Reviewer: Nate Tkacz

1) Is the subject matter relevant?

While this essay is explicitly on the topic of P2P architecture and design, I do not think that it is suitable for Critical Studies in Peer Production. The reason is due to the article’s level of generality and the nature of the argument.

The stated premise of the essay is to make a methodological case for the importance of humanites scholars and social scientists to take architectures seriously. In itself this is a valid point, but on its own it is not enough to sustain an entire essay, nor does it made a novel contribution to existing debates. The essay’s main point is already well taken and underpins a plethora of existing studies. The reason that I think the essay is not suitable for JoPP is because the journal itself is already based on this premise and thus, restating the argument to its readership would not be of great value. That is, JoPP is about investigating exactly these relations between architectures and emergent modes of sociality, and thus to make a general argument about the value of such activities is preaching to the converted. Therefore, I think the essay would be better placed in a journal where such ideas are less common.

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?

The treatment of the subject matter is competent, but I do not think readers of the journal will find the content interesting.

The author mentions much of the relevant literature, but this could be better handled. The first section in particular needs to be better organised and is missing some key reference points – most notably anthropoligcal works (Kelty; Coleman), and works from software/hardware studies (Galloway; Kirschenbaum; Chun; Hayles; Parikka; Bogost and Montfort; Soderberg). Some of these are mentioned in passing, but they need greater attention as they already significantly extend the arguments made by the author.

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

No.

4) Is the article well written?

Generally, the article is well written.

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

If the author was to pursue publication with JoPP, I suggest greatly condensing the first sections and greatly extending the final section. That is, the article does not need to make a lengthy case for the value of studying P2P architectures and instead, should actually do it (as it begins to, in a very truncated way, in the final two pages). This would allow the author to make some novel insights about the “plumbing” and engage in a more sophisticated way with recent developments in these debates.

Reviewer: Vincenzo D’Andrea

1) Is the subject matter relevant?

The topic of this paper is certainly relevant to the journal audience, pointing the readers’ attention to a relevant perspective for social scientists studying P2P (or even, more broadly, phenomena related to the Internet).

2) Is the treatment of the subject matter intellectually interesting?

I’m uncertain on how I would classify the paper: it calls for a new methodological approach, but it is not a methodological paper because methodology is hinted and suggested rather than defined; it refers to several empirical studies, but it is not an empirical paper because there is no connection to a field research; it looks like a position paper, but there is a huge effort for showing that the proposal is so well founded that it looks not disruptive as I would expect from position papers. I understand the author’s decision not to describe empirical work (and not to refer to her/his work for preserving anonymity), so what I would like to see is a better definition of the proposed methodology. In other words, the author does a very good job in providing several convincing references supporting his/her approach (that is, to consider the infrastructure / plumbing / architecture) but at the end of the paper I felt it was little more than a bibliographical review. Supposing there is a reader that was not already convinced of the importance of caring about the plumbing, now I have read the paper and understand the issue and feel I should change my approach to field work but have no clues how…

Are there citations of bodies of literature you think are essential to which the author has not referred?

The literature presented in the paper is quite extensive and (for what I could check) related to the subject. On the other hand, in some parts of the paper I had the impression of a long list of related but not-too-connected references (Section 2.2 for instance). I think that there are so many authors that have supported the need for “caring about the plumbing” (great title by the way!) that the author should do a better effort in helping readers in navigating that literature. I was tempted to suggest several more references (several papers in Participatory Design literature, some STS, several paper published by tripleC – including the recent issue with paper on P2P), but the main point is not what is missing but rather the map of what is there and the motivation connecting what is presented with the paper focus.

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

No.

4) Is the article well written?

So, the main improvement to the paper would be to state in greater detail what is the author’s take on how to actually take into account the architecture: how can we do “a sociology of networks that is not afraid of its subject of study ” (differently from what has already been suggester by others)?. This said, the paper is generally well written and I had no problem with the language and the overall organization.

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

The only part I found both redundant (for JoPP) and not too connected with the main point of the paper is the part of Section 4 and 4.1 that argues for legitimate use of P2P and the need to avoid confusion when studying it.

I add more comments on what seems the main effort of the paper, that is the discussion of the relevant literature.

1.A revision of some references would be useful. I think the referred works could be placed in two broad container: one for authors that directly support the paper perspective (such as Star and STS); the second for work that do not directly support the paper argument but are enrolled as supporters because they implicitly care about plumbing. I would place in this second set Agre, van Schewick, Niva Elkin-Koren, and so on. Some of the works in this second group seem to me to contradict the paper’s point. For instance, Agre argues that “Architectures and institutions often  are related, and systems analysts work  largely by translating institutional  concepts into system architectures.” This seems a rather deterministic view of the relationship between technology and politics, not supporting an intertwined view of the relationship between plumbing, plumbers, water,… A similar argument could be raised about van Schewick: I could only access reviews and excerpts of the book, but it seems van Schewick broad goal is to discuss the impact of Internet architecture changes to the economic environment for innovation. Once more a rather specific, directional view of the relationship between architecture of the internet and social studies.
2.Some references Section 2.1 could be revised. The reference to Schollmeier is a two pages poster (maybe a bit overestimated for “a great deal of attention”, even if it is from the First Conference on P2P Computing). The reference to Oram is a bit broad, being connected to eight different chapter, each by different authors, presenting P2P applications (not specifically related to just “optimization for specific purposes”).
3.Given the careful work on providing a complete picture of the literature, it is somewhat surprising to see no supporting references for claims such as the ones in the opening paragraph of Section 3 and of Section 3.2.
4.Minor points: Cardon 2009 in Section 2.1 should be 2008; Elkin-Koren 2006, cited in Section 3.2, starts from page 15, but the reference points to pages 6-9; the reference to Arbor Networks 2009 is missing; DeNardis 2009  and the works by Strahilevitz are not cited in the text.
—end—

Reviewer: Matthijs den Besten

The paper has two inter-twined goals. The first, explicitly stated in the abstract, is to show that it is important to understand the technological structure of a device in order to appreciate its potential social impact. The second, emerging between the lines, is to advocate a more positive attitude toward peer-to-peer applications (P2P). As it is, the reasoning seems rather circular: we appreciate P2P since we understand the social implications of its architecture and our appreciation is proof of the importance of this link.

As it is, the structure of the paper is rather convoluted. In order to straighten in out, it might be beneficial to focus on what seems to be the central research question: why is the client-server architecture still dominant despite the benefits of peer-to-peer? More specifically still, why has Napster been replaced with Deezer?

The paper repeatedly makes the point that peer-to-peer technology is unfairly reduced to illegal file sharing in popular opinion. Would P2P have been able to realize its potential as disruptive technology if people would have cared about its architecture? I am not sure.

Van Schewick (2010), from which the paper quotes elements of the introduction and conclusion, writes quite extensively on P2P. Her focus, however, is on the technological impediments to P2P. “In the 1990s,” she writes, “various aspects of the Internet’s architecture were optimized for the needs of a particular group of applications called client-server applications, which where the dominant class of applications at the time. […] Given that most of the applications that were known at the time followed this pattern, network providers designed the access links (the links connecting the home to the network provider) of their broadband networks to provide asymmetric bandwidth. […] Today, asymmetric bandwidth creates problems for applications that send and receive an equal amount of data; these applications are called peer-to-peer applications” (p. 70). Furthermore, asymmetric bandwidth is not the only problem that P2P faces. Also other aspects of the Internet infrastructure make life difficult. “For example, peer-to-peer applications […] may have difficulties passing through NAT’s or firewalls” (van Schewick, p. 386). Finally, P2P have faced active resistance from network providers. See van Schewick (p. 261) on Comcast’s interference with BitTorrent. The motivation for this interference, however, does not seem to have been the use of BitTorrent for illegal file sharing, or otherwise Comcast would not have tried to hide its interference from its customers (see also p. 351). The issue, rather, seems to have been the cost of P2P to the provider: “Peer-to-peer file-sharing applications challenge traditional assumptions about use: they often send and receive continuous streams of data over longer periods of time (violating the assumption that traffic will be bursty, discontinuous and asymmetric), and they often download or upload data to and from computers on other networks (increasing the network provider’s interconnection charges)”(van Schewick, p. 266). To me this sounds like path dependency (David, 1985). To make an analogy: it is not because people thought that Dvorak was a communist that they stuck to the QWERTY keyboard.

When van Schewick talks about architecture as an “alternative way of influencing economics systems” (p. 3), she talks about network architectures as a whole and in particular deviations from the end-to-end architecture (see Chapter 7). This is not the same as pitting client-server applications against peer-to-peer applications: “Though differences in location, in ownership, or in control may be relevant […], they do not matter for the broad version of the end-to-end arguments […] Similarly, some assume that the broad version […] favors peer-to-peer applications. […] Again, these differences are not relevant” (p. 110). What matters, however, is whether design choices that have been made in favor of one type of applications at the detriment of others: “The broad version of the end-to-end arguments requires the lower layers of the network to be very general”, but in violation of this principle network engineers have moved to “optimize the performance of the network or save costs in the short term while disregarding the effects of their actions on the long-term evolvability”, and P2P has been one of the victims of this move (p. 286).

Now, are there reasons to expect that peer-to-peer applications are better suited than client-server applications for knowledge sharing, collaborative work and social networking, as the paper states? With respect to this, the paper seems to fall into the trap that consists of juxtaposing architecture to the stories of institutions, individuals and groups, which it warns against itself. The client-server model seems to work quite well for Wikipedia and Github. How would the communities that are organized around these sites be more productive if they would stick to a pure peer-to-peer model?

In short, I suggest that the paper be revised in order to focus more clearly on one line of argument and to engage more seriously with its sources, van Schewick’s book in particular. In that way the originality and the relevance of the points made by the paper will be able to show through more clearly as well.