usage restrictions or servitudes" (p. 4). However, because of the presence of the copyrighted software, manufacturers can control software's (and therefore also product's) usage even after the sale: judges have indeed recognized the possibility of usage restrictions on software-embedded goods (pp. 3-4).
E.g. (pp. 4-5):
- Canon has been allowed to prevent digital camera owners from lending their cameras to others;
- Google has been allowed to forbid resale of Google Glass; moreover, the contractual terms concerning the Google Glass "attempt to both limit how transfers of the device can happen,and to bind future owners of the device to the terms" (p. 32);
- Nest has been allowed to require its thermostat to be used only for personal, non-commercial purposes.
This ability of manufacturers of software-embedded goods to restrict usage and transfer of the goods commercialized makes possible for them to unbundle features of the products and to implement price discrimination (p. 5). For what concerns price discrimination, think about the example of the license for Microsoft Office: Home and Student Edition nominally forbids houseguests from sitting down at one's computer and using Word or Excel (p. 50).
On the other hand, this idiosyncratic licensing ability may create economic and social harms, because the attributes of property rights become to varied, and therefore information costs (associated with investigating the scope of property rights: p. 12) accrue (p. 6). Information costs may be both discovery costs (= acquire the relevant information) and processing costs (= understand the information acquired) (p. 14).
In order to understand this statement about digital property, let's think about traditional property: restrictions, servitudes on chattels, on personal property, are really limited, while they are not on land, on real property (p. 8).
In fact, optimal property rights require a degree of flexibility that allows people to achieve their goals, but that doesn't make the transaction costs of property use and transfer too high (pp. 12-13). And it is undoubtful that variety in property rights "increases the cost of investigating and understanding the scope of one's property interests" (p. 12). Now, we have to remember that "one goal of property law is to promote efficient uses of property, and legal rules that encourage inefficient expenditures are against that goal" (p. 21). Numerus clausus of property rights tries to provide such limited degree of flexibility (p. 13). Some people may argue that recording systems or other forms of notice may obviate the need for standardization, but that's not true: "the costs of understanding a property right still rise as the terms of the records get more lengthy, complicated and difficult to understand" (p. 13). This argument may help us to sustain also that - in case of purchase of IoT products a non domino (because the terms and conditions accessible by anyone on the Internet say that title remains to the provider) by a second acquirer - the possesso in buona fede vale titolo principle (see my notes in Essay: A Transactional View of Property Rights) may apply, because of the costs of understanding the property rights in a world where forms of notice exist, but they are lengthy, complicated and difficult to understand: good faith of the second acquirer would therefore be possible, and this may moreover have the positive effect of encouraging the providers to be more clear and simple in the redaction and placement of their terms and conditions of use: put simply, to implement more transparency.
If more restrictions are allowed on real property than on personal property, this means that information costs to understand the scope of personal property rights rise more quickly than real property ones when standardisation decreases (p. 13).
This is due to several reasons:
- Absolute information costs. Chattels are different from land because of their small size, their mobility and their fungibility (p. 15): even if not all chattels possess these features, most chattels possess the three qualities, and nearly all possess at least two. And each of these qualities raise the information costs of property usage (p. 16): indeed, because of fungibility, if different property rights exist on different pieces of the same type, it would be difficult to identify to which piece what property rights apply; and this problem is compounded by the small size and the mobility of chattels; we need unique identifiers (think about the computer market) in order to overcome this problem (pp. 17-19).
- Relative information costs. Normally, chattels have a value that is lower than the value of a piece of land; this makes the information costs of learning about the property interests in a chattel "frequently higher than the value of the chattel and any loss from incorrect usage " (p. 21). Moreover, personal properties' value normally declines as they age (p. 22): this makes information costs greater than the value of the chattel, and this situation may discourage the recycling or repurposing of an otherwise obsolete chattel (pp. 22-23). This may obstruct and harm sustainability attitudes.
- Aggregate information costs. Real property is normally expensive, and it is rarely purchased and sold, and therefore it is reasonable to expect people to spend time and money to investigate the property interests on it. On the contrary, people interact with chattels everyday (p. 23): they may want to sell or otherwise transfer them, they may want to purchase them, and they may want to use them in many ways (alter or destroy them, for example). See The Zero Marginal Cost Society, pp. 74,75.
It is true that we may have chattels that are easy to identify, valuable and with which people rarely interact: think about cars. In those cases, a greater variety of property interests may be allowed, because information costs wouldn't be too high and therefore wouldn't make the use of property inefficient (p. 25). Therefore, the distinction made by property law between land and chattels is not explained by something exceptional about land particularly; the numerus clausus principle could have been developed in other ways; but the separation chosen between chattels and land sounds reasonable, because it permits beneficial flexibility while keeping information costs low (p. 26).
Let's come back to IP-embedded goods.
Traditionally, these goods "have functioned under rules similar to chattel property" (p. 27): the first sale doctrine (which prevents manufacturers from "placing restrictions on how a good may be used after its initial distribution": therefore, "it becomes inexpensive, from an information cost perspective, for an owner to allow it to be repurposed or transferred to someone else": p. 27; however, we have to remember that, if the possibility of a new transfer is ensured by this doctrine, the repurposing or, more generally, the use of the thing is not considered by this doctrine: we should therefore wonder if i. restrictions of use established by the sales contract may apply to the customer or not; are they compatible with the attributes of property?; ii. these restrictions, if valid for the first customer, can apply also to subsequent owners: are they only contractual clauses subjected to the privity of contract - which would be justified because of the numerus clausus of the property rights -, or do they restrict the object of the sales contract, and therefore apply also to subsequent owners, because of the principle nemo dat quod non habet?) applied.
However, court's treatment of software has evolved: it allows "downstream restrictions on how software can be used and when it can be transferred" (p. 28). This evolution is motivated by the fact that software usage requires to make a temporary copy of it, and copyright grant to the copyright owner the exclusive right of reproduction (pp. 28-29). The same happens for the others digital copyrighted works. It is true that the copyright statute grants owners of copyrighted software the right to make copies that are an essential step to using the program, but most copyright owners license their software instead of selling them, which makes the exception inapplicable, because the customer is not the owner (pp. 29-30). We have to highlight, however, that in the EU legal framework the exception to the exclusive right of reproduction applies not only to owner, but to every "lawful acquirer" (art. 5(1) DIRECTIVE 2009/24/EC on the legal protection of computer programs), which must be interpreted as every "lawful user" (Report from the Commission to the Council, the European Parliament and the Economic and Social Committee on the implementation and effects of Directive 91/250/EEC on the legal protection computer programs).
But IP-embedded goods are similar to traditional personal property ("Pieces of personal property that incorporate an intellectual-property-protected element tend to possess the qualities of property best suited to standardization."), so this ability of the provider to restrict the property interests of usage and transfer on them should be regarded with scepticism (p. 30).
Moreover, with the Internet of Things, processing capacities (and therefore software) are more and more embedded in traditional objects, and it becomes therefore uneasy to "distinguish the act of using software from using other objects", and therefore the information costs concerning use and transfer of software-embedded goods is no more segregated from other types of personal property (p. 31). It is true that frequently we have separated contract terms for the object and the embedded software, but sometimes the language used in those terms "blends any distinction between the copyrighted software and the physical device" (p. 32).
Given the fact that today every object can potentially run software, "the amount of resources a person must expend to learn how to appropriately use the devices in their possession will increase, whether the objects in fact run software or not" (p. 33). In fact, when the variety of property rights increases, the information costs increases also, and not only for the good bit by the idiosyncratic restrictions, but for all the goods, because potential customers have to verify if restrictions apply or not apply to them. So, "software embedded goods [...] subject to conditional sales [...] [raise] the information costs associated with using not just these software-embedded [...] goods, but the costs of interacting with all chattels which may or not [...] contain software"" (p. 33). "As licenses on chattels become more common, people will come to expect their presence and be more likely to expend resources learning if a license exists and understanding its scope" (p. 33). See The Zero Marginal Cost Society, pp. 74,75.
One solution could be to bring software-embedded products back to the world of non-digital chattels and of their really limited flexibility in posing restrictions to transfer and usage (p. 34). Some authors (Perzanowsky and Schultz) suggest that section 106(3) of the copyright statute - concerning the distribution right - opposes sale to more temporary transfers, so - if the "license" gives perpetual possession without ongoing payment, this relationship should be regarded as a sale; in the grey-area cases, courts will be able to set appropriate distinctions (pp. 34-35). Unfortunately in the EU context, the DIRECTIVE 2009/24/EC on the legal protection of computer programs, art. 4(1)(c), deals with " any form of distribution to the public, including the rental, of the original computer program or of copies thereof", so it doesn't opposes sale and temporary right to use as the sole possible and alternatives categories of distribution acts: however, art. 4(2), concerning the exhaustion principle, opposes sale and rental (only the sale produces exhaustion, and anyway further rental requires the authorization of the copyright holder); this opposition exists also in the CJEU's jurisprudence: case C-456/06, Peek & Cloppenburg, 2008, holds that exhaustion doesn't apply in case of rental, but only in case of transfer of ownership (see also § 56 of Opinion of Advocate General Bot delivered on 24 April 2012. UsedSoft GmbH v. Oracle International Corp.). Moreover, the Oracle v. UsedSoft case states that lump-sum + perpetual possession constitute a sale; however, this statement is given dealing with the exhaustion principle: therefore, it considers only contractual limitations on transfer, not further limitations on use (also § 56 of Opinion of Advocate General Bot delivered on 24 April 2012. UsedSoft GmbH v. Oracle International Corp. deals with this summa divisio only with regard to the exhaustion principle). For the American doctrine cited, on the contrary, not only the first sale doctrine, but the entire distribution principle are interested by the opposition sale/temporary right to use, and therefore by a broader and less formal definition of "sale": therefore, if a "license" is qualified as a sale, it should be a "sale" not only under a re-transfer perspective, but to all interests and purposes, which means that on the software-embedded goods any restriction not only on transfer, but also on usage/enjoyment can't exist. Mulligan states that - in the IoT context - this solution (of qualifying as a sale the contract formally identified as a "license", but which established a perpetual possession in exchange for a lump-sum) could particularly appropriate - because "[c]opyrighted software is intricately tied to the corresponding device": the embedded software will be probably with no value without the particular good in which it is embedded, so the risks for the copyright owner to be harmed by user's behaviour are minor (p. 35).
"The combined package of a smart toaster's software and chassis constitutes a rivalrous, scarce resource, just like any other piece of personal property" (p. 35). "And because the software in inextricably connected to a particular, specialized device, one can see how a system governing "smart devices" will tend to work similarly to how personal property transfers have historically functioned"" (p. 41): in fact, the price is unique, and the percentage paid for software and for hardware respectively is not specified (we should verify if the same happens for the general-purpose computers + operating system embedded in them; and, even if the distinction or specification isn't made, we have to remember that the Italian judges - in the Hewlett-Packard v. Pieraccioli case - have anyway stated that if the customer doesn’t want the embedded software, he can obtain back the part of the price relative to it: may the same reasoning be applied with reference to an IoT good + service contractual relationship? The price is for the entire product, but if the provider ceases to provide me with the service, I can claim for a restitution of a part of the sum paid, the one which may be referred to the service). In the IoT context, the software (a good whose copies are digital by nature) becomes a piece that has the same characteristics of analogical items, because it is incorporated in a particular analogical good, without which the software has no reason to exist (in the past, on the contrary, software could exist - in the sense that they could be useful - independently from the tangible support in which they were contained and/or run: a support was necessary, but it could be a support different from the one in which they were originally commercialized; thus, the risks for the copyright owner of being harmed in remuneration because of further reproductions by the customers were higher). Therefore, the teleological arguments that in the past have justified the introduction of the EULA tool (i.e. the easiness of reproduction by the possessor of a single copy) may fall down, because - even if the easiness may remain - the reproduction act would be nearly useless. But pay attention to the fact that today IoT products are for the most proprietary devices, walled gardens on which limitations of hardware, software and network are imposed; however, theoretically they could be general-purposes devices; in the future,we might have general-purpose IoT products; for example, we already have apps downloadable on some IoT devices (even if they are not essential to product usability); in the future, we may have different operating systems among which we may choose. For what concerns the apps, we have problems concerning the ability of re-transfer (through transfer of the device and through online transfer); for what concerns the operating systems, we'll have tying issues. We have also to reflect about the fact that probably, in the IoT context, even if it is unlikely that the embedded software would be reproduced in order to be utilized by other people who don't have paid the copyright owner, it may be reproduced in order to make the good able to interact with a service different from the one provided by the product provider (or a partner of him); and this may harm the provider, who may become unable to receive the data collected by the item sold (perhaps for a reduced price) and therefore to retrieve the value that this data would have brought. However, DIRECTIVE 2009/24/EC on the legal protection of computer programs states - at article 6 - that an exception (mandatory) to the exclusive right granted by copyright is decompilation to allow interoperability. So, this exception makes anyway useless every limitation contractually established by the provider in order to prevent this kind of behaviour, and moreover, the three-step test (remembered also by art. 6(3) of the Software Directive) takes care of the copyright holder, impeding those decompilation acts which may lead to an unreasonable prejudice of the copyright holder's legitimate interests.
Mulligan presents also an interesting explanation of why, in the US courts, licensing is more easily recognised in the software field than in other categories of copyrighted works: the difference between the UMG v. Augusto case and the Vernor v. Autodesk one is that in the first the recipients did nothing to assent to the license, while in the second they did (p. 39). This is a liberal approach. However, we have to wonder when it is possible to recognize that "consent" is present: is the power of choice between purchasing and not-purchasing sufficient? moreover, at least in the Italian legal framework, the terms should be accessible before the conclusion of the contract; instead, in many cases you buy the good in the shop, and at home you can find in the Internet the Terms of Service: it is true that potentially you can consult them before the purchasing, but the information costs are relevant, because for every object you should look for the legal Items on the Internet, and they are not always easy to be found; moreover, the assent to the terms usually takes place with the access sand use of the services and/or of the device, and not with the purchase of the device; therefore, the alternatives, depending on the content of the terms, are two: 1. the good becomes property of the customer, because also the Terms don't state that the title remains in the hands of the provider, and therefore servitudes/restrictions should not exist; 2.the terms say that ownership remains in the hands of the provider: but the access and use by the customer comes after the purchasing, and - if it is the access and use which implies the consent - would that mean that the customer acquires the ownership on the good when he pays for it, and then it loses it, transferring it to the provider - when he uses it?
It is true, however, that in an IoT world, there are technical measures that could be employed in order to reduce the information costs of the flexibility in property rights: products commercialized could be equipped with technologies able to lower and absorb those costs (RFID, that enables unique identification, or DRM, which enable the good to refuse to function in a disallowed way...) (pp. 45-46). Think about the Xbox 360 Kinect: it is equipped with a camera and , if more people than the number for which the film has been licensed are present in the room, remedial actions may be taken (p. 47). This technological reduction of information costs has negative consequences: "greater surveillance of everyday life and home life, of how and where one uses objects, as well as who uses them" (p. 45) (think about the Xbox 360 Kinect: it is equipped with a camera and, if more people than the number for which the film has been licensed are present in the room, remedial actions may be taken: p. 47).; undermined unanticipated, necessary uses of property (p. 48) (think about the cars equipped with engine shut-off devices, which stop cars from running when payments are late: p. 47). Those are the social costs of flexibility (and of the subsequent technological enforcement of the restrictions following it) in IoT products' property rights.
Let's develop this second issue: "under the common law, one could invoke "necessity" as a defense to claims of trespass or conversion" (p. 48). The same happens in civil law: as CHRISTIE G.C., The Defense of Necessity Considered from the Legal and Moral Points of View, Duke Law Journal, V. 48, i. 5, March 1999, p. 989, note 77, says:"For example, section 904 of the German Civil Code denies the owner of property the right “to prohibit the interference [with his rights of ownership] if the interference is necessary for the avoidance of a present danger and the damage threatened is disproportionally great compared to the damage caused” to him as the owner of the property destroyed. Section 904 BGB, translated in THE GERMAN CIVIL CODE 169 (Simon L. Goren trans., 1994). Section 904 also declares that “[t]he owner [of the property destroyed] may demand compensation for the loss suffered by him.” Id. [...] Article 122-7 of the French Penal Code provides that a person who, faced with a present or imminent danger to himself or another or to property, performs an act necessary for the safety of a person or of property is not subject to criminal responsibility unless there is a disproportionality between the means employed and the seriousness of the threat. See CODE PÉNAL [C. PÉN] art. 122-7 (Fr.) (author’s translation)." Mulligan explains the purpose of the necessity defense: "there are circumstances where it is preferable, or even morally requisite, to break a law in order to avoid a harmful result" (p. 49).
There may also be circumstances in which there is no "necessity" to violate property, but anyway the breach could be efficient, because the "actual damages to the owner are virtually non-existent, but the benefits to the borrower may be enormous": it is the contract-law notion of efficient breach (p. 50). According to Mulligan, the choice of how to use a chattel should remain with its possessor, instead of a licensor, because "the current possessor will have far better information about whatever situation she is than the license drafter did" (pp. 50-51). Moreover, the licensor "lacks the incentives to try to reach an optimal result" (p. 51): in fact, some tempering effects may be exercised by the market on these limitations only if they are frequently cause of frustration, but "so long as the chattel doesn't stop working often, there likely won’t be enough market pressure to alter how a particular device is permitted to function" (pp. 51-52).
For what concerns surveillance, and therefore privacy issues, they may exist even without flexibility in property rights; however, "suggesting massive private surveillance as an alternative to property standardization" (i.e., as a mean to lower the information costs associated with flexible property rights) "will likely encourage further privacy intrusions" (pp. 53-54).
Finally, DRM and other technological measures have limits: if a password is required to use a product, password may be revealed by the customer to a friend in order to use it (p.56). They may also be ineffective in the sense that they prevent authorized behaviours, and therefore create incentives to circumvent DRM or to find unauthorized and infringing copies not covered by DRM (pp. 54-55).