Kelty (2008) Two Bits, Chapter 7
Kelty, C. 2008. Two Bits: The Cultural Significance of Free Software. Durham, N.C.: Duke University Press.
Chapter 7. Coordinating Collaborations
210 Last of four practices/components of FS dealt with in this chapter: coordination. A lot of hope invested in ‘gift economies’, ‘peer production’, etc.
210 Will look at coordination in 1990s
211 Unique about this coordination is that no goals – favour adaptability over planning. This not same as chaos, rather way of reconciling individual hacking virtuosos and collective coord you need ‘to create and use complex software and networks’.
212 Recursive public not just about people and discourse but also about giving ‘concrete infrastructural form to the means of expression itself’. Geeks see programming and hacking as ‘variants of free speech and freedom of assembly’.
212- From UNIX to Minix to Linux
212 Both Linux and Apache = coordination experiments; both supposed to be fun [on the rewards of practice, see Warde 2005, this blog]; both key to Internet expansion 1990s; 213 represent a global movement of sorts, as branded in 1998-9
213 Linus Torvalds seen as new generation of FS, post Stallman and Raymond. He stressed fun, meaning adaptability over planning. Right time and place to become an “accidental revolutionary”. Not much theorising going on, except perhaps around ‘community’ eg Rheingold [ah yes, community again].
214 late 80s-early 90s a lot of experiment with Internet tools, academic-commercial hybrids epitomised by UNIX
214 never quite clear what meant by coordination, e.g. level of explicitness/implicitness of licensing issues
215 Linux started as student project in Helsinki, not intended as contribution to FS Foundation; thrived cos already in place infrastructure/moral-technical geek order [notice key importance of unsung mailing lists]
216 piggybacked on Minix, itself UNIX based
217 perhaps cos of EMACS past troubles, Torvald didn’t want to reuse any code, didn’t want restrictions
217 Design and Adaptability
Tanenbaum (Minix) usually strawman role in Linux story; old comp-sci prof against young Turk of Torvalds
218 Tanenbaum didn’t want thousands of strangers improve on his Minix: to him useful for teaching
218 In contrast, Torvald no goals. Debate not really showing Tanenbaum’s conservatism, but contrast tween two ways of coordinating and collaborating
219 it seems that Torvald accepted just about any contribution, didn’t make decisions
220 hierarchical system though, but strictly voluntary
220 all that mattered was whether patch (pieces of code) submissions worked or not; again adaptability over planning; 221 evolutionary metaphors often used to explain this, no premeditated design
222 Linux = recursive public of entwined operating and social systems
222 whilst adaptability is all about critique, goals and planning are about negotiation or autocratic decision-making [cf. Manchester School 1966 definition of politics as struggle over public goals]
222- Patch and Vote
2nd example of coordination: Apache Web server and Apache Group
223 this is story of ‘progressive evolution’ of coordinating ‘people and code, patches and votes’.
224 Patching a bit like debugging, ‘but more like a form of ex post facto design’ [reverse engineering?].
225 story of discussion about patches over mailing list, voting innovations, 226 tension individual virtuosity vs. group decisions about developing the same software
227 disagreements over forms of collaborating, presumed forks, etc; list poster 1 ‘modesty’ in beavering away alone, list poster 2 frustrated with this silence and sudden revelation
229- Check Out and Commit
229 Source Code Management systems (SCMs): tools for organising code and people. Shows problem of recursive-depth: FS still free if made with non-free tools?
230 SCMs anyone can check out code, but only some people can ‘commit’ it; 231 used by both Linux and Apache, 232 e.g. Apache voted to elect trusted committers, those with ineffable “good taste” [same story in every field of practice, e.g. journalist ‘nous’ in study by Nottingham Trent ethnographer]
232 Sep 98 fight Linux kernel developers: Torvalds not up to speed, posssible forking of Linux – ‘there is only one Linus’ [egocentric social field; not to be confused with egocentric network]; Torvalds avoided fork after came back from holiday [mensch, wish I could get a holiday, this blog is a slave-driver]
233 Stallman’s ideology vs. Torvalds’ fun/pragmatism
234 settled vs unsettled practices: GPL was stable document, whereas SCMs coordination still in flux
235 like EMACS controversy of 1985 [see earlier chapter], 2005 Bitkeeper controversy Torvalds created own SCM – in common, the issue of how recursively deep the meaning of free. Experiment failed, but lesson learnt about not to use SCM to coordinate people and code. Also that adaptability not matter of genius invention but of ‘critique and response’.
236 both Torvalds and McVoy figuring out limits of FS but differently: bottom-up, nondesign vs. top-down, planned.
236- Coordination is Design
hype about ‘self-organizing systems’ when people don’t know how things work
237 same point about FS coordination reiterated: all about adaptability not planning; debugging can mean lots of things
238 key problem of coordinating: how do you collect and redistribute changes made by contributors?
239 FS finds niche tween spaces of design and debugging
239- Conclusion
Both Linux and Apache are social experiments with technologies, legal tools, governance and coord systems, moral-technical orders
Important cos central to expansion of Internet, which in turn changing how we think about governance, knowledge and power [big claim about the Net, cf. Boellstorff’s claim about Second Life]
As a recursive public, FS proposes and provides alternatives [I’m not clear now whether only FS is a recursive public or Linux, Apache etc as well?]
240 all this to be seen in historically specific not universalist terms
Trackbacks