From October 2016 to March 2017 the team is joined by Guest Kats Rosie Burbidge and Eibhlin Vardy, and by InternKats Verónica Rodríguez Arguijo, Tian Lu and Hayleigh Bosher.

Friday, 8 June 2012

Who Uses Open-Source Software: Are Some Scholars Late to the Party?

Twice a year this Kat shares the podium with an esteemed IP colleague as we jointly lecture the new batch of articled clerks on principles of IP law and their application in practice. One of the topics that always draws particular interest is the topic of open-source software. The attraction seems clear enough. Over and against the commercial model resting on the licensing of proprietary software, the open-source model, in all its variations, challenges the young lawyers to think about how copyright can be employed commercially in a quite different manner.

What serves as a point of discussion for our articled clerks underscores the more general question -- why do people select open-source software over proprietary software? A recent article, "Who Chooses Open-Source Software", by the distinguished Stanford professor Mark Lemley and his co-author Ziv Shafir (The University of Chicago Law Review, vol. 78, Number 1), discusses this question. The context of their study is well-explained as follows:

"The rise of open-source software poses an important challenge to the classic account of the production of intellectual public goods. Instead of using IP rights to optimize monetary benefit, open-source production relies on IP rights to keep software,and any improvements or additions to it, free and widely accessible. Open-source software is provided to others for free; providers profit, if at all, not by selling the software or improvements to it but by providing consulting or other services.

Scholars vigorously debate whether open-source software represents a fundamental new means of collaborative production potentially extendable to other forms of human endeavor or an altruistic fringe to the dominant market-based model of production."
Lemley and Shafir pose the economic question on the basis of what kind of economic person is attracted to open-source software. Is he or she a rational economic actor, actuated by rewards expected to exceed costs, in which case "open-source production is likely to be limited to the creation or relatively low-cost or small scale products ....?" Or, to the contrary, do we have a situation in which the actors "are collectively motivated to create nonfinancial incentives, of if there is a sustainable market for the provision of services ancillary to open-source products", whereby the open-source is both scalable and extendable both within and beyond computer software only?

The authors consider these questions by exploring the kinds of use following the release by a biotechnology company (Affymetrix here), of similar software under both proprietary and open-source licences. The software provided by the company complements its product for use in certain DNA-based laboratory tests. Such a dual release strategy is apparently common in the open-source licensing world. For those who are interested, the details of the study are described in the article. Sufficient for this Kat's current purposes is a summary of the author's main findings and conclusions.

With respect to an analysis of publications that made reference to the software algorithm used for interpreting the results based on the Affymetrix product (for our purposes, open-source or proprietary software), the results were as follows:
1. Academic users are much more likely to use the open-source version of the software.

2. Commercial users made far less use of the open-source version and, even when they did, fully one-half of commercial users did so within the context of academic-commercial collaborations.
With respect to survey data of users of the Affymetrix product (which captured more non-academic users), the results were as follows:
1. There was no material difference between commercial and academic users as to whether open-source or proprietary software was being used.

2. For those who preferred open-source software, the two main factors are cost and the ability to modify the software. However, the differences between open-source users and users of proprietary software on these two parameters was only modest statistically.

3. Users of the proprietary software were more likely to mention the presence of an existing institutional licence and convenience of the software.
The implications offered by the authors based on these results reveal, not for the first time, that the "pure" legal view of the issue at issue differs from what seems to be going on in practice.
1. In practice, there appears to be a significant amount of mixing of both open-source and proprietary software. "Users [of the software studied] seem largely indifferent to whether the software they use is open-source or proprietary." Stated otherwise, "they seem to use those software tools as they will -- and not as the niceties of open-source contracts would suggest."

2. Those who make improvements seem to be indifferent to whether open-source or proprietary software is used, thereby apparently "ignoring the fundamental constraint of open-source software--that you release your software to everyone."

3. As summarized by Lemley and Shafir, contrary to what legal scholars might think, "[l]aw may matter to makers of open-source software [or not--IPKat], but it does not appear to affect the behavior of software users."
This Kat is grateful for these results because, quite honestly, they comport with his anecdotal impressions of how open-source software is used. Indeed, his colleague, who spoke to the clerks on open-source software, pretty much said the same thing to these young lawyers based on her rich experience in the field. Maybe what we need, à laJohn Le Carré, are more scholars who have come in from the open-source software cold.

More on John Le Carré here.


JH said...

Strange piece.

Some comments:

* Users mix proprietary and open-source software

Why not? I can follow a Google link to a Wikipedia article one moment, then pull a printed reference book off the shelf the next. Does that make me a conflicted person? Similarly, I've got some great open-source software on my computer; I've also got some fine paid-for software. What's the problem?

* "the fundamental constraint of open-source software--that you release your software to everyone."

Actually, rather few open-source licenses require you to release your improvements. Under BSD / Apache / Android type licenses, the choice is entirely up to you: you can release your changes or not as you wish. Even the more "viral" GPL license only requires you to make available the source code of your changes if you are releasing new binary versions to customers or third parties. If you are just modifying the code in-house for your own use, then even under the GPL there's no obligation to release that.

But there may be several good reasons why it may be sensible to release your changes -- for example, you may then benefit if somebody else uses your code to make even better changes, that perhaps make the software you're interested in work even better (or make the hardware you're trying to sell work even more slickly). Also, it can be very useful to get your changes integrated into the main development trunk, so that then they get maintained and updated by the central development team, rather than you having to produce new versions of your changes yourself every time the underlying software gets an update.

* open-source production is likely to be limited to the creation or relatively low-cost or small scale products

I'm sure that would come as most interesting news to all those who have contributed fixes and patches and new developments to Linux and any number of considerably large-scale open-source projects, often from stone cold sober self-interested rational economic motivations.

Anonymous said...

In point 2 of the list of what is happening in practice, you make a quote which I assume is from the paper:

"ignoring the fundamental constraint of open-source software--that you release your software to everyone"

That sounds to me like a complete misunderstanding of (one of) the fundamentals of open source, at least as I understood it. The point is not that if you modify a piece of open source software, you are obliged to release it to the world (you're not). The point is that if you DO decide to distribute your modified software, you must also distribute the modified source code - it's "open source", not "open software".

(Of course I may be being unfair - it's hard to tell without reading the actual paper, the quote may be misleading out of context.)

Gentoo said...

I agree with JH "strange piece".

In so many ways the results may be filed under "no shit, Sherlock? As with music and DRM end users don't think about it providing it "just worked". The backlash happened when the consequences of the the loss or damage of a £150 music player made it difficult or impossible to recover £1,500 worth of music files plus time and effort.

Not everyone has to care about everything for it to be important.

However if you really think OSS matters only to a few you should ask yourself why proprietary software vendors and their friends from the telecommunications industry applied so much pressure (in public and behind the scenes - for some evidence read Mark Ballard, Computer Weekly) on Cabinet Office and its consultation on FRAND v Royalty Free open standards.

We all know the the majority of OSS is released under one or other of the GPL and is thus incompatible with FRAND. If OSS really didn't matter why spend so much time lobbying to block it?

Pamela Chestek said...

I agree with others that there is a fundamental flaw in the authors' premise, that a GPL license requires that the modifications be made available to others. That is simply not true; the requirement is that IF one distributes binary code, one must also provide the matching source code. The survey question asked was fine: "Why did you choose the version you did?" with one possible response "It gave me greater freedom to modify the software." Quite possibly a respondent knew that he or she wanted to customize the software for the person's particular application, but from that one cannot possibly infer that the individual understood that to include a burden to publish the changes because there simply is no such burden.

The authors accurately stated one reason for a dual licensing scheme: "the belief that customers would want proprietary software because they hoped to improve on it and sell the improvements." (p. 152). The GPL requires that all derivative works be also licensed under the GPL, so a dual licensing scheme provides an alternative licensing vehicle that will allow the licensee to lock up its modifications under a proprietary license. Whether this assumption was an accurate one would have been interesting to learn, but, alas, we'll never know from this study. After determining that there was distribution of modified software by both open source and proprietary licensees, what the authors never probed is what license was actually used for these derivative works, i.e., whether the redistributions of the code acquired under the proprietary license was indeed subsequently distributed under a proprietary, revenue-generating license.

Confusingly instead, the authors characterized the distribution of modified software as "under their own brand" or not which, as far as I can tell, was an attempt at a proxy for whether the modified software was revenue producing or not. However, the question is meaningless. It may have been the legal conclusion of those redistributing the software that the original brand could not be used for modified software, whether redistributed under an open source license or proprietary license. It is therefore false to suggest that the branding has anything to do with either revenue or the type of license used for the redistributed software.

So under "Implications," we have some troubling conclusions. "[The users] often use multiple programs and algorithms, some of which are open source and some of which are proprietary. That sort of mixing is a big worry for open-source lawyers ...." (p. 152). No, it's not. A user can mix and match however he or she pleases, because the licensing requirements for open source software occur on the distribution of the code, not the mere use. Software distributors must ensure that they can and do comply with the licensing obligations for all their inbound licenses, whether those licenses are open source or proprietary, but suggesting that there is or should be any concern at the user level is patently false.

"Curiously, even these improvers seem largely indifferent to the open-source–proprietary line, to the extent that many of them appear to be ignoring the fundamental constraint of open-source software—that you release your improvements to everyone. In fact, it is not clear that they are even aware of the limits on behavior that open-source licenses impose.... Law may matter to makers of open-source software, but it does not appear to affect the behavior of software users." (p. 153). As noted, this is a completely inaccurate statement about open source licenses. It is the authors who are unaware of the behavior that the licenses impose; the users appear to be doing just fine.

Where on earth was peer review on this article?

Anonymous said...

Personally, I use open-source software (when available) rather than proprietary software because it saves me money. Sometimes it's not quite as good, but the rapport qualité-prix is just fine.

Subscribe to the IPKat's posts by email here

Just pop your email address into the box and click 'Subscribe':