Reviewer: Jakob Rigi, Central European University

31 October 2011

Overall evaluation:

The strength of the article is its ethnographic case studies and its emphasis on the study of critical junctures/junctions. Both cases are interesting and authors have picked very relevant issues.

This positive side withstanding, the article has the following weaknesses.

I.
Authors ignore both the general and specific contexts of the quotes from the authors they are criticizing for confusing proprietary with commercial software. The general context is the opposition that Richard Stallman established between Free Software (SF) and proprietary software through the invention of GPL. Stallman did this in reaction against commercial companies that made codes, freely exchanged before, industrial secrets. Companies made the codes secret, because they wanted to sell the software. In this sense GPL emerged as a reaction to the commercialization of the code. And therefore, the proprietary software was a commercial software and SF non-commercial one. It is true that GPL also permits the use of the code for commercial purposes but this does not make the software commercial. While proprietary software is fully commercial the Free Software may have a commercial use. In most contexts it works like a non-commodity (commons) and in some context it can become a commodity. Therefore, Free Software is a hybrid form between commons and commodity, and in different contexts of exchange the one or the other aspect may become its predominant feature. This hybrid form can be contrasted to the pure commercial form: proprietary software. This is, I think, what all scholars, criticized by the authors, do without being ignorant about the hybrid nature of FS.

<Authors’ response:> We agree partially with the comment. This is because in the paper we do not claim that authors are “ignorant” about the issue. More simply and more modestly we say that their use of the term “commercial software” is largely different from that of FLOSS stakeholders. Isn’t this something very singular and worth looking into it?

This is just an empirical consideration and not an accusation. We also of course ask ourselves the question “why” there is such a difference of views between academics and stakeholders, and try to provide an answer in the conclusion of the essay.

The hybridity (commons-commodity) and novelty of FS presents a challenging problem for both scholars and developers. The complexities that FS generates become more obvious if we consider it in the context of a full range of real and hypothetical ideal types software (in Max Weber’s sense).

1. None commoditized, gratis software, with free code. (NCS).
2. FS, a hybrid form. Any software that uses any code from a FS must release all its codes under a GPL license.
3. Open source allows for the combination of free software and proprietary software (FS-PS). It requires that only the components which are derived from FS should be released under GPL.
4. Proprietary software. (PS)

As we see (1) is a pure non-commodity and (4) is a pure commodity form. (2) and (3) are hybrid forms. 2 is leaning towards commons and (3) towards commodity. It seems to me the authors level the difference between (2) and (3) under FLOSS. FLOSS, indeed, can be many different things, depending on relative proportions of propertied code and free code which are included in it.

So the contrast that the criticized scholars establish between the FS and commercial software is a relative one, and in this sense a correct one. In dealing with work of other scholar on FS and FLOSS we need to see how they deal with their hybrid nature. The authors of the article do not do this. They do not specify the particular context of each quote which, indeed, determines the meaning of the opposition free-commercial for each scholar. More careful reading of each scholar may show that they present different meanings or at least different shades of meanings of this opposition.
<Authors’ response:> Thanks for pointing these aspects out. We just base our work on an empirical consideration: for FLOSS stakeholders it seems to be misleading to oppose FLOSS to commercial software. FLOSS stakeholders are clearly against this opposition and are building a different view point in which FLOSS is commercial.

Of course you are perfectly right in saying that there are mixed models and various gradients (BTW we have clarified this in the text in order to address your comment, including some references), but our point is another one: for stakeholders opposing FLOSS to the term commercial software is a mistake.

Further, we do provide different contextualization of the opposition between FLOSS and commercial software which is given by authors (licenses discussions, code access control, organizational aspects and so forth). We have better clarified these aspects in the new version of the essay, in order to answer this comment.

II.

The case studies are exciting but do not support authors’ conclusion that FS is commercial.

Let us look to each case separately.

Case 1. GRAS is a software supplying geographical data formats.

It is run currently, according to the article, under GPL /GNU license.

The Open DWG library is an import/export library produced and maintained by Open Design Alliance. “DWG is the native, and proprietary, format of several CAD packages including the well known AutoCD” (p. 14). Many maps come in OpenDWG format. GRASS used v.in.dwg command from Open DWG. Then a developer of GRASS posted an email saying that this violated GPL. As result the v.in.dwg was eliminated from GRAS. After a year asecond developer of GRAS discovers that Open Design Alliance permits the use and distribution of its library in free of charge Software. He gets the following confirmation from OpenDWG: “We allow access to our libraries for research for research purposes and development of free or internally used software.” (p, 17). Then, this second developer suggests that they may include v.in.dwg in GRAS without violating GPL. He tells a third developer “ He [Aron Dahlberg from Open Design Alliance] assured me that the alliance’s intent was only to restrict or control commercial use of their libraries, not use in educational or free software.” Then this third developer intervenes and claims that Dahlberg does not know the definition of Free Software and confuse it with proprietary gratis software.

Then the following statement taken from Open DWG website is posted in the mailing list of GRAS.

Open Design Alliance members have created the following free utilities, based on the OpenDWG libraries, for your unrestricted, non-commercial use. Please note that inclusion of any utility in a commercial product requires a commercial licensing.
(p, 20)

Then the authors conclude:
That the post clarified that GRAS could not use OpenDWG due to its own commercial nature. They claim that OpenDWG conditions bring to the fore the identity of FS as a commercial software for developers in opposition to non-commercial software.

What is wrong with this conclusion? v.in,dwg could not be included in GRAS, because GPL permits commercial uses, but the Open Alliiance does not permit it. But as I argued in section I on GPL such a permission by no means make it commercial. It is a hybrid form. GPL is an assembly composed of two heterogeneous components, faced to a truly non commercial license (the hypothetical type 1 in the above typology) it may appear as commercial, but faced to (4, and 3) in above typology it will appear as non-commercial or less commercial. Let me remind it again that Stallman invented GLP/GNU to combat the commercialization/commoditization of the code. This history cannot be abstracted from any discussion on the relation of FS with commerce, because it is still an important part of the debates on FS among communities.

We cannot deduce a general commercial nature for GPL/GENU from this case. If the actors do so they are confused.
<Authors’ response:> Your statement/comment clearly illustrates that the problem relates with the type of investigation that we do (epistemology) and not with substantive problems. If we did not put the actors opinion in the foreground we would not be able to appreciate the difference between academic and stakeholders discourses.

We would like to clarify our position here: we do not claim that authors are wrong in opposing commercial software to FLOSS. Rather, we claim that in many cases FLOSS stakeholders see the opposition between commercial software and FLOSS as largely wrong. The actors might be confused, but this is not the point. The point is: why for academics there is an opposition FLOSS vs Commercial Software , whereas this opposition is wrong for FLOSS stakeholders? This is a legitimate question to ask, we think.

 With ethnomethodology we do not impose out view (the researchers view) onto actors. Rather we try to identify the actors own theories about reality. If actors say that FLOSS is not in opposition with commercial software, then we can just recognize this and account for it. What you point here is a substantive issue (actors are confused and academics are not), whereas on the contrary we approach the problem from an epistemological point of view (we do not care if actors are confused, we just account for the actors ethnomethods without imposing our own theory).

Stallman and his followers still contrast FS not only to the type 4 (proprietary software) but also to the type 3 (FS-PS). I know personally of developers who are allergic to the term Open Source, and whenever I use the term Open Source to describe their practice, they immediately correct me by saying “not open sources, but free software”.

<Authors’ response:> Yes. But there are thousands of papers that describe this ideological opposition. This (describing this ideological confrontation) is not our goal and we also point references that the readers can use in order to learn about this. In order to address this comment we added a reference to the Open Source definition (which was required by another reviewer) and explained briefly the differences between FS and OS.

Obviously there is an ideological battle around the interpretations GLP/GNU and Open Source and their use in practice: Richard Stallman versus Erik Raymond orientations. Stallman’s followers are dismissive of Open Source Software, because they think it corrupts the FS by subjugating it to commercial intersts.

You should have said, subjugating FS to proprietary interests. Because this is what Stallman and the other say. We made clear that the opposition to FLOSS is proprietary software and provided and explicit definition.
Case 2
Open Solaris

Sun microssystem releases Solaris under a combination of propertied and GLP licenses on file basis. Solaris license is called (Common Development and Distribution License ) (CDDL). The Sun aim from transforming the previous proprietary license to a new license that contains both GLP and propertied licenses for different files is commercial gains. This, as authors quote, is stated clearly by the Sun. From this and the use of commercial in the Sun’s declarations and statements the authors claim that the actors in the Sun assign to FLOSS a commercial nature. Let us accept this from authors. But does this make those codes system under GPL commercial? Not necessarily. Actors claim cannot be taken at face value.

Sun adopts the new license with a view that voluntary cooperation among many corporations and individuals produce better software. Now assume that a group of companies located in a region build a public road because they want to transport their goods. They use it for transportation of their goods but the public use it for free for other purposes. Although the road is used for commercial purposes itself is not an item of commerce. Actually when companies and volunteers cooperate in producing a software under GPL, the companies use the software for commercial purposes and other actors for other purposes. The software itself does not necessarily become an item of commerce. It is unlikely that someone pays for software which is available for free on the internet. GLP permits every one to publish the software in any form.

The selling of a complex system requires that some of its components are propertied. This is the logic of CDDL. Now we can ask the following question. Is this a corruption of GPL? The question is of both political and historical importance. The authors need to delve into CDDL, and map out its hybrid structure and compare it with FS. What files and why are licensed under GPL? What files are proprietary and why? How do these two types work together? What is the extent of each type? How does this system serve commercial interests? What type of legal and technical problems, and contradiction does it create?

<Authors’ response:> Thanks for pointing this out. You are perfectly right. This is also why we did this (comparison between the 2 licenses and how they are used in these 2 case studies) precisely in another – already published – essay, that we also explicitly quote in this paper, See: http://firstmonday.org/htbin/cgiwrap/bin/ojs/index.php/fm/article/view/2064/2030 .

Authors are perfectly correct in saying that the academic and actors construction of the meanings of GPL are important for the control and the direction of its development. But the actors points of view cannot be privileged. They should be examined critically, in their diversity. From these two cases we cannot conclude that all actors in the field construe software under GLP as commercial .

<Authors’ response:> Again, it seems to us that very much depends on the perspective. We use ethnomethdology in order to identify actors ethnomethods and show the differences between these ethnomethods and the academic discourses. Your comment seems to be directed against ethnomethodology as a research perspective/theory in general and not against our paper which is an application of that perspective/theory.

Recommendation

I suggest that the authors make the hybrid nature of various FLOSS such as GPL, CDDL, … the central focus of their description and analysis. They also need to explicate the difference between GLP and CDDL, and other type 3 licenses (FS-PS).

<Authors’ response:> Thanks for pointing this. You are perfectly right. As we tried to clarify above, we have already published an essay in which we compare the two licenses within the case studies, see: http://firstmonday.org/htbin/cgiwrap/bin/ojs/index.php/fm/article/view/2064/2030.
This paper does exactly what you ask and addresses your comment.

The paper we are submitting to JoPP is an extension of that work that looks at a different issue: how FLOSS stakeholders define what they do.

And also need to look at the hybrid practices/discurses of the Open Alliance in relation to OpenDWG, though their hybridity might be of a different type than that of FS. Compare these two types of hybridity. They need to identify specific forms of hybridity case by case and show why and how each form is produced and contested. Then, they can examine the views of both actors and scholars critically in the light of their findings and present their own conclusions, though not as the actor’s point of view but their own.

<Authors’ response:> See above comment. We have already published on this.

They need to give a more complex and contextualized accounts of the work of scholars they criticize.

<Authors’ response:> Thank you for this recommendation. We have improved the essay taking in account this consideration.

They also need to analyze in some depth their case studies and compare them with cases studied by others.

<Authors’ response:> Case studies have been analyzed in-depth (they are qualitative research conducted over a quite long period).

The discussion should be considered in the context of the history of GPL and the emergence of peer to peer production.

<Authors’ response:> This is quite far from our work. The history of the GPL is not really an issue for us. Besides we have already analyzed this in other publications. We feel we cannot really take in account this recommendation.

Reviewer: Johan Söderberg
The authors are homing in on one recurrent error in the terminology used by many academics. Very often, FOSS is opposed to commercial software, as the authors correctly note. An observer does not have to spend many hours on the Internet to find out that most FOSS developers would protest against such an antinomy. The correct way to talk about FOSS is to contrast it with proprietary software. This observation is confirmed by the aboundant amount of commercially driven FOSS projects. The authors are right in making this point, and it might be worthvile having such an argument in CSPP.

The more pressing question is what strings one can pull from this observation. The authors take it as a pretext for advocating the ethnomethodolgical approach espoused by Actor Network Theory (ANT). ANT prescribes the scholar to put the narratives of the practitioners at the foreground. Accordingly, the researcher should strive to start her investigation without any prefigured theoretical system. Otherwise, it is said, she will read her theories into the stories told by the informants. It is this idea which has attracted the two authors to ANT. Latter on they endorse Grounded Theory for much the same reasons. I can sympathise with this ambition. They go astray, however, in thinking that the pledge to respect the stories told by the practitioners is the property of one theoretical school, setting it apart from all other traditions within the social sciences. Rather, I would argue, it is a mark of craftmanship which the scholar may or may not acquire through experience and devotion, quite independently of the theory abided to. Theory cannot be written in such a way that it guards against poor craftmanship.

<Authors’ response:> We have erased almost all the references to ANT. We do not care about pushing this or that theory, this is not our goal. We would only like to – as you correctly point – put the narratives of the practitioners at the foreground. This is our general perspective.

This reflects back on the argument of the authors in the following ways:

1) Is the subject matter relevant?

The authors do not seem to extract more from ANT than they find in Grounded Theory and the stricture of ethnomethodology, which is to leave out a whole lot of claims made by ANT researchers. This might be for the better, still, the authors need to justify why they draw on this extensive, ontological worldview about symetry between humans-nonhumans, and develop its relation to grounded theory. A interesting article could be made if the authors approached in a grounded theory way to test if the end result will be anything resembling a symmetry between humans and non-humans. But frankly, I believe that the empirical material presented in this article points in a different direction, namely: that the academic system with its ”publish-or-perish” is broken and leads to a lot of shoddy research were academics do not take care about practitioners’ terminology.

<Authors’ response:> We do not enter in the ”publish or perish” issue of course, this is not the goal of this essay. In any case we have removed most of the references to ANT, which we thought were not adding much to our paper. We would really like thank you for your comment which helped us to recognize that much of ANT was not really used in the paper.

We just kept the ethnomethodological perspective. By erasing the ANT parts the paper has become much more clear and easier to be read.

2) Is the treatment of the subject matter intellectually interesting? Are there citations of bodies of literature you think are essential to which the author has not referred?

Perhaps to strenghten their own case, the authors treat the failure in the academic literature as a monolith (a bit of their own medicine would be good when researching this empirical material). In fact, they are writing for a journal where the readers and contributors (almost all of them academics) know that free software is not anti-commercial in a simple sense. The authors need to recognise the existence of this literature too.

<Authors’ response:> Thanks for pointing this out. We use some of this literature, and some has been added: Lin (2005), Kelty (2011), Coleman (2005).

3) Are there any noticeable problems with the author’s mean of validating assumptions or making judgements?

The long list of quotes certainly brings home the point that many academics have used a sloppy terminology. Perhaps this list overdoes it a bit and could be shortened, highlighting the influential academics (von Hippel for instance) to develop their critique against them, and removing those names who hardly can be called academics (Raymond). In fact, Raymond is an insider, and might be more of a problem for the authors’ argument, it shows that ”natives” too are confusing the terminology.

<Authors’ response:> We have erased Raymond and have also cut some other references. The table is shorter now.

4) Is the article well written?

In the concluding section, ANT and social construction (scot, etc) is put on parity, while in fact they are not. The authors need to show that they are aware of the difference. Next, the emphasis the authors give to discourse, i,e, ” […] the role of words in this realities construction process”, suggests that in fact they are not so steeped in ANT after all. And since this theory does so little work for them in their argument, they might better drop it and stick to ethnomethodology.

<Authors’ response:> We have erased the references to scot and ANT. We agree that they are different, but given that they are no more in the essay we believe this point is addressed.

5) Are there portions of the article that you recommend to be shortened, excised or expanded?

If the article drops ANT and qualifies its critique of the flawed academic discourse, it leaves us with a valid critique of some authors and an interesting case study, but not much of a generalised argument. Hence the authors would need to develop some other claim to make the article worthwhile. I have a suggestion: Having criticised the academic discourse, the authors should not stop with the same binary non-commercial/commercial. After having concluded that FOSS is not just against commercial tout court, it remains to be shown what nuances of negotiating the ways to be commercial that is going on within FOSS projects (the line between free and open, the cases refered to in the paper is likely to be full of such examples) The authors can have a look at O’Mahony’s article on ”guarding the commons” as an example of a nuanced academic highlighting the nuances of non-commercial/commercial within FOSS projects.

<Authors’ response:> Thanks for these suggestion.

We dropped ANT.

We have alredy published about negotiations in other essays. See for instance http://firstmonday.org/htbin/cgiwrap/bin/ojs/index.php/fm/article/view/2064/2030
http://www.sciencedirect.com/science/article/pii/S0198971510000852

O’Mahony has been added and was helpful to clarify how communities manage to protect their work. Thanks.

Reviewer: George Michaelides

1) Is the subject matter relevant?

Yes.

2) Is the treatment of the subject matter intellectually interesting? Are there citations of bodies of literature you think are essential to which the author has not referred?

Yes. The treatment is intellectually interesting and the literature review is adequate. However the most important document in this discussion is the open source definition and this has not been utilised. See my 2nd argument in the “additional issues” below.

<Authors’ response:> Thanks for pointing this out. We have added proper references to the Open Source Definition.

3) Are there any noticeable problems with the author’s mean of validating assumptions or making judgements?

Yes. The main problem is that the authors argue that the ethnomethods that emerge in FLOSS communities imply that FLOSS can also be commercial. However in my opinion this is not a matter of the shared schemata of FLOSS developers being different from shared schemata of academics. It is simply a matter of “a few” academics either (A) misinterpreting the FLOSS licenses or (B) misusing the word commercial as a synonym of proprietary.

<Authors’ response:> Thanks. We believe that this comment does not undermine our position with the paper.

Furthermore, some of the examples of misuse that the authors provide are not all really misuse:

e.g. quote by Bird et al (2009) on page 6 does not imply that commercial is the antonym of FLOSS. It simply says that they studied distributed development in one commercial ENTITY.

<Authors’ response:> We have erased this reference and a few others.

4) Is the article well written?

Yes. Can be improved by being more explanatory in the introduction what is the potential contribution of the paper.

<Authors’ response:> We have improved the introduction and better explained the contribution of the essay. The text had also been revised by a proof-reader.

Can also be improved by removing jargon and acronyms not related to the central thesis of the paper or not used very frequently in the paper (e.g. STS, ANT)

<Authors’ response:> Indeed, you are right. We have removed most of the jargon and references to ANT. The paper is definitely better now.

Also the whole article is very verbose and the authors can make their point with a much reduced version of this paper

<Authors’ response:> We have shortened the essay by almost 2000 words. Also the essay has been revised by a proof reader.

5) Are there portions of the article that you recommend to be shortened, excised or expanded?

Yes.
P8. The fact that this paper comes from 2 doctoral dissertation and their description is irrelevant to the paper

<Authors’ response:> We did cut this part.

Section 3. The whole section provides a lot of details about the case that are not necessary to make the argument. Similarly a lot of the examples from online discussions are not necessary and the authors could simply summarise the issue regarding OpenDWG and only provide the relevant pieces of the discussion. The section starts at page 14 but it only gets to the point at page 20-21
Overall the paper can be reduced substantially to make it more readable (which will in my opinion also strengthen the argument)

<Authors’ response:> We reduced very much this part, which is now shorter and clear.

Additional issues:
1. The word commercial is used in different ways throughout the paper which makes the whole argument confusing and weaker. Among others the word has being used in different places to describe commercial entities (as in for profit organizations), their contributions to communities, commercialization (as in monetization) of the product.

<Authors’ response:> We have clarified this aspect in several part of the essay. It should be clearer now.

2. Is this really a matter of how the communities and the developers view their products or this simply a matter of understanding the licensing or the Open Source Definition?
For example, clause 6 of the open source definition (No Discrimination Against Fields of Endeavor) explicitely states that:
“The license must not restrict anyone from making use of the program in a specific field of endeavor. For example, it may not restrict the program from being used in a business, or from being used for genetic research.”

<Authors’ response:> The matter is the differences between stakeholders and academics discourses.

3. Need to reformat the reference section in a consistent way. For example some papers follow the format of
Surname, Firstname, other are F., Surname yet others are Surname, F.

<Authors’ response:> Thanks. Done.

4. There are many ways in which something can be said to be commercial. I would really like to see in this paper a table or list of some examples of how this happened in different communities. Based on point 2 all licenses (and by extension communities) should allow for commercial participants. What are some examples of this. Similarly what are some examples of projects being used for commercial purposes and how e.g. by providing proprietry add-ons (Revolution R), dual licensing (MySQL), providing support (e.g. Red Hat, Apache, etc), or by selling hardware leveraged by open source software (e.g. Samsung Android smartphones).

<Authors’ response:> We mention throughout the paper several examples: the GNU/Linux distributions, Nexenta, Firefox. And so forth. We believe this will take much more space.