Open Source Interpretation
Eli Greenbauma
(a) Partner, Yigal Arnon & Co.
Abstract
This Article offers an alternative to the standard assumptions concerning the interpretation of Free and Open Source Software licenses – that such licenses should be interpreted as boilerplate agreements applied by the licensing parties without having negotiation regarding the language of the license. The Article considers some of the consequences of this approach to license interpretation.
Keywords
Law; information technology; Free and Open Source Software
More specifically, this article examines a line of cases in which such “template” interpretive principles have been applied to the boilerplate language of financial contracts. FOSS licenses and financial contracts obviously differ significantly in their structure and goals, and the communities that employ these documents themselves differ in their organization and practices. Nonetheless, this article proposes that the interpretive cannons applicable to financial boilerplate can be usefully applied to understand the meaning of FOSS licenses. This article also outlines some of the consequences of applying this mode of interpretive analysis to FOSS licenses. To be clear, the intention of this article is not to assert that courts will use such “boilerplate” principles to interpret FOSS licenses to the exclusion of other interpretive methodologies. Rather, the intention is to propose that these principles can be usefully employed in understanding and interpreting the language and usage of FOSS licenses, and that courts and commentators should consider employing these principles in their interpretation of these documents.
The typical FOSS license describes a very different transaction than a complex debt agreement. Nonetheless, despite these differences, the Sharon Steel principles applicable to interpretation of the latter can be usefully applied to the former. As with the boilerplate provisions in financial contracts, FOSS licenses are almost always used as template agreements, and rarely are the terms of any specific FOSS licensing transaction subject to any amount of negotiation. Indeed, FOSS licensing transactions would seem to depend on transactional templates to a larger extent than the financial contracts at issue in Sharon Steel. First, FOSS licenses are often applied simply by reference to the applicable licensing template, without detailing the specific legal language of the template license. Second, FOSS licenses are applied as templates throughout a wide range of transactional relationships. For example, FOSS licenses can be applied at multiple points in a supply chain such as between the original developer and an open source project, between the open source project and a distributor of the software, and between any of the foregoing to the end user of the software. Moreover, template FOSS licenses are often used to describe licensing transactions both between large and legally sophisticated technology companies and also with legally unsophisticated individuals. The language of FOSS licenses remains constant despite this heterogeneity of transactional categories and transacting parties.
This Section briefly outlines some of the possible ramifications of adopting a “boilerplate” theory of interpretation in the context of FOSS licenses.
FOSS licensors may occasionally pepper template FOSS licenses with their own understandings of the applicable license. Such glosses could either relax or reinforce the obligations under the license – for example, a licensor could provide a lax interpretation of license obligations or alternatively could describe a strict enforcement policy that goes beyond ordinary community understandings of the license. These glosses provide insight into how the licensor itself understands the parties' obligations under the license, even if those intentions may not always be consistent with community understandings of the license text. The FAQ document made available by the FSF can be seen a prominent example of an interpretive gloss – even if the FAQs are not construed as providing an objective interpretation of the license text itself, they are certainly evidence of the specific licensing intentions of the FSF when acting as a copyright holder.
The classical model of contractual interpretation involves a search for the intent of the contracting parties. But the boilerplate nature of FOSS licenses distinguishes them from the classical model of contracts. In a standard FOSS licensing transaction, the parties may have little regard for the linguistic details of the license, and an excavation of the fossil record of the parties’ intentions with regard to the specific language of the license may raise little of substance. This short article has shown that in courts in other contexts have often recognized the unique interpretative challenges posed by such boilerplate language, and suggested that courts may employ such interpretive techniques in the context of FOSS licensing as well.
About the author
Eli Greenbaum is a partner in the Jerusalem office of Yigal Arnon & Co and adjunct faculty at the Interdisciplinary Center at Herziliya
Licence and Attribution
This paper was published in the Journal of Open Law, Technology, & Society, Volume 12, Issue 1 (2021). It originally appeared online at http://www.jolts.world
This article should be cited as follows:
Greenbaum, Eli (2021) 'Open Source Interpretation', Journal of Open Law, Technology, & Society, 12(1), pp 1 – 10
DOI: 10.5033/jolts.v12i1.143
Copyright © 2021 Eli Greenbaum
This article is licensed under a Creative Commons Attribution 4.0 CC-BY available at
https://creativecommons.org/licenses/by/4.0/
1To the extent United States courts have had the opportunity to interpret FOSS licenses, they have generally not explained their reasoning. See, e.g., Jacobsen v. Katzer, 535 F.3d 1373, 1381 (Fed. Cir. 2008) (interpreting the terms of the Artistic License, without explanation of the underlying interpretative principles). Some United States courts have had the opportunity to interpret Creative Commons licenses (which are not geared towards the licensing of software), and such courts have generally assumed that they can "property rely on traditional tools of contract interpretation." Drauglis v. Kappa Map Group, LLC, 128 F. Supp. 3d 46, 53 (D.D.C. 2015). See also Great Minds v. Office Depot, Inc., 945 F.3d 1106 (9th Cir. 2019); Great Minds v. FedEx Office & Print Servs., 886 F.3d 91 (2d Cir. 2018).
2See Lawrence Rosen, Open Source Licensing Software Freedom and Intellectual Property Law 120 (2004); Heather Meeker, Open Source for Business 98 (2017) (stating that “there is no reason to think that the rules of interpretation” applicable to the GPL “would be different from those applied anywhere else in the law”); Lothar Determann, Dangerous Liaisons—Software Combinations as Derivative Works? Distribution, Installation, and Execution of Linked Programs under Copyright Law, Commercial Licenses, and the GPL, 21 Berk. Tech. L.J. 1421, 1491 (2006) (asserting that, in interpreting the GPL “courts would likely apply contract interpretation rules and try to determine the intent of the copyright owner who selected the GPL and the licensee who selected the program”). But see Van Lindberg, OSS and FRAND, Complementary Models for Innovation and Development, 20 Columb. Science & Tech. L.R. 251 (2019) (since licenses are chosen and not negotiated, "some typical canons of license interpretation may not apply in the open source context").
3Rosen, supra note 2 at 120 (asserting that “[u]nder the law, only the common understanding of a licensor and his licensees matters, as reflected in the written terms and conditions of the license agreement between them”); Lindberg, supra note 2, at 257 (using the Open Source Definition to "illuminate the 'meeting of the minds' between licensor and licensee to the extent it exists"). Jason B. Wacha, Taking the Case: Is the GPL Enforceable?, 21 Santa Clara High Tech. L.J. 451, 487 (2004). (asserting that “even if the FSF's language is additive, in analyzing a contract, a U.S. court would look to the intent of the parties”); Determann, supra note 2, at 1491 (asserting that in a dispute about the GPL, "courts would likely apply contract interpretation rules and try to determine the intent of the copyright owner who selected the GPL and the licensee who selected the program").
4Rosen at 120 (asserting that “it is legally unnecessary to know what the drafter of a license – usually just an attorney with no stake in the matter—meant to say”); Lindberg, supra note 2, at 257, n.18 (asserting that for the interpretation of open source license, the "comments of license stewards are interesting but of questionable relevance"); Determann, supra note 2, at 1491 (asserting that the "examples, explanations, and programmatic and ideological statements" of the Free Software Foundation are likely to be of limited relevance if the FSF is not itself a party to the dispute).
Robert W. Gomulkiewicz, De-Bugging Open Source Software Licensing, 64 U. Pitt. L. Rev. 75 n.185 (2002) (asserting that “once a hacker applies the license to his or her software, it is the hacker’s intention that controls” interpretation of the license).
5Rosen at 120 (asserting that “If there is an ambiguity or uncertainty of interpretation in a license, the license will generally be interpreted against the licensor regardless of what the license drafter meant to say”); id. at 282 (under general principles of contractual interpretation “courts may protect individual consumers from unfair license conditions where they wouldn’t bother to protect a sophisticated company…”); Meeker at 100 (interpretation of the GPL may employ the canon of contra preferentum”); Determann at 1491. See also Gomulkiewicz n. 40 (noting that the ordinary canon that contracts should be interpreted against the drafter may, in the context of open source licensing, be modified somewhat by intellectual property law).
6Similarly, it may sometimes be possible to characterize the often terse “fair, reasonable and non-discriminatory” (FRAND) patent licensing commitment in similar manner – as template language meant to evoke community expectations rather than negotiated terms to be construed as per the parties’ expectations. Admittedly, the specific language used to implement FRAND obligations may not always be unmodified boilerplate. See, e.g., Jorge L. Contreras and Anne Layne-Farrar, Non-Discrimination and FRAND Commitments in Cambridge Handbook of Technical Standardization Law: Competition, Antitrust, and Patents 186, 190 (Jorge L. Contreras, ed., 2017) (suggesting differences in the meaning of FRAND commitments that prohibit all "discrimination" as opposed to only "unfair discrimination"). But see Stephen J. Choi & Mitu Gulati, Contract as Statute, 104 Mich. L. Rev. 1129, 1149 (asserting that under a boilerplate analysis "it is not at all clear that the small differences in language … should be given much weight at all").
7Sharon Steel Corp. v. Chase Manhattan Bank, N.A., 521 F. Supp. 104, 107 (S.D.N.Y. 1981), aff'd 691 F.2d 1039 (2d Cir. 1982), cert. denied.460 U.S. 1012 (1983).
8Id.
9Id. at 109.
10Id. at 108-109.
11Id. at 112-13.
12Id. at 113.
13Id.
14Id.
15Sharon Steel Corp. v. Chase Manhattan Bank, N.A., 691 F.2d 1039 (2d Cir. 1982).
16Id. at 1048.
17Id. at 1048-50.
18Id. at 1051 (holding that the interpretation of the clauses at issue should "best accommodate[] the principal interests of corporate borrowers and their lenders"). See also Choi & Gulati, supra note 6 at 1156 (noting that the court in Sharon Steel recognized the “need to defer to market understandings” in interpreting boilerplate contracts).
19Id. 1048
20Id.
21Id.
22Kolbe v. BAC Home Loans Servicing, LP, 738 F.3d 432 (1st Cir. 2013); Feaz v. Wells Fargo Bank, N.A., 745 F.3d 1098 (11th Cir. 2014).
23Id.at 1105.
24Id.
25See, e.g., Open Source initiative, Report of License Proliferation Committee and Draft FAQ.
26Id.
27See Rosen, supra note 2, at 235 (stating that open source “licensors should use an existing template license. Please don’t invent your own. The open source community is not seeking new licenses to analyze and interpret”).
28For a summary (and criticism) of the standard view concerning the ownership and licensing of FOSS projects, see Pamela S. Chestek, A Theory of Joint Authorship for Free and Open Source Software Projects, 16 Colo. Tech L.J. 285 (2018)
29See generally, Eli Greenbaum, The Non-Discrimination Principle in Open Source Licensing, 37 Cardozo L. Rev. 1297 (2015)
30Id. at 1302.
31See Section 5(c) of the General Public License version 3 (requiring that works “based on the Program, or the modifications to produce it from the Program” may be conveyed only “as a whole, under this License”).
32See supra text accompanying notes 22-24.
33691 F.2d at 1049
34Concord Real Estate CDO 2006-1 v. Bank of Am. N.A., 996 A.2d 324, 331 (De. Ch. 2010)
35Id.
362015 Bankr. LEXIS 3991 (S.D.N.Y.)
37Id. at *36.
38Id. at *40. But see Great Minds v. Fedex Office & Print Servs., 2017 U.S. Dist. LEXIS 24063 (E.D.N.Y.) (not allowing the Creative Commons Corporation to file an amicus brief, since the court was not then required to interpret the license text).
39Id. at *42.
40GNU Operating System, Frequently Asked Questions about the GNU Licenses, https://www.gnu.org/licenses/gpl-faq.en.html.
41This documentation is available at Free Software Foundation, Welcome to GPLv3, http://gplv3.fsf.org/. For commentators advocating use of such material in interpreting boilerplate, see Choi & Gulati, supra note 6 at 1167-68 (asserting that courts should look to the historical record of the drafting process).
42See, e.g., Apache Software Foundation, Frequent Questions about Apache Licensing, http://www.apache.org/foundation/license-faq.html; Mozilla, MPL 2.0 FAQ, https://www.mozilla.org/en-US/MPL/2.0/FAQ/.
43Indeed, courts may also find interpretive support in statements from such institutions regarding the historical background of FOSS licenses. For an example of a commentator attempting to use such evidence in the interpretation of FOSS licenses, see David J. Kappos & Miling Y. Harrington, The Truth about OSS-FRAND: By All Indications, Compatible Models in Standards Settings, 20 Columb. Sci. & Tech L.R. 242, 247 (2019) (using recollections of current staff at MIT and University of California, Berkeley to interpret the MIT and BSD licenses, respectively). For commentators advocating such an approach in general, see Choi & Gulati, supra note 6 at 1167 (asserting that the "starting point" in the analysis of boilerplate terms should be the "historical understanding" of the clause, including "discerning the intent of the original drafters").
44Courts construing FRAND licensing commitments have also looked towards the understandings and goals of standards organizations in interpreting that commitment. For example, in FTC v. Qualcomm, 2018 U.S. Dist. LEXIS 190051, *38-39 (N.D. Cal.), the court made use of the guidelines and "statements of purpose" of the relevant standard-setting organization in interpreting the FRANC commitment, even though the court framed this inquiry as a search for the intent of the contracting parties. Id. at *26.
45See, e.g., Gregory Klass, Boilerplate and Party Intent, 82 Law and Contemporary Problems 105, 127 (2019) (noting, in observing courts’ contrasting approaches to the view of the Insurance Services Office and the ISDA, that “[t]he ISO’s partisan nature distinguishes it from the ISDA, which generally represents the interests of both sides in a derivatives sale. Consequently, whereas it can be appropriate for courts to look to the ISDA Master Agreement’s drafting history and to the ISDA’s present position on its meaning, it would be inappropriate to give similar weight to the drafting history of ISO standard forms or to ISO’s present understanding of them”).
46A licensor may not have unbridled freedom to annotate a FOSS license. For example, section 7 of GPLv3 allows licensees to ignore any "further restrictions" that the licensor may impose in addition to the license terms. It is unclear how a court would approach a situation in which a licensor added a "further restriction"- would a court indeed allow the licensee to simply "remove" that further restriction as permitted by the original text of GPLv3, or would a court find a way to reconcile the contradictory texts? A situation more relevant to the topic of this article, and perhaps subject to even more legal haziness, would be a situation where a licensor publishes a restrictive interpretation of the GPL, and a licensee counters that such restrictive interpretation should be viewed as a severable "further restriction".
47 738 F.3d 432 (1st Cir. 2013).
48Id. at 436. See also Kaiser Aluminum Corp. v. Matheson 681 A.2d 392 (Del. 1996). In Kasier, the Supreme Court of Delaware was called on to interpret the “hopelessly ambiguous” language of a corporate certificate of incorporation. Despite the opacity of the contract, the court noted that it was reluctant “to rely on extrinsic evidence” of the parties’ intentions, since that could introduce “inconsistencies” in the interpretation of such template documents. Id. at 398
49Sharon, at 1048.
50See, e.g., Intel Corp. v. VIA Technologies, Inc., 319 F.3d 1357, 1363 (Fed. Cir. 2003) (applying contra proferentem in the interpretation of Intel’s template royalty-free license agreements for patents applicable to a technological standard, because “Intel alone drafted the agreement and had complete control over the language of its terms”). While this case involved a template license, the usage of the template at issue differed from FOSS licenses in that the template was only used by a specific licensor (Intel) in the licensing of patents for a specific standard.
51For some commentators suggesting the application of contra proferentem in FOSS licenses, see supra note 5.
52To some extent, seeing the FSF as the sole “drafter” of GPLv3 is not completely accurate. Rather, the GPLv3 license was drafted in a public process which included many participants. Though the FSF seems to have had sole control over the final license text, the concerns animating the intermediate and final drafts of the license text were made publicly available. See generally, Free Software Foundation and Software Freedom Law Center, GPL3 Process Definition (2006), at http://gplv3.fsf.org/gpl3-process.pdf. The author has not found court cases that address the application of contra proferentem in such situations.
53Broad v. Rockwell Int'l Corp., 642 F.2d 929, 947 n.20 (5th Cir. 1981)
54nother concern raised by courts discussing the application of contra proferentem in template documents is that the language of contractual boilerplate was not “actually discussed and thought about by the parties.” Id. In other words, in a transaction in which the parties simply applied boilerplate, neither party actually negotiated the language of the agreement and so there is no “drafter” against which ambiguities can be interpreted. As noted above, FOSS licenses may actually suffer from the opposite concern - while FOSS licenses typically have a license “steward” that publishes the license, FOSS licenses have also been drafted in a public process in which a surfeit of participants are able to submit comments and concerns regarding the license language. See supra note 52.
55See also David Horton, Flipping the Script: Contra Proferentem and Standard Form Contracts, 80 U. Colo. L. Rev. 431, 436 (2009) (arguing that without the rule of contra proferentem, the meaning of mass-produced terms would fluctuate with the particulars of each deal, leading to perverse results”).