THE GOVERNANCE OF CODE: OPEN LAND VS. UCITA LAND
Imagine two network societies. In one society, the transfer of information and use of software is governed by the various licenses used to protect open code today. For convenience, we dub this society "Open Land". The other society recently passed a law identical to the new Uniform Computer Information Transactions Act ("UCITA", passed in two U.S. states and pending in several others) to control its information and computer software. We will call this society "UCITA Land".
This paper looks at the ways in which Open Land and UCITA Land differ. Although it might initially seem that a licensing framework is trivial in terms of the actual conduct of society, we posit that such frameworks in many ways define the mode of governance of network society. The establishment of a market involves the development of a bundle of rights that both create property and define the rules under which property-based transactions might occur. In Open Land and UCITA Land, fundamentally different approaches to the establishment of those rights lead to vastly different societies.
In this work we focus on code, rather than on information goods as a class. Doing so provides for us a framework in which to explore the implications of different mechanisms of governance. We define the two worlds, looking closely at their licensing schemes, and at the governance models those schemes imply. To understand Open Land, we need to examine several common open code licenses, including the GNU Public License, the BSD license, the Mozilla license and the artistic license. Our consideration of UCITA Land requires some background on the UCITA legislation and an analysis of its effects. Before turning to the two worlds, we begin by demonstrating how licenses underpin governance in network societies.
THE ROLE OF LICENSES IN GOVERNANCE
In real space, we have seen much effort go into establishing governance mechanisms. Nations have created different law-making bodies - often known as the executive, legislature and judiciary - and worked hard to establish their roles and the relationship between them. A nation's
governance mechanisms can be expected to preserve and support its political philosophies, whether they favor absolutism, monarchy, dictatorship or democracy. But law does not spring from a vacuum. In real space, governance is created by law, but also by the social realities and common experiences that make a particular form of governance possible and acceptable. To take two oft-quoted examples, Land War I was the necessary background for the formation of the League of Nations (later the United Nations), and a troubled monarchical history in England led the United States' Founding Fathers to adopt a constitution mistrustful of concentrated power. Further, laws will not stand where the social realities of a country overwhelmingly reject them, such as in the case of revolution.
What are the equivalent mechanisms in the information society? Stallman and more recently Lessig have argued that code is the prime governance mechanism and establisher of the political, commercial and relational environment in an information society (e.g., Lessig, 1999). While the analogy has strengths, the flaw in comparing code to governance is that each person can choose his or her own code. In a Lotus Notes shop dominated by Windows with no support for Linux or Macintosh, we wrote this article on StarOffice on a RedHat Zoot box. In real space, a United States resident cannot similarly choose to bind herself by the laws of Belgium and not those of the USA. Code is not law, because code allows "opt out", while law is binding.
In our view, code can more usefully be seen as the environment that gives rise to governance, in circumstances where the relevant licensing regime operates as the law in the information society. In real space, law arises from its environment, governs that environment and binds citizens. Similarly, a licensing framework springs from the code, governs the code, and binds all network society participants. Participants can each experience code differently, and work around or opt out of some of the constraints code may create. But licenses are binding. This paper proceeds on the basis that code and licenses together establish governance in this new world, much as the environment and law establish governance in real space.
It is now appropriate to look at the licensing framework that prevails in Open Land, before considering the governance mechanisms operating there.
OPEN LAND
As the name suggests, Open Land consists of open code. Although a tremendous amount of attention has been focused on open code, there has been little corresponding discussion regarding the modes of open code, their differences, and the substance of those differences. Our discussion of Open Land attempts to develop those differences more fully.
We begin by outlining what Open Land does and does not comprise. In Open Land, all code is governed by licenses satisfying the Open Source Initiative definition for open licenses (see Perens, 1999, for the full definition). It is commonly but falsely assumed that a society like Open Land would consist of public domain code. This type of code is released to the public without any copyright (or more correctly, with an express rejection of the author's copyright). That enables users to do whatever they like with it, including re-copyrighting it as their own work. The result is that users may then convert the program into proprietary software, with major or minor modifications, and redistribute it in that form, taking it outside of Open Land. Such action will strip away the freedom originally attached to the program.
By contrast, "copyleft" code is distributed with a license that, in simple terms, uses copyright to prevent it from being made proprietary in the future. As a result, it is arguably "more free" than public domain software, because all derivations of copyleft code in Open Land will remain in that world. Richard Stallman, founder of the Free Software Foundation, coined the term copyleft, asserting that it enables the crucial freedoms that define free software to become inalienable rights (Stallman, 1999).[3]
The open code license developed by Stallman, known as the GNU General Public License (GPL), contains a strong copyleft. As some of the code in Open Land is governed by this license, not all Open Land code is public domain code, and copyright continues to have a role in Open Land. However, not all of the open code licenses include a copyleft - some, such as the BSD License, would allow code to be modified and taken outside Open Land.
It is also important to note that the code and information in Open Land are not necessarily free in terms of price. The terms "open" and "free" refer to the lack of proprietary restrictions on ideas underlying the code. "Free software" requires the freedom to use, modify, and distribute the software. "Open" means open to examination and alteration. Neither term has any bearing on the price charged for code or information. Programmers are entitled to add value to open code and then charge for that value. For example, there are fee-paying GNU-Linux distributions that include a GUI installer and thirty days of installation assistance. The alternative free distribution does not include the installer or any assistance. Further, RedHat offers a constant update line whereby patches are offered for any vulnerabilities, free, to those who have purchased the code. There is nothing in the free redistribution requirement to prevent versioning or value-added services. Such fee-paying services are consistent with the parameters of Open Land.
The Licenses
By definition, licenses play an important role in Open Land. In this section, we briefly outline the main licenses used.
The first license, the GNU General Public License (GPL) is often considered to be the archetypal free software license. As mentioned above, it contains a copyleft to protect the freedom of code down the generations. The LGPL, the lessor GPL, is similar to the GPL and was designed by GNU to cover software libraries. For this reason, it does allow protected code to be incorporated into proprietary software.
The BSD License is a non-copyleft free software license used by the various Berkeley Software Distribution (BSD) projects and by Apache. Behlendorf characterizes this license's laissez-faire approach as follows: "Here's this code, do what you like with it, we don't care, just give us credit if you try and sell it" (Behlendorf, 1999). The most common criticism of the original BSD
License relates to its requirement that attribution be made to Berkeley in all uses or advertisements of the software. Developers applying the BSD License to their own software have tended to copy this clause and change it to refer to their own company or university, meaning that in large projects many lines of advertising must be used to acknowledge all contributors. The Modified BSD License addresses this concern by deleting this requirement and eliminating references to Berkeley, while retaining all of the other characteristics of the BSD license. This modified license is compatible with the GNU GPL, but is a non-copyleft license. The Modified BSD License is the version examined in this paper.
The Mozilla Public License (MozPL) was developed to cover the public release of the Netscape Navigator code (dubbed "Mozilla") in 1998. Other free software licenses could not be used for this purpose because many of the third party components contained in the Mozilla code were covered by private licensing agreements. The Mozilla license needed to be compatible with those private licenses and also allow commercial developers to contribute code to subsequent versions of Mozilla (Hammerly et al, 1999).
The first version of the license - the Netscape Public License (NPL) -- was posted on a public newsgroup and amended in accordance with the comments received. These comments resulted in the development of a second license intended to work with the NPL, namely the MozPL. The NPL is identical to the MozPL except that it confers additional rights on Netscape. The NPL and MozPL both allow modifications to protected code to be "taken private" and not kept in the open software domain, but they are flexible in allowing the combining of protected code and proprietary code. Both the NPL and the MozPL are free software licenses, but they do not provide a strong copyleft, no doubt in part because of the desire of Netscape (now AOL) to incorporate any technology integrated with Mozilla. Netscape saw this as necessary given that the bulk of contributions to Mozilla were initially made by Netscape employees. As a result of this need by Netscape/AOL to protect its own perceived investment, these licenses are not compatible with the GPL. The licenses are fairly opaque in their drafting and would benefit from a plain English rewrite.
The Artistic License was developed for a language, not to protect a product produced in a particular language. It was created by Larry Wall for the Perl language, and has since been heavily criticized but widely adopted. The FSF comments that "We cannot say that this is a free software license because it is too vague; some passages are too clever for their own good, and their meaning is not clear" (Free Software Foundation, 2000d). The Artistic License is considered to be an open software license, but not copyleft.
We have already mentioned the Open Source Initiative definition for open licenses. Before examining the licenses in more depth, it is appropriate to consider this definition, so as to obtain a clearer picture of how code and information are dealt with in Open Land.
The Open Source Definition
Bruce Perens of the Open Source Initiative (OSI) describes the open source definition he prepared as a "bill of rights for the computer user" (Perens, 1999, p.171). A license must fall within this definition to be certified as "open source " by OSI.
According to the OSI definition, the distribution terms of an open source software product must satisfy the following conditions:
1. Free Redistribution
The license must not restrict the sale or disposal of the software, or require any fee in connection with such a sale (however, an individual distributor may charge a fee if he or she chooses). This requirement is unequivocally satisfied by all licenses except the Artistic License. Most licenses include some requirements regarding attribution and the inclusion of notices (copyright and disclaimers).
The free redistribution clause of the Artistic License (Article 1) may not apply to the code if a third party has modified it in a non-standard way, so the reach of the license may be quite limited in practice. Also, the license allows a fee to be charged for the software if it is aggregated as part of a larger software distribution (Article 5). The free redistribution condition of the OSI definition was worded specifically to include this "aggregation" scenario, even though it is not strictly within the spirit of free software. Note also that if the protected code is fully embedded in another distribution, the rights no longer appear to apply (Article 8). For these reasons, there is some doubt as to whether the Artistic License genuinely allows free redistribution, although it does appear to come within the OSI definition, in letter if not in spirit.
2. Source Code
Open code must include or allow easy and free access to its source code, and this code must be in the preferred form for programmer modifications. With the exception of the BSD License, all licenses considered here clearly satisfy this requirement. The BSD License allows redistribution in either source or binary forms - binary forms are not the preferred form for modifications.
3. Derivations
The license must expressly allow the software to be modified, and must allow any modifications or derived works to be distributed under the same license. All of the licenses satisfy this requirement. The GPL ensures that modifications are covered by the same license, which also means that GPL-licensed code cannot be mixed with non-free software -- a significant restriction on its use.
4. Integrity
The license may require any modifications to the source code to be documented and/or included in patch files. This is a concession from the derivation condition rather than a condition in its own right - it allows authors limited control over derivations in order to protect the integrity of their code.
All licenses except the BSD License require that modifications be fully documented. As this is a concession rather than a condition, there are no consequences for not including this requirement - this would not disqualify the BSD License from participation in Open Land.
5. No discrimination
The license must not discriminate against persons or groups, or against use in a specific field of endeavor. One example of discrimination, described by Perens, is a license issued by the Regents of U.C. Berkeley preventing particular software developed there from being used by the South African police (Perens, 1999). That restriction was perhaps understandable during the period of apartheid, but persists although it no longer has the same relevance. Open code proponents regard it as preferable to avoid even such well-intentioned discrimination. It is true that the non-discrimination condition might prevent the use of open source code in the development of some restricted kinds of software, such as classified software, but the open code movement considers that a necessary evil. This is exemplified by the original Java code. Although arguably superior in terms of governance, this code could not have originally been considered open code, because there were limitations on the use of Java 1.1 in safety-critical systems - not all libraries had been tested to assure reliability in multi-threaded use.
All licenses except the GPL are clearly non-discriminatory. The GPL does allow the license to restrict distribution in certain countries, if due to the intellectual property system in those countries, distribution there would infringe the GPL. Although this may be considered to be discrimination, it is better viewed as a rejection by those potential recipients of the terms of the GPL.
6. License Distribution
The rights must apply to all recipients of the software, without the need for execution of an additional license. None of the licenses requires recipients to execute an additional license.
7. Non-Specificity
The rights must not depend on the software being part of a specific distribution. All of the licenses clearly satisfy this criterion.
8. Non-Contamination
The license must not impose restrictions on other software distributed with the licensed software. Except for the GPL, the licenses unequivocally satisfy this condition. The GPL does impose restrictions on software that forms part of a modified work together with some portion of the licensed software. But it will not apply to identifiable sections that "are not derived from the Program and can be reasonably considered independent and separate works in themselves", and neither will it apply to separate works that are aggregated with licensed software on a storage or distribution medium (Article 2). Consequently, the GPL arguably satisfies this requirement. However, there is some debate on this point within the open source community and among open source scholars.
Governance of Open Land
We hope that this discussion of the various open license criteria has provided some insight into the rules governing code in Open Land, as well as demonstrating that the licenses considered above would all be useful in that world. We move now to outline the governance philosophy in Open Land, with reference to each of the OSI criteria.
1. Free redistribution
A parallel could be drawn between the free redistribution requirement in Open Land and the minimal "welfare" accommodations offered by the state for many elements of life: food, shelter, and health care. In both cases there is arguably basic functionality that should be denied to none, although the quality and associated service may be considerably less than the market usually provides.
However, this is a flawed analogy, as it compares real and information goods. Because of code's status as an information good, a better argument is that free redistribution of code is comparable to the provision of basic democratic information. Without access to code, participants do not have a voice in a network society where information resources are key. Thus this freedom is perhaps as basic as the right to vote in a traditional context, or the right to examine and comment on the law. The information embedded in code contains the keys to powerful systems of control and authority in a virtual world. Open code advocates would argue that access to that information is essential to full and free citizenship in that environment (though of course, many "regular users" would not require the full degree of information that might seem essential to a programmer).
Another approach would be to regard code as providing an understanding of how network society works, and so equipping users to exercise creativity and develop new solutions. A real space informational analog might be basic education, which provides children with the building blocks of communication and computational skills, such that they can engage with the world as independent adults. Preventing user access to code may at one level be as serious a denying real space citizen's education in reading, writing and arithmetic.
This approach suggests that freedom of code is a fundamental characteristic of open code, preventing the other characteristics of code from being denied by prohibitive pricing. Free distribution is enfranchisement and basic education, not welfare, in Open Land. Its inclusion in the Open Land charter creates an enfranchised, able populace.
2. Source availability
Source availability is described by some (e.g. Lessig, 1999) as the most fundamental of all open code requirements. The availability of code offers the potential but not the guarantee of transparency in the governance sense.
In engineering, a system is transparent when it works so well that it is invisible to the user. By contrast, in governance a system is transparent if all steps can be examined and nothing can be hidden from the examiner. Thus open code can be seen as an analogy to the various sunshine laws, which require that government advisory bodies open meetings and records. Code availability effectively creates structural transparency in Open Land.
3. Derivations
The fact that Open Land expressly allows derivations opens the way for progress. However, the fate of those derivations is a hotly contested issue. The debate echoes environmental arguments over the national heritage. Is the ideal goal of the national heritage to ensure it remains unchanged for all generations? Or is the heritage something to be modified according to maximization of profits? In the United States, this conflict has existed over physical lands through out history, as exemplified by the Homesteading Act and various laws granting mining and logging rights on national lands.
So what of the information contained in Open Land? Is it a national asset to be used only with an expectation that all advances be returned, or is it raw material ripe for harvest? In fact, either case could be true, as some code can be made proprietary and taken outside Open Land, and some (e.g. code protected by the GPL) must remain free. To the free software view the removal of code from Open Land is theft. The core value seems to be that derivations allow for the progress and development of Open Land's public commons, but that proprietary derivations are the privatization of those commons.
4. Integrity
Integrity maintains the trustworthiness of Open Land's code base. Consumer protection regulation is embodied in integrity. According to licenses with integrity requirements, any changes should be documented and redistributed with the original code. Integrity can be seen as a high level of control on altering that part of governance embodied in code, as any changes must be well documented and explained. In this way integrity provides the enforcement of claims of authenticity in Open Land.
5. Non-discrimination & non-specificity
Non-discrimination means that there is no group or application that may not use the code. This provision ensures the inclusiveness of Open Land and prevents the code from becoming involved in interline disputes. Non-discrimination allows there to be ferociously competing camps at particular level or function in open code communities (e.g., BSD vs. Linux, KDE vs. Gnome) but where those releasing software not immediately concerned with that layer or function are required to be agnostic. In effect, it prevents mandatory bundling.
Similarly, non-specificity prevents the rights attaching to Open Land code from being limited to any particular distribution. Again, it removes the temptation to discriminate and promotes an environment in Open Land of equality of opportunity and technological agnosticism.
6. Non-contamination
While non-discrimination and non-specificity prevent mandatory bundling, non-contamination encourages promiscuous bundling in Open Land. Non-contamination requires that it be possible to release the software in conjunction with other software. It enables consensus and community to form or coalesce around any subset of software without requiring that other elements be included.
Conclusions
To sum up, Open Land's licensing framework creates a governance model that enfranchises and equips participants, ensures transparency, allows for progress (and even privatization or "opt out"), protects consumers by maintaining the trustworthiness of the common code, fosters inclusivity, and allows consumer and developer freedom around code bundling. This approach is best compared with a libertarian view of governance -- the existence of a liberal set of governance rules requires continued transparency, a balance of power, and open processes. Because there can be nothing coerced from the user community through enforced bundling, or sold through guile or lack of integrity, the image of governance presented in the open code scenario is one of minimal governance and maximal individual autonomy.
Having shown that Open Land experiences a highly libertarian form of governance, in particular a government of code plus licenses without the power to enforce or obligate the user to take action, we now use the principles of code plus licensing as governance to examine UCITA Land.
UCITA LAND
As already noted, UCITA Land has just passed a law identical to the Uniform Computer Information Transactions Act (UCITA), which is now law in Maryland and Virginia and pending as a bill in several other U.S. States. UCITA has been introduced in real space at the urging of the Business Software Alliance, and over the objections of numerous consumer bodies.[4] In UCITA Land, even though contractual parties can choose not to be bound by UCITA, that law is now the default licensing framework for all software.
In December 2001, in response to the submissions of advocacy groups, the official UCITA Standby Committee recommended several legislative amendments to the National Conference of Commissioners on Uniform State Laws (UCITA Standby Committee, 2001). Where these are relevant to the discussion below, they are included as notes. However, these recommendations do not yet have the force of law, and so the concerns raised below continue to stand. The recommendations are also not reflected in the laws currently in force in Maryland and Virginia.
The introduction of this new law means that UCITA Land differs from Open Land in many respects. The Open Land licenses view source code, to one degree or another, as conversational creations - creative speech acts capable of modification and development. UCITA views code as a service to be licensed for a specific purpose and duration. The most significant differences created by UCITA can be summarized as follows:
1. Coverage
In UCITA Land, UCITA will cover all contracts whose primary purpose is to create, modify, transfer or license "computer information" (section 103). Although "computer information" is defined very broadly, UCITA does not automatically apply to goods unless the main purpose of the transaction is to obtain software embedded in those goods. In practice, UCITA covers contracts for software, on-line databases, Internet applications, HTML source code, computerized music/MP3s, and storage devices that exist primarily to hold computer information.
UCITA will not automatically apply to books, newspapers, magazines, television or movies. However, the opt-in provisions allow a vendor or publisher of information to bring any information-based transaction within the Act (section 103(e)). Even though UCITA expressly states that opting into its coverage will not undermine the effects of any applicable consumer protections, the law may well end up applying to transactions never contemplated by its drafters (Kaner, 1999a).[5]
2. The Shift from Purchase to License, From Good to Service
In UCITA Land, information transactions are licenses for specified uses of information, rather than sales of copies of the information. This is not an entirely new development, as software has been sold under license for some time. However, UCITA is intended to iron out some of the legal uncertainties surrounding this method of transfer -- especially mass market licenses -- and so solidify it.
The fundamental difference between purchasing a copy of software and receiving a license to use it is that a vendor has little ability to impose downstream controls on a purchaser's use of the product, while a licensor can impose many such controls on a licensee. A purchaser of software may enjoy full ownership rights subject to the requirements of applicable intellectual property rights, such as copyright. A licensee may only use the product subject to the contractual terms of the license, which may be considerably more restrictive than intellectual property limitations (Samuelson, 1998). The preamble to UCITA makes it clear that this change in philosophy was fully intended by the drafters (NCCUSL, 1999). UCITA's support of mass market licensing signifies a shift in power between copyright and contract law. Several commentators have argued that this shift is unnecessary and detrimental to consumer interests (e.g., Samuelson, 1998).
3. "Shrink Wrap" and "Click Wrap"
In UCITA Land, a mass-market license (a standard form contract used in dealings with the general public) is enforceable against a customer provided that the customer assented to it after having an opportunity to review it. "Shrink wrap" and "click wrap" licenses are enforceable provided that the customer knew the terms were coming, was given a right to return the product, cost-free, if s/he did not like them, and was reimbursed for any costs reasonably incurred in the process. The license is not enforceable to the extent that its terms are unconscionable, contrary to fundamental public policy or in conflict with the parties' actual terms of agreement.
These provisions validate post-sale warranty disclaimers and remedy limitations, possibly to the disadvantage of consumers (see, e.g., Kaner, 1999b). In UCITA Land, the onus is on consumers to show that terms are unconscionable or contrary to fundamental public policy. Further, competition is limited in UCITA Land because customers do not have the opportunity to compare and evaluate terms prior to purchase.[6]
4. Consumer Protection
UCITA Land consumers may enjoy less protection than Open Land consumers. Although the real space drafters of UCITA argue that the Act preserves existing consumer protection statutes and does not alter U.S. consumer law (Ring & Nimmer, 1999), many commentators and consumer advocates disagree.
UCITA Land consumers ostensibly enjoy a right of return and refund if not satisfied with license terms, and protection for electronic errors in contract formation. But it is arguable that both of these protections are less favorable to consumers than pre-existing case law (Kaner, 1999a). Further, because UCITA defines computer information transactions as licenses rather than sales of goods, these transactions are removed from the scope of several U.S. consumer protection laws, including the federal Magnusson Moss Warranty Improvement Act and California's Song-Beverly Act (Kaner, 1998).[7]
5. Warranties
UCITA includes warranties based on those applicable to the sale of goods under Article 2 of the U.S. Uniform Commercial Code. It also includes implied warranties -- for instance, if the licensee relies on the licensor's judgment as to compatibility of products, an implied license as to compatibility of products will be imputed.
However, UCITA Land software does not necessarily have to conform to documentation provided with it, as representations are only considered to be express warranties if they become "part of the basis of the bargain" (Section 402 (a)(1); see, e.g., ARL, 1999). The result is that users may have no remedy if software fails to confirm to the manual provided with it. A computer industry group, the ACM, charges that UCITA "makes it too easy for software publishers to avoid facing any legal consequences for defective software" (ACM, 1999).
6. Establishing breach
In UCITA Land, either party may refuse performance and cancel the contract in the case of a material breach by the other party. But if the purchase is not "mass market", a party is unlikely to be considered to be in material breach of its contractual obligations if it has substantially performed them (s.701). So if a UCITA Land software vendor delivered a product that did not conform to agreed specifications or differed from documented capacity, and a court considered that the vendor had substantially met the agreed terms, damages might be payable to the customer but there would be no right to terminate. UCITA drafters state that this protects smaller software contractors against capricious corporate customers. Critics argue that this undermines the "perfect tender" rule under the U.S. Uniform Commercial Code, which allows a buyer to refuse a product and terminate the contract if it fails to conform to the agreed specifications (e.g., ARL, 1999).
7. Self help
UCITA Land licensors may exercise self-help remedies following cancellation of a license, including by repossessing information or electronically terminating access to that information, without judicial process (s.815, 816). The remedy is not available unless the licensee agreed to that part of the license, the licensee must be notified at least 15 days before exercise, and the licensor must not exercise it if it will result in a breach of the peace, a foreseeable risk of personal injury or significant physical damage to information or property other than the licensed information (s. 815).
The self-help principles contained in UCITA have been borrowed from the Uniform Commercial Code's treatment of leases and secured parties (UCC s. 9-503, 2A-525). The inclusion of such a provision in UCITA effectively gives a software producer an interest in licensed products equivalent to a secured or lease interest. This substantially extends the rights of producers. User groups are concerned about the potential of self-help to unbalance contract negotiations and disputes -- faced with the risk of business disruptions, a licensee may be intimidated into concessions that it is not obliged to make (ARL, 1999). This remedy also appears inappropriate where the dispute involves a level of complexity.[8]
8. Criticism and Reverse Engineering
In UCITA Land, consumers are arguably unprotected against boilerplate terms purporting to limit their ability to publish results of tests of software or other criticism, communicate non-copyrightable information, or conduct reverse engineering to fix problems (Kaner, 1999a). When copies of software are sold or transferred under open licenses, as in Open Land, intellectual property law protects these user freedoms. The possibility that UCITA will restrict reverse engineering activities is a particular concern. In Open Land, developers can freely reverse engineer mass-market products, but in UCITA Land, there are limitations on this practice that can only be clarified or removed by litigation. These restrictions are likely to make it very difficult to develop free software in UCITA Land (see, e.g., Stallman, 2000).[9]
9. Duration, Transferability and Breadth of License
In UCITA Land, if the parties do not agree on the length of a software license, it will endure for a "reasonable time" subject to termination at will by either party (s.308). This is contrary to pre-existing commercial practice that any limitations on duration need to be expressed in the license (ARL, 1999). In addition, the transfer of software licenses in UCITA Land is likely to require consultation with licensors, which complicates corporate mergers or software financing arrangements. This change in default assumptions is to the detriment of software consumers (Samuelson, 1999).
Further, if a UCITA Land license does not expressly limit the number of software users, it is taken to permit "a number of users which is reasonable in light of the informational rights involved and the commercial circumstances existing at the time of agreement" (s.307(c)). Again, this is an imposition of licensor-friendly default positions in a case where the license is silent (ARL, 1999).[10]
10. Choice of Law and Forum
Parties to a UCITA Land software transaction may choose any legal system to govern their agreement - there need not be any natural nexus between the governing law and the transaction. Where there is no stated choice, the law is either the law governing the vendor's principal place of business or, where a tangible product is delivered, the state of delivery. Under pre-existing law, the agreement of the parties on this point is unenforceable unless they select a jurisdiction with a reasonable relationship to the transaction.
Given that licenses are rarely negotiated in a mass-market context, UCITA Land consumers will often be required to comply with the licensor's choice of law and forum. Consumers are very unlikely to scrutinize these provisions before entering into a licensing contract, and would only become aware of this issue when considering taking legal action. The cost and inconvenience of taking action in a foreign jurisdiction are likely to discourage smaller UCITA Land consumers from pursuing their rights.[11]
Governance of UCITA Land
We have already described the governance philosophy applying in Open Land. In this section, we present, by way of contrast, the governance philosophy that can be extracted from the licensing approach in UCITA Land.
In UCITA Land, users never have ownership rights over code, and in fact can lose ownership rights over their own data. There are no public commons as in Open Land, because all information goods continue to be owned by the corporations that produced them. The goods are only used by permission of those corporations, and rights of use are temporal and constrained. Thus the governance paradigm best suited to UCITA Land is that of the corporate collective.
UCITA Land's corporate collective model of governance brings with it a number of other contrasts with Open Land. UCITA Land consumers do not enjoy the enfranchisement of consumers in Open Land - there are no public commons, so it is up to the corporations whether or not consumers are entitled to access information. Another basic conflict occurs at the level of transparency. UCITA forecloses the possibility of examining the product or licensed service purchased by being based almost entirely on object code. Further, consumers do not have a full opportunity to examine and compare the terms of their bargain prior to purchase.
The progress allowed in Open Land by means of derivations is not a possibility in UCITA Land. No user can own alterations to the code, should they be permitted. All ownership remains with the corporation, and all additions or alterations can be claimed by the corporation. There is no sense in which consumer improvements or additions to code can be contributed back to the common good. Although UCITA Land does have consumer protection, it is arguably weaker than protections in Open Land, and there is no way for consumers to peruse code to ensure integrity or monitor developments. Consumers must trust the "black box" of code, and if it fails, the outcome is legal action. There is little inclusivity or customer freedom in UCITA Land - all such matters are up to the discretion of the corporations. Although parties may choose to opt out of the UCITA framework, mass market consumers' rights to do so are very limited indeed.
UCITA Land limits core speech rights by enabling restrictions on customer complaints. UCITA Land challenges concepts of citizenship by enforcing a vendor's choice of law and forum. In essence this would mean that a citizen of one nation, purchasing from a company incorporated in and resident in that nation, could be required to abide by the rights provided consumers in another nation. Finally, UCITA Land alters the fundamental question of autonomy with controls on reverse engineering and self-help. There is a tradition of the home as inviolate, under which even the state is only allowed entry under highly controlled circumstances. Self-help allows corporate access to personal electronic spaces, physically within the home, in order to enforce contract provisions. When viewed through the lens of governance, UCITA is radical indeed.
Although UCITA Land supporters argue that UCITA's principles simplify and modernize the law, providing a roadmap for consumers and businesses alike (BSA, 2000), Open Land proponents looking at UCITA Land see only a disaster for free software developers and consumers. If free software developers venture into UCITA Land, they will be bound by default rules that impose liabilities they cannot sustain, and which prevent development techniques (such as reverse
engineering) essential to the creation of free software inter-operable with major commercial programs (Stallman, 2000).[12] As for consumers, the U.S. Federal Trade Commission and numerous consumer groups charge that the UCITA framework undermines well accepted consumer protection principles. They strongly question the appropriateness of its approach (see, e.g., FCC, 1999).
CONCLUSION
Most disturbingly, UCITA Land is not a fantasy. UCITA Land already exists in Virginia and Maryland, and may soon exist in other states. At this early stage, the law has not been tested and we can only imagine its full impact. But UCITA introduces a style of governance that would not be tolerated in real space, and that is fundamentally opposed to software culture, which has historically had more in common with Open Land. We are not suggesting that Open Land is utopia, and UCITA Land its dark nemesis. But in an environment where licensing and code together give rise to governance, the choice of a licensing framework has powerful implications. We leave to the reader consideration of those implications, not only with respect to governance, but with respect also to the ability to choose to "opt out" of oppressive frameworks and let code remain a source of ideas, and an object of personal choice.
ENDNOTES
REFERENCES
*Unless otherwise noted, all web sites listed were last viewed in March 2002.
ACM, 1999, ACM letter on UCITA, www.acm.org/usacm/copyright/usacm-ucita.html.
ARL, 1999, Association of Research Libraries, Summary of Key Problems with UCITA for Business Users of Software by Principal Financial - Fall 1999, http://www.arl.org/info/frn/copy/keyprobs.html.
Behlendorf, 1999, "Open code as a Business Strategy" in C. DeBona, S. Ockman and M. Stone (eds) Open codes: Voices from the Open code Revolution, O'Reilly, 1999.
BSA, 2000, UCITA Signed Into Law in Maryland, http://www.bsa.org/usa/press/newsreleases/2000-04-25.132.phtml.
Free Software Foundation, 2000a, Various Licenses and Comments About Them, http://www.fsf.org/philosophy/license-list.html.
Free Software Foundation, 2000b, Non-Copylefted Free Software, http://www.fsf.org/philosophy/categories.html#Non-CopyleftedFreeSoftware.
Free Software Foundation, 2000c, The BSD License Problem, http://www.fsf.org/philosophy/bsd.html.
Free Software Foundation, 2000d, The Artistic License, http://www.fsf.org/philosophy/license-list.html#ArtisticLicense.
FTC, 1999, Bureau of Competition and Policy Planning office letter to Mr. John L. McClaugherty, Chair, Executive Committee, National Conference of Commissioners on Uniform State Laws, July 9, 1999: http://www.ftc.gov/be/v990010.htm.
Hammerly et al, 1999: J. Hammerly, T. Paquin, S. Walton, "Freeing the Source" in C. DeBona, S. Ockman and M. Stone (eds) Open codes: Voices from the Open code Revolution, O'Reilly, 1999.
Kaner, 1998, Comments on Article 2B, October 1998, www.badsoftware.com/kanerncc.htm.
Kaner, 1999a, Letter to NCCUSL, June 1999, http://commons.somewhere.com/rre/1999/RRE.UCITA.formerly.Artic.html.
Kaner, 1999b, Objections to UCITA, July 1999: www.badsoftware.com/debate.htm.
Lessig, 1999, Code and Other Laws of Cyberspace.
NCCUSL, 1999, Uniform Computer Information Transactions Act, http://www.law.upenn.edu/bll/ulc/ucita/ucita1200.htm (June 2001).
NCCUSL, 2000, UCITA fact sheet, http://www.nccusl.org/nccusl/uniformact_factsheets/uniformacts-fs-ucita.asp.
Perens, 1999, "The Open Code Definition" in C. DeBona, S. Ockman and M. Stone (eds) Open codes: Voices from the Open code Revolution, O'Reilly, 1999.
Ring & Nimmer, 1999, Series of Papers on UCITA, http://www.ucitaonline.com/docs/q%26apmx.html#ES. It should be noted that the authors of these papers were involved with the drafting of UCITA; Carlyle Ring was the Chairman of the NCCUSL Drafting Committee on UCITA, and Professor Nimmer was the Reporter to the Committee.
Samuelson, 1998, "Legally Speaking: Does Information Really Want to be Licensed?", Communications of the ACM, September 1998, http://sims.berkeley.edu/~pam/papers/acm_2B.html.
Samuelson, 1999, Letter to NCCUSL, http://www.jamesshuggins.com/h/tek1/ucita_psam_19990709_letter.htm.
SEI, 1999, Letter to NCCUSL from SEI, www.badsoftware.com/sei.htm.
Stallman, 1999, "The GNU Operating System and the Free Software Movement " in C. DeBona, S. Ockman and M. Stone (eds) Open codes: Voices from the Open code Revolution, O'Reilly, 1999.
Stallman, 2000, 'Why We Must Fight UCITA', Linux Today, Feb 6, 2000, http://linuxtoday.com/stories/15948.html.
UCITA Standby Committee, Report to Executive Committee, National Conference of Commissioners on Uniform State Laws, December 17, 2001, http://www.nccusl.org/nccusl/UCITA-2001-comm-fin.htm.
