Thursday, December 13, 2007

Alfred North Whitehead meets an anthropologist of science and technology.

So perhaps I will start to do some blogging on Whitehead. Here is a first try:

I know, it's something like a love story. Fifteen years ago I was hooked on Gregory Bateson. In between it was Donna Haraway, Bruno Latour and «ANT», Actor-Network theory. Some said that I did ANT too devotedly. And then, the feelings for Whitehead is much more warm blooded than it ever was for Haraway and Latour. So I am possibly even more lost now than I was then?

I don’t know, I think it I am more playful with Whitehead than I was with Latour. It is more fun with Whitehead, in it’s totally incomprehensible beauty, than it was with ANT. Perhaps we – or I – was too serious about ANT? Perhaps we made it too much into the Solution? A Serious commitment to fight the Evil of «Flat Realism» and «Social Constructivism»?

Anyway, Whitehead is no less important to the well being of the many creatures of earthly existence than ANT ever was. On the contrary. The good thing about his philosophy is that it is a sweet medicine, even if it took me half a year of hard, meticulous reading to sense its sweetness. And that was just to get through «Science and the Modern World».

Wednesday, January 24, 2007

___________POWERS AND REPOSITORIES___________

Ubuntu and Debian

This is a paper that I have also published as a blog post – to allow for comments. But it is 14 pages long, so you may want to print out the pdf-version. You might also want to jump right to the comments.

Introduction: The authority and power of software repositories

This paper deals with the authority and power of software repositories, especially Launchpad and Rosetta, the repositories that contains the source code and translations of Ubuntu. It does this by extending some of sociologist Max Weber's notions of authority. Weber's idea of “charismatic authority” is well known to many free software developers, and to those who comment upon and study this phenomenon. The informal, personally achieved authority of the charismatic leader, like that of Linus Torvalds, has been well known since Eric Raymond wrote his papers The Cathedral and the Bazaar 1 and Homesteading the Noosphere.2 Eric Raymond's message is substantiated by a detailed empirical account and states that Free software projects stick together, rather than to fork, pretty much because this authority is respected. Raymond has made some very fine observations.3 But they are not enough. In the project that this paper is part of, the focus is on the two other, well known authorities of Weber, the traditional and the bureaucratic-rational one.

The basic idea of this paper-to-be is that repositories are traditional and bureaucratic-rational authorities, and that they do important coordination and integration work in free software-projects. As long, that is, as they are respected by their users.

By the term “tradition” I don't think of something from which one could, should or inevitably will emancipate oneself (in a modern progressive sense). I use the word “tradition” in a wider, more in communication-theoretical way. When a population share a language, this, following language philosopher David Lewis, means that there is a convention in the given population of truthfulness and trust in that language4 (Lewis 19755). To respect and be a part of such a convention of truthfulness and trust means to respect the authority of a tradition, the tradition of that language. And to share a convention of truthfulness and trust in a language means to respect the standards – the rationality, the protocols and bureaucracy – of that language. A note on words can here be in place: A “repository-in-use” is a bureaucracy because it is a particular kind of rationality which is partly externalised into formal and mechanical procedures. A “repository-in-itself” is nothing but a dead machine. A repository-in-use is a bureaucracy and it is made up of some actions, norms and representations that are internalised by its users, and some actions, norms and representations that are externalised into the machine. When I speak about “repositories” in this paper, I always mean “repositories-in-use”, unless otherwise stated.

These repositories, then, share some important features with languages. As languages, they are traditions and rationalities, and by being so they make communication possible.

It seems to me that an important problem that some (or many?) Debian Developers and Debianites in general have with Ubuntu is related to the way in which Ubuntu challenge the authority and tradition of Debian. Thus, Ubuntu pose a treat of making communication impossible. The treat of non-communication, the treat of forking.

Now it might be argued that the Debian fear of Ubuntu is precisely just that; fear, given by some of the Seven Deadly Sins, like envy, pride or greed. This is speculative. I don't know much about people's motives and I don't want to speculate about them. Rather I would like to take the arguments against Debian seriously, not to uncritically buy into them, but neither to dismiss them at hand.

The rest of this paper is mainly empirical. It first presents some of the criticism against Ubuntu and Launchpad. Focusing on Rosetta in particular I then go on to explore that criticism in more detail. I partly do this by drawing on my own experience with translating KDE into Norwegian.

Launchpad: Evil or just Adding Value (tm)?

Now, you all know that the appearance of Ubuntu stirred up some feelings in parts of the Debian community. One of the major starting points of the emerging conflict was an interview and a blog entry by Ian Murdock, spring 2005. This quote pretty much sums up Murdock's concern:

[Ubunbu is] diverged so far from Sarge that packages built for Ubuntu often don’t work on Sarge. And given the momentum behind Ubuntu, more and more packages are being built like this. The result is a potential compatibility nightmare.

Mark, this doesn’t end well. If you want a glimpse of what will happen, take a look at the RPM world, where software developers and ISVs have to build a different RPM for every RPM-based distro (either that, or the ISVs have to choose the one or two most popular RPM-based distros to the exclusion of all others–or perhaps that’s what you have in mind?). (Murdock at

Several debates on Debian-devel followed suit, with this tread as probably the most famous one: Is Ubuntu a Debian derivative or is it a fork?6 This tread itself forked into multiplicities, but some of its important branches discussed the distinction between “fork” and “derivative” with examples of how Ubuntu could, should or actually did feed its changes back into the Debian repositories. Another major thread discuss the way in which Ubuntu gives proper credit to Debian developers, see the tread Ubuntu and its "appropriation" of Debian maintainers. At the same time (May – June 2005) the thread Canonical and Debian discussed, to put it very briefly, the power of Canonical versus that of the Debian elite (known with slightly self-ironic paranoia as “the Cabal”).

Later the same summer, at DebConf5, in Helsinki, summer 2005, Mark Shuttleworth and several others did a lot of work to try to reconcile the parts and resolve the conflict, for example in the two plenary sessions “The Ubuntu Talk”, held by Shuttleworth, and “Debian Derivatives”, a plenary chaired by the Debian/Ubuntu developer Benjamin Mako Hill. Very briefly summarised, Mark and other Ubuntu/Debian developers worked quite persistently to establish a legitimate position for Ubuntu, in the very heterogeneous and partly connected landscape that makes up Debian as a totality.

Talking to Debian developers towards the end of DebConf5 I still met a lot of scepticism towards Ubuntu and Canonical, and the scepticism was not directed towards Ubuntu per se, but most often to the way in which its repository, Launchpad, was organised, and the way in which it did, or did not communicate its changes back to the Debian and upstream repositories. People were also sceptical to the fact that Launchpad is not Free or Open Source software. “I would never submit my code”, a developer told me, sharing taxi to the airport after the conference, “to a repository where I have no insight into the flow of information inside it”.

Now, Christian Perrier, who is central in the internationalisation of Debian, is one of the Debian developers who has attended the Ubuntu-Debian discussions with a constructive attitude, working quite persistently, it seems to me, to bridge conflicts rather than to add fuel to the fire (see for example this thread7). However, in March 2006 he is making an emotionally engaged, slightly frustrated and clear statement on the importance of the non-free-ness of Launchpad, and its language tool, Rosetta, in particular:

Rosetta, langpacks... Jordi, we know that Rosetta is good (though Javier raised interesting concerns about it and more generally translation portals). But, eh, please spend some time convincing Mark that releasing the damn thing as free software would be the best Canonical could do to the community. I know you guys are probably nagging him nearly daily with this but, believe me, that would be the best that Canonical could do for i18n. Mark, listening? :-) (Perrier, Tue, 28 Mar 2006, at his blog, Bubulle's weblog8)

Perrier must have met many sceptics, like the one I shared taxi with, most certainly may more than I have.

There have also been a lot of blogging activity around the Debian-Ubuntu issue, and Launchpad/Rosetta in particular. Without having the slightest chance to cover the multitude of this writing, I will present an argument made in two clear blog entries from the time around DebConf5. Both blogs are written by Debian developers.

Sunday, July 17, the last day of DebConf5 in Finland, Debian developer Joachim Breitner posted a blog-entry called Launchpad, Google and why Microsoft is not the problem. I site it at length, to show the full argument:

I just came out of Scott's talk on revision control and package management. He also talked about Launchpad, a project by Mark Shuttleworth's Canonical Ltd. Launchpad collects, as far as I can tell, all (relevant) Free Software in source format from everywhere including the complete history, as well as all modifications done by the distributions, and puts it in a single revision control system. Supposedly, this has happened to a certain extend already, and fills up 1700 Gigabyte on their drives. This can be accessed then, for example, by the Debian maintainer to very easily integrate the changes that the Red Hat maintainer did to his sources, and the other way around, and so on. The technical usefulness is almost unlimited, and it seems as if this might be a service to the Free Software community larger than and freshmeat combined.

So why would I blog about it? Because this is as dangerous as it is useful. What Canonical is trying to achieve here is to collect all the work that the free software have made and will made, and put them under their control. Sure, it is all accessible, sure, the code might even be free software someday (where someday is probably the moment that launchpad is so established that competitors, even with the same software, have no chance of establishing themselves), but the data is in Canonical's hand, and by this, we all depend on Canonical to continue providing the service. [...]

And why is this more dangerous than Microsoft? Microsoft did a pretty clever thing: By good marketing, clever programming and commercial pressure, they locked in a lot of people to use Microsoft products and formats. This is evil, but we have found a solution to that: reverse engineering and Free Software. Microsoft can be coped with. But Google and Launchpad were even cleverer. Instead of locking something in, they open everything: It is free, it has nice usable APIs to integrate in applications, and they suck all that opened data in and keep ahold of it. And while Microsoft has a record of doing a bad job when it comes to technical quality, thus helps the alternative, non-evil, Free Software, Google, and probably Launchpad, excels in what they doing, thus reducing the immediate need for alternatives altogether. (See: Launchpad, Google and why Microsoft is not the problem9)

It seems that Canonical with their Launchpad, just as Google, may be able to “control” information which is freely available to all, just by organizing a whole lot of it. And then, apparently, there is the problem that Launchpad is not free software, and thus hides the way in which it works.

Mark Shuttleworth responded to this blog entry, in the commentary section of Breitner's blog. Her is a part of his reply:

This is a well-written blog post, I can't fault your logic, and I understand where your concerns come from. I can only say that I hope, in the fullness of time, you'll be very happy with the way we handle Launchpad.

Over time, it will be open sourced. Right now we compete with Progeny and Red Hat and other companies, so we need to have a unique offering to do so effectively, and that's Launchpad. There are already libraries and tools in LP that we have open sourced on request, especially in Rosetta, the translation infrastructure. ... (Shuttleworth at Breitner's blog)

Another Debian Developer, Anthony Towns, commented on this remark by Shuttleworth:

... to me, the only logical conclusion to that is that LaunchPad will be free when Canonical/Ubuntu are the only players in the market, or when Canonical’s current business model fails and they switch to a different one. Which is fine: if you write some software from scratch, it’s your choice what you do with it; but unless you’re an underpants gnome or a slashdot commenter, the above doesn’t qualify as a “plan” to free LaunchPad. (Towns at indolence log)

While another active hacker, Dick Davies, commented:

So have I got this right - Google (and Launchpad, although I've no experience with that) are evil because they don't suck?

They're just Adding Value (tm) to Free stuff and choosing not to make themselves free. That's not evil, that's their choice, surely? (Dick Davies at Breitner's blog)

Now, I cannot judge from these entries who is right and who is wrong in this debate, or indeed if there is any real difference in the positions presented above. Because Shuttleworth is going a long way to give Breitner right in his analysis, and basically just legitimatises his plan with reference, in the latter part of his entry, to their common enemy:

.... One thing I can say, though, is that a web service (or even a remote app service) can never create the same level of pain that a proprietary OS can do. Having watched what Microsoft has done, I'm largely motivated by a desire to ensure that countries like South Africa never have to pay a tax like that again. (Shuttleworth at Breitner's blog)

As an anthropologist and an outsider I could be content with presenting the different positions with respect to Launchpad and the Ubuntu-Debian relationship, and abstain from taking side in the difference, or from any deeper understanding of the conflict. I am not quite content with such a view from a distance. Rather I want to dig a bit deeper, to see if I can get a better understanding of the stakes. Is Launchpad, and Rosetta in particular, “evil” or is it just “Adding Value (tm)”? I have not found any clear cut answer to this question, and I am not claiming that there is an easy, final or universal answer to it. However, by studying and taking part in the translation of KDE to Norwegian I have reached something like a tentative, partial and temporary “conclusion”. That conclusion, to put it very briefly, is that there is something ... (fishy?) about Launchpad. Now, here is the story that has led me to this ad hoc and provisional conclusion.

Rosetta and SVN – a Norwegian case

The translation of KDE into Norwegian is administered by a project called “Skolelinux”, also known as Debian-Edu. Notably, this administration consists of two things: Skolelinux maintains the SVN-repository of the language files, and they arrange a series of gatherings. There has been 35 gatherings since early 2002, which gives an average of seven a year. They have to a large degree been held during a weekend, often at schools that run Skolelinux, mostly in Norway but increasingly in Germany and France. At most of these gatherings several translators have met, and worked with translating the small strings of text from English to the two versions of written Norwegian. Skolelinux considers its SVN-repository of Norwegian KDE as the upstream site for these language files, and it has been generally accepted as such.

In May 2005 Mark Shuttleworth attended a Skolelinux gathering, held in Bergen, Norway. He presented the idea of Rosetta to the gathering. Petter Reinholdtsen asked Mark how to solve the problem of feeding translations done in Rosetta back upstream. Mark had no answer to this question. Not long after Rosetta version 1.0 was released. (And at that time accusations of the following kind already circulated: “Rosetta is evil and eats babies”.)

A year later I decided to try out Rosetta in some more detail, to try to see how it worked, and how it worked together with the Skolelinux upstream repository. So I did this little experiment:

4. August 2006 I translate 6 strings of K3B (of 40 untranslated) in Kubuntu 6.06, using Rosetta. I am invited to do the translation with this warning: “You are not an official translator for this file. You can still make suggestions, and your translations will be stored and reviewed for acceptance later by the designated translators”. Two months later I attend a Skolelinux gathering, and continue to translate k3b. This time I download the files from the SVN-repository of Skolelinux. The 40 untranslated strings present in Rosetta repository of Kubuntu Dapper Drake are still untranslated, including my six untranslated ones, but there are 42 untranslated strings in the Skolelinux repository10. At Rosetta my six translations are still awaiting an approval from a registered contributor. I translate the 42 untranslated stings, using Kbabel, an check the changes back into the repository. Three months later (8. January 2007) there are still 40 untranslated stings in Rosetta's repository of K3B (both in the 6.06 and the 6.10 release).

So, there was no communication between the repositories during this half-a-year, and my work in Rosetta was wasted. I was invited by Rosetta to do the translation, and I was not informed by any text or link that there existed another translation group and another repository – a repository, that is, that considers itself as “upstream” of this translation. I was invited to take part in the translation, and I wasted my time in the attempt.

Now, in the specific case of translating Ubuntu and Kubuntu into Norwegian the story is in practice less harmful than I have made it look like above. Norway is a small country, with only 4.5 million inhabitants. This means that the community of free software translators are also quite small, and that people tend to know each other. Among those who have started to translate Ubuntu into Norwegian there are several people who have also taken part in translating Skolelinux. Thus these two communities know about each other very well. (E.g.: The administrator of “Team Ubuntu Norwegian Translators”, Karianne Fog Heen, is married with Debian Developer Tollef Fog Heen, who took part in the early development of Skolelinux.) In an email exchange at the Email list of Skolelinux, Karianne Fog Heen agreed with people from Skolelinux to leave KDE to Skolelinux, to let the Ubuntu-team concentrate on GNOME, and to coordinate translations at common Email-lists.11 Of course, no-one wants to do “evil”, everyone knows that it is better for all if double work can be avoided, and the Ubuntu/GNOME and Skolelinux/KDE translators don't want anything but a set of consistent and effective translations of these free software desktops into Norwegian.12

Obviously, it would not take long before a Norwegian newbie – someone who did like me – to run into the Norwegian “Ubuntu translation page”, probably before getting to Rosetta, and read about the translation effort that also takes place with Skolelinux. Thus, a set of social relations and web pages outside of Rosetta would most certainly prevent an unwanted fork of the kind that I provoked in my little experiment. The point, however, is this: The web, in the small community of Norwegian translators, that prevents this forking exists outside Rosetta and Launchpad. Rosetta itself does nothing to prevent a fork. There is no systematic feedback from Rosetta to the upstream translation. No system or systematic attempt of information that prevents double work. No system for merging work done at different places, and for resolving conflicts if there is double work that cannot be merged automatically. Rosetta acts and exists as if it is independent of upstream.

Some other cases of Rosetta trouble

I have found recent traces of how the relative independence of Rosetta may cause problems to other translations.

Matthew East is concerned about the Dutch translation of GNOME, and recently wrote to the ubuntu-translators list:

> An interesting discussion on the #launchpad irc channel yesterday has been making me think about the question of quality assurance for Ubuntu translation teams. To put the question into context, this is how the discussion arose:

An upstream GNOME translator for the Dutch language mentioned that often there are complaints on their mailing list about the quality of translations of the GNOME desktop environment for Ubuntu. He says that these flame wars are particularly irritating given that the translations complained of are not made by GNOME translators at all, but are introduced by the Ubuntu translation team, overwriting the upstream translations. One of the Ubuntu translation team joined the conversation, and it became clear that there was very little QA going on to ensure that the members of the Dutch team are (a) good translators, and (b) familiar with the GNOME and other upstream translation guidelines. (Quoted by Carlos Perelló Marín, 31 Oct 2006, at

Marín, who is involved in developing Rosetta, confirms that the problems raised by the Dutch translators are exactly the same in the Spanish translation work of Ubuntu. A German translator, Sebastian Heinlein, raises a similar problem to the launchpad-users list:

Hi, I am a member of the German translator team.

In the last week before release there was some noise about KDE upstream string that have been overwritten by some translators in Rosetta. To avoid this situation in the future I would like to make the following feature request.

Upstream strings are of a quite high value:

Firstly they are often of a higher quality compared to our Rosetta translations - regarding level of review and consistency.

Secondly upstream feels disrespected if we change their work without any communication. Currently we don't even see how much was changed in Rosetta. The German team has got about 250 members, so quality assurance is quite a hard business.

So I would like to see a feature in Rosetta, that would allow us to easily select translations that have been changed in Rosetta. A corresponding option could be added to the show combobox of the editor. Furthermore it would be nice to see the percentage of changed upstream strings in the progress bar or in a statistical table. (June 2. 2006, at

In these two cases the upstream translation is overwritten, “forked” one may say, by people who may not even know that there is such a thing as an “upstream” to fork off from. Moreover, when as much 250 people take part in the translation of the Ubuntu-version of GNOME we have an indication of another social complexity than in the neat small Norwegian community. We may also note that Heinlein's posting to the launchpad-users-list was called “Request: How we could pay more respect to upstream translation”. It remains unanswered today, half a year later.

I want to postpone to the postscript the question of whether upstream stings are “of better quality” than Rosetta-translated strings – strings that allegedly lack “quality assurance”. First, and more importantly, I want to address this issue: None of the replies to the emails above present any solution to the challenge of feeding Rosetta-translations back upstream. The only solution addressed is the other way around: to make sure that Rosetta is more frequently updated with the most recent upstream. Thus, my conclusion, based on both my experience with the Norwegian translation, and by the recent challenges with the Dutch, Spanish and German translations of Ubuntu, is this: Rosetta works institutionally – systemically if your like – to produce another authority, another tradition and another bureaucratic rationality. That it, it works institutionally to “fork” translations. Rosetta, as a machine by itself, encourage forks. The way they are avoided, at least in the Norwegian case, is by making local “gentleman agreements” outside of Rosetta. In the Norwegian case; translation of KDE has not forked in spite of the possibility of forking that Rosetta enables.

It does seem to me to be one way – perhaps only one way – to make sure that Rosetta translations will not fork from upstream, and that is, quite simply, by turning Rosetta into upstream. I do not know if this solution is intended by Canonical Ltd, but I find it hard to believe that they have not considered this possibility, and this is the simple reason for my vague worry that “there is something ... (fishy?) about Launchpad”.

– * –

It seems like I am confirming Debian Developer Joachim Breitner's worry that Launchpad is equally imperialist in its ambitions as Microsoft and Google. We are back to the unanswered question, then, if this is “evil” or if it is merely “Adding Value (tm)”. I don't know, but Rosetta (as part of Launchpad) seems to be founded on an imperialist ambition. Is Ubuntu becoming big by becoming powerful, and is it becoming powerful (perhaps even attempting so) by becoming big? Is this, as an ambition, a power that implies the loss of power elsewhere, for example in the other sites that now administer upstream repositories? Does this ambition include “taking control”, as Breitner worries, in the same way as Google is “taking control” and as Microsoft took control? Is this power “evil” or is it merely “Adding Value (tm)”? And is it a problem if Debian, in the future, becomes a derivative of Ubuntu, rather than the other way around?

Ps: Why Rosetta is neither evil nor Adding Value (tm)

This section presents a practical and social argument for why Rosetta will not replace SVN-kind-of repositories as an upstream site for translations. Thus this may also be read as an advice to how Rosetta may be made better (given that I am able to make an argument that Canonical have not yet thought of).

Several comments on Rosetta starts by praising the way Rosetta makes translation easy, while at the same time warns that it must not lead to bad translations. I do however not think you can praise the ease of the present version of Rosetta while at the same time ask for better quality. I think in some way it is the very easiness that leads to bad translations, yes perhaps even less translations. I think the Norwegian Skolelinux (Debian-Edu) has been able to produce good quality translations of KDE into two minor small languages because it is using a tool (KBabel) that demands a minimum of initial effort for someone to get started, at least when we look at this tool in its social context. So, let us have a second look at how Skolelinux have worked to translate KDE into Norwegian.

First, we note that a few devoted people have done a lot of work. Gaute Hvoslef Kvalnes did single handedly an almost complete translation of an early version of KDE into New Norwegian, and Axel Boyer have put a lot of effort into making the Bokmal (“Dano-Norwegian”) version. Then a group of between 2 and 8 people have attended the 35 gatherings held the last 5 years. This has not been a stable group. A few of the central translators have attended almost all the gatherings, but many have been doing translation for a year or two, then to be replaced by new people. I have counted all together 28 people that have taken part in the translation work during those five years.13

There is no formal criteria and no formal process to go through in order to become a Skolelinux translator, and to be given write access to the repository. People have just attended a gathering, often by bringing their sleeping bag to some Norwegian school to spend the night on some hard madras in a classroom. In the days they have been translating, chatting and orienting each other. In the evenings they have had a good time at some bar or restaurant. People who translate software may not be very skilled in computer science, so even if setting up KBabel and learning the basic commands of the SVN-repository is easy, it can still be a threshold for some potential translators. However, attending a gathering significantly lower the threshold level. We all know why: Asking the experienced guy next to you how a program work, and having it demonstrated, is infinitely much easier than reading the handbook.

Thus the gatherings have worked as “initiation rituals” for new translators, and probably to new developers as well. Through such a weekend commitment and trustworthiness is both demonstrated and established. A sense of community and belonging is created, including a good reason to keep on translation when the weekend is over. The person – the new as well as the experienced translator, or at least many of them – who return home from such an event and keep on contributing, get a sense of making a contribution not only to an abstract “common good”, but also to a concrete group of people that he or she has come to know. That can be significant, and during the gatherings I have attended I have heard many testimonies of the usefulness of meetings in Real Life.

By comparison, the occasional and accidental user of Ubuntu who drop-in to Rosetta to translate a dozen or two of strings, is most likely to never see his or her translations again, and not to get the sense of belonging which is so important to produce commitment, and thus provide both quality and quantity in the translations.

In addition to this difference of commitment, working with SVN and KBabel – and often doing it at gatherings – enables another kind of quality assurance work than what Rosetta does, namely vertical coordination and quality insurance of strings. Here is an example: I started to translate KOrganizer during a gathering in 2005. By that time earlier versions of if had been partly translated by 9 different people, since 1998,14 and I found a very inconsistent translation. Notably the terms “event” “to-do” and “item” (the last one generic of the two first), as well as derivatives of them, where both mixed and variously translated. This was serious. These are the most central terms in KOrganizer, and using them inconsistently (for example on a function button and its help text) will create confusion, and most certainly anger, with new users. I decided to sort out the mess. That took the weekend. And it required me to search vertically through all the .pot-strings, for the English original terms (“vertically”, as opposed to “horizontally” within each string). There where 1193 strings in that version of KOrganizer. Paging manually through these strings, ten by ten, as in Rosetta, would have made the job impossible.

When a new version of KOrganizer is released, and some of the strings containing “event”, “to-do” and “item” needs re-translation, the translation may again start to loose consistency. If Rosetta is then used as intended by its technical design (that is, also by “drop-in”-translators), such inconsistency is almost unavoidable. KBabel is certainly not a guarantee for quality, but it does enable work on vertical consistency, and the fact that it takes a minimum of effort to learn how to use it need not be a problem. It can be part of an “initiation rite” that produce both community and commitment, and thus both quality and quantity in the translation work.

At the moment, then, Rosetta seems not to be Adding Value (tm). It is just adding mess. Neither is it evil. It is just bad.

Rosetta may of course improve technically, perhaps to match KBabel in functionality. And people may add social institutions to it, to produce the same kind of community and commitment as that of Skolelinux. (The Italian translation theme of Ubuntu has added such institutions, more akin to Debian-kind of formal commitment procedures than to the sociable gatherings of Skolelinux, see, again, this email:15) Thus the quality that Rosetta delivers may greatly improve. And I am in no doubt that it will improve. As I write this (January 25. 2007) there is a email-tread called “Ubuntu vs. Debian translations” going on , at the Ubuntu-translator list, and Canonical employee Carlos Perelló Marín just replied to someone:

I agree that coordination from Ubuntu with upstream not using Launchpad directly is not perfect, but as Sebastian already pointed, we are working hard to improve it every day. (Carlos Perelló Marín soon to be found at

So, we may assume that Rosetta/Launchpad will really start to “Add Value (tm)” not only to Ubuntu. Will it, then, also become “evil”? Well, that question brings me back to my initial discussion, and to my indeterminate conclusion that “there is something ... about Launchpad”.

Does Launchpad in general and Rosetta in particular embody an ambition of becoming not only useful, but also powerful to its owner – by becoming upstream of more than just Ubuntu? “coordination from Ubuntu with upstream not using Launchpad directly is not perfect”, Marín write. So we better all start to use Launchpad as upstream? And will this potential power be a democratic problem? That is, it is easy to see that such an upstream position may become commercially powerful. But in a market that includes actors such as IBM, Novell, Sun and Red Hat one can hardly be worried about the commercial growth of Canonical. Debian Developer Gunnar Wolf blogs on Ubuntu:

“I'm not [...] blaming them for selling their principles – Of all the commercially-oriented Linux distributions, Ubuntu is clearly the one that stays closer to what I'd like. And it's not by mere chance that it derives and keeps in constant sync (at least until now) with Debian. ...” (See Are Debian people real Free Software zealots)16

So, I guess, we cannot worry about the potential commercial power of Canonical/Ubuntu. But can we worry about undemocratic power?

What do you think?

3He has also made some very unnecessary reductions of Free/Libre/Open-Source Software into socio-biological naivety, but that need not concerns here.

4A convention of truthfulness and trust does not assume that everything said has to be true. Rather, it is a condition for lies.

5Lewis, David 1983 [1975]: ‘Languages and Language’, in Lewis 1983, pp. 163–88 Philosophical Papers, volume I. Oxford: Oxford University Press

6The tread continues in June, see

10Apparently someone has translated two strings using Rosetta. I did not investigate which one.

11See the exchange that starts with this posting (in Norwegian):

12Thus, the Ubuntu-translators use and recommend the language-guidelines of Skolelinux to avoid that similar technical terms get different translations in GNOME and KDE, something that would harm both. (See

13The number is based on the list of people who have signed in to attend the gatherings, see

Saturday, July 30, 2005

WhatTheHack, second day

I got up early today to catch a talk by Yuwei Lin from Vrije Universiteit Amsterdam. I was looking forward to the talk, since several people in Genova suggested I look into her work. Her qualitative approach to the FOSS community uses concepts like social worlds and communities of practice which seem very relevant for my own research. Unfortunately, her talk was canceled without explanation. I then attended a talk about Tor, an anonymity service run by the Free Software Foundation (FSF), which means anybody can use Internet services anonymously, i.e. without revealing their true IP address. As a Tor user, you can eventually choose to appear to be surfing from the location of your choice, bypassing any filtering based on IP origin, such as gaining access to Olympic Games results for Americans, or gaining access to the Republican party web site which is normally blocked for traffic from Europe. This is perahaps controversial in these 'war on terror' times, but it seems authorities and security agencies are willing to tolerate the existence of such a service in order to use it for their own ends. The speaker mentioned in passing that he had presented the system to Norwegian police and they were suitably impressed... I also had high hopes for a lecture on 'Digital Identities and the Power of Hacking' by Stephan Humer, a hacker-turned-sociologist and PhD student from Free University Berlin. However, while the talk might have been appropriate at a sociology conference, it failed to capture the interest of the hackers. The tent was quite full at the start, but emptied rapidly, and especially when the poor sociologist spouted off long lists of references that were meaningless to the hackers. I also think that the talk was held at an abstract level that was alien to this group of hands-on technologists. The feel of the conference today was much more quiet than the first day. I may have to revise my estimate on the number of people to closer to 1000. People are very relaxed and sitting around in small groups around the network switches focused on their laptops. I am a bit surprised that there is nn apparent mobilising of the hackers either in terms of political or technical projects. But maybe that isn't the point of this conference - or maybe the crowd is just to diverse in its goals, beliefs and technologies of choice?

Thursday, July 28, 2005

WhatTheHack, first day

I am now at WhatTheHack, an outdoor hacker conference that traces its roots back to similar conferences that have been arranged every fourth year since 1989. These events have been called 'the Woodstock of the hacker scene'. The whole thing takes place in a field some kilometers north of Eindhoven in the Netherlands. Hackers of all persuasions have put up their tents around four 'circus tents' where the talks take place. In addition there are tents with bars, restaurants or just for hanging out. Most (all?) of the tents are equipped with power and network outlets, and there is supposed to be a WLAN spanning the whole site. For some reason, I am unable to connect wirelessly, so I found a free network cable in a tent that boasts a bar as well as a row of good old-fashioned pinball machines and video games. The conference started with a two hour long welcome speech by Rop Gonggrijp and Emmanuel Goldstein (aka Eric Corley), two 'old-school' hackers who have been active since the 1980s. They clearly wanted to draw the lines back to their pioneering days and wanted to see this conference as part of a hacker culture that goes back to the good old days of 'phone phreaking' and the early computers. This didn't seem to be too controversial among the crowd, but I'm not sured everybody followed when Goldstein wanted to bring in political activism under the hacker umbrella. He drew in the current political situation in the US and brought up issues of surveillance and the Patriot Act as something hackers should actively fight against. I then went to another tent to listen to Biella Coleman talking about a report she has been part of for the US Social Science Resarch Council about the politics of open source adoption, with emphasis on NGOs in the developing world. The bottom line was that those who want to fight for FOSS adoption will have to appeal both to idealism and pragmatism in their approach. In the first case by appealing to ideals of openness and accountability, but also to the pragmatics of 'clams (money), control and choice' that can be obtained by using FOSS software. I quite enjoyed a talk by Arjen Kamphuis on 'Free Software in the Boardroom'. This was a 'tutorial' on how to talk to company board members and big-whigs about FOSS and how to sell the idea to non-technical manager types. Lesson one: Wear a tie! Ties were distributed and we were taught how to tie a perfect Windsor knot... The rest of the advice was basically about hiding your idealism, staying away from technical details and being clear. Board members are like spoilt children with short attention spans, apparently... The atmosphere here is quite relaxed. I guess there is around 2000 people here, but they don't make much out of themselves. Bermudas, sandals and a hacker T-shirt seems to be the attire of choice. People are shuffling between listening to talks, relaxing in their tents or in the hammock area (wired for power and network of course), and having a snack or a beer (or something stronger - I detect more than the occasional whiff of marijuana...). My impression is that this is first and foremost a social event, where people get together and relax.

Friday, July 15, 2005

DebConf05, sixth day, 16. July I talked about dress code the other day, and there is something I have to add to that. Among this plurality of not very dressed up Debian-hackers there are indeed codes of commonality. It is too easy just to celebrate the heterogeneity of this “wild bunch” (as New Scientist called them some years ago). I'm not going to try to draw a complete picture. I'll just mention two codes of commonality. The first one is the cotton T-shirt, and not just any kind of T-shirt that half the planet would tend to wear on a warm day. There's the special kind of geek- or hacker-T-shirt. The most common one at this conference is a Debian T-shirt. It has the Debian-sign on it. It might be the T-shirt we got for this conference, or perhaps a T-shirt from earlier Debian- or Linux-occasions. And then there is the self-ironical geek-T-shirt. There are many of them, like this one: One the chest it is saying <.geek.> (without the dots - I had to add some noice to prevent my blog machine to treat this as a real html-tag), which means “geek begins” in the html-programming language (just as < .i.> without the dots means “italic begins”). On its back it says <./geek.>, which means “geek ends” (again without dots). Within these tags is enclosed the body of a geek. There is no strong normativity connected to the T-shirt. No-one makes a comment when you do not wear it. Your operating system, however, will be commented upon. I booted my laptop at the hack-lab the other day. “Are you using SUSE!!”- “The worst!!” I have installed Linux on my laptop, as dual boot, which means that I may choose between booting Windows or Linux. I did this dual boot installation using the German SUSE-Linux, because at the time (some years ago) I thought SUSE was user-friendly enough for a non-geek like me to be able to do it. Debian, certainly, was not. I don't know what these guys would have said had I fired my Windows, but SUSE was bad enough. They offered me help to install the newly releases Debian 3.1, with codename “Sarge”. I refused but offered them my hard-disk if they installed KDE-Ubuntu (Kubuntu), the new end-user friendly version of Debian. They refused. But I am sure someone will help me to do Kubuntu tonight. (If I need help at all, Ubuntu is not very difficult to install. Except I may still need some help to get the wireless card to run. That is still a hassle under Linux.) No one will hurt you if you don't wear a proper suit and tie at a wedding, but you may expect a comment or two. Its the same with your operating system here at DebConf. Debian is plain, taken for granted normality, and Debian developers wear their operating system. It is a sign, I guess of knowlwdge, skills and connection. But I have to add: They all wear their Debian (on their T-shirts and laptops) because they are passionately and professionally convinced about its technical superiority.

Thursday, July 14, 2005

OSS2005, final day

The OSS2005 conference has just ended (for me anyway, there is a post-conference workshop tomorrow that I won't attend). Today was the PhD symposium, where 10 PhD students (including me) presented their research. The project spanned a wide range of topics and approaches, but I was pleased that there were 2 or 3 other projects that had a sociological, qualitative approach. Chengetai Masango and James Howiston from Syracuse University in New York are doing research on socialization practices in FLOSS teams and coordination of unreliable collaborators, respectively. What was interesting was their qualitative, ethnographical method which is what I will be doing myself. Maurizio Teli from the University of Trento was perhaps the one I had most in common with theoretically. He too bases his research on concepts from STS, actor network theory as well as John Law's notion of 'modes of ordering'. He will be studying the OpenSolaris project. I think we both felt that our own 'mode of ordering' was a bit different from the rest, coming mainly from information science and business/economy backgrounds. Both Maurizio and I were grilled a bit on the generalizability and practical importance of our projects, the selection of 'representative samples' and issues of objectivity and bias in doing participant observation. I felt that my own presentation went reasonably well after the initial nervousness had subsided. Several people came to me afterwards to say that they found my research interesting and wanted to draw my attention to related research. Especially the notion of 'communities of practice' sounded relevant and I will look more into this. After all the presentations we pretended to be 'the European commission' and had to select two projects that deserved further 'funding'. My project got 2 votes, which was actually quite good, since one project focusing on release management in FLOSS projects got almost all the votes... All in all it was a good conference that clearly showed the breadth of research being done on FLOSS these days. A few themes emerged: No one is much concerned with the motivations of FLOSS developers anymore. The emphasis seems to be on understanding how the collaboration works, how the development process can be improved, and how standard software engineering can learn from FLOSS projects. I leave the conference with a lot of inspiration, several theoretical threads to follow up and some new contacts within the FLOSS research community.

Wednesday, July 13, 2005

OSS2005, Genova

Genova is quite hot, too! Why is it that countries that really need air conditioning never has it while in Norway any lowly office building is fully equipped? But I digress... I'm here for the OSS2005 conference, aka the 1st International Conference on Open Source Systems. Today is the second day of the main event. This is a more traditional academic conference with the occasional industry participant (most notably Microsoft, IBM and even BSA). Most of the people here seem to be researchers within computer science and information science, including a colleague from the University of Oslo, Knut Staring. I have met one or two who are fellow ethnographers/sociologists/anthropologists, but the vast majority seem to be into modelling and quantitive approaches. A lot of papers yesterday dazzled us with utility formulas and graphs from mathematical simulations of open source projects. A Japanese researcher had done a cluster analysis of the 107 different open source licences in existence... Someone else asked the pertinent question: What had these presenters learnt from their models and simulations which they didn't know going in? The keynote speakers were quite good, pointing at the fact that the OS field is developing, becoming more professional, and the struggle is on to stay true to community values at the same time as making money, making the development model more professional and predictable - not to mention the issue of software patents. After the reception and the free drinks, yesterday night was spent with Italian colleagues at a Genoese restaurant, sampling the local delicacies...

DebConf05, Third day, 12. July, 2005

I ended up speaking about (US) Americans the other day. The next day was much more in the name of the global. The sessions was introduced by an Indian, and one of the morning sessions was about Free software in the third world. “The third world”, of course, is an ethnocentric generalization by some others, but this talk was not. We got to hear some very specific stories about how people try to “push” Free Software in Latin America. One challenge, just to take an example that I remember, is the highly political position of Free Software. Some political parties in Latin America, notably left-winged parties, are eager to promote Free Software. Its easy to give them our support, and their case is obviously good: Cheap software and local control of the innovations is clearly what Latin America – as many other countries – need. The problem, however, is that changing governments, political instability and the rhetorical need to produce difference means that the opposition tend to be very much “against” Free Software once one party has declared its passionate support to it. So, as the case have been in Mexico, Governmental support tend to come and go every third year. This makes the necessary long term development of locally adapted Free Software very difficult. But back to DebConf. It is an international conference, and this was more visible the second day. The Germans are as visible as the Americans, and there are Japanese, a few Chinese, some Latin Americans and Europeans of all kinds. I want to talk about the heterogeneity of the gathering, but not particularly about the national or ethnic heterogeneity. There is something about heterogeneity in relation to uniformity and knowledge. I'll talk about dress code, not because I think it is very important, but because it is not entirely unimportant either. There is a liberalness about dress code that I have not seen many other places. Punks or Gothic heavy rockers may be very unconventionally dressed. And many hackers share their unconventionality: But punks and heavy rockers do not diverge much internally. There is nothing as predictable as a heavy rocker. Their dress code tolerates nothing. There are a few “heavy rockers-light” here, a few “cowboy”-kind of Americans, a bit more “hippi-light” people, and some almost boringly conventional people. No dresses, a lot of T-shirts – it's still hot summer. Almost only males. I sat next to some male, mid-30ish people at the hotel the other night. They were probably attending something at a Nokia site (not far away) or something. Talking about some small portable PC-kind of gadget, looking like a large PDA (running Linux!). They where looking casual but smart, and I understood that there are engineers elsewhere that look very different from this untidy bunch here at the “hack lab”. By the way, it's midnight and still there are something like 30 people working with their PCs here, jacked in to a provisionally mess of Internet-wires. Here is a picture: So what is the point I am getting at? It is this, and it is late and I have to go to bed, but this: The hacker community is a “do-ocracy”, a meritocracy. That is how they describe themselves, and they are not much wrong. They are “inside” because of what they know, can and master, not so much because of what they look like. They are what they master. There is something about gender here, because they are almost all males. And if its a strict meritocracy, then it may be that female exclusion is merely a matter of lack of devotion to skills. A women at “Debian Women” I talked to objected to this, and she will give talk in a couple of days. I'll return to the issue. Good night.

Monday, July 11, 2005

DebConf05, First day, 10. July, 2005

So, I am attending Debconf05. Helsinki is warm, but I won't speak about the weather. This is a conference, an international one, and I have attended quite a lot of international conferences. This one is different.

An international conference, to me, has a certain formal style. There is a certain distanced politeness about them, probably due to the fact that many people don't know each other very well. The Artificial Life conferences were sometimes very formal, I suspect much because of the high concentration of Japanese reasearchers. These "hackers" or "geeks", however, ... not that they aren't polite :), certainly not, but they have a very informal style. They do seem to know each other fairly well. Some of them know each other very well, and some of these make up the core of the conference. This core, then, often talk to each other as if they were a group of good friends chatting, which they are, of course, and that makes the whole athmosphere very relaxed. But they also do it in the large plenum-hall, with 200 people listening in. It seems to me that this group to a large degree is made up of US Americans, or at least they tend to take the word in this very informal way. I know, or I think I know, of Europeans that are very central to the Debian project. So, why are the Americans so present in the public, informal space of the talks, making informal comments, being personal? Am I talking about differences in national ethoses here? The difference between Japanese Artificial Life researchers an American geeks fits well with received national differences, such as those that Sharon Traweek describes in her ethnogrphy about high-energy physics. So, perhaps the Americans are more relaxed, than the more formal Europeans -- but then also a bit more dominating, as they sometimes tend to occupy the public space.

Well, sorry for making national sereotypes here, I may be wrong in my analysis, and it certainly is the case that there are quite a few US Americans in the core of Debian, independently of what "the American ethos" may be, or not be.

Nationalities aside then, the fact remains that some people know each other very well at this conference. They are at home. At home in the Debian, and thus in the DebConf05.

I am an outsider, no doubt about that. That's the nature and also the sad fate of doing anthropological fieldwork. I also felt lost when i studied "ALifers" (people doing Artificial Life). Here I feel lost in another way: lost because the others are so wery much at home, at least the most visble part of the others. Lost despite the fact that this group of people are open, inclusive and truly friendly. But they are still exclusive, because they are so extremely skilled and devoted. This is perhaps a more scientific community than any other community I know of. "Scientific" for tree reasons: For their celebration of open knowledge, for their empatic celebration of systematic knowledge, and for their deep identification with being a knower on those terms of knowledge. They are taking and preserving within computer science some of the best sides of science.

Sunday, July 03, 2005

The Bergen Gathering

Imagine a world where IBM is bragging about using $ 100 mill on building Open Source innovation labs in China and Brazil, where those two states have large development programs for rolling out Linux in their state administarations, and where HP is providing complete Open Source systems, where the same firm they develop all their printers in labs only running Linux machines, thus also unofficially supporting the Linux world with top quality drivers for all their printers. But HP does not ship these drivers with the printers (and with official support). They feed them to the distributors, such as SUSE, who then feed them to the end users, free of charge and with no warranty. But it might be that HP is writing their printers drivers for Unix before porting them to Windows and Mac. Isn't that strange? Well, anyway, and to me even mor fantastic: In this environment of coming Big Business, Open Source / Free Software is still also made by enthusisats. A couple of week ago, I slept on the floor in a classroom, thogether with about 20 "hackers", one week end in Bergen, at St. Paul school. People live cheap, travel cheap and eat cheap, in order to afford the gathering of as many people as possible. They hack, translate and talk about free software for a week-end, all for free, and to the benefit for the many schools (including St. Paul) that have installed "SkoleLinux" or similar Linux systems on their computers. Are you worried about the health of political engagement? Your worry should not be general. In a place like the free software there are plenty of engagement -- it is just a bit difficult to see it as it escapes classical definitions of politics. The size of open source / free software is vast. The impurity in terms of actors, intersets, projects and political commitments is equally facsinating. Above, there is an image to capture some of that size and impurity. The selection of actors represented here is arbitrary and makes up just a tiny fraction of a much bigger collage. And here is a bigger one: Collage (big) To give you an idea of the content og the gathering, here is a comment to the event, made by Frode Jemtland at the list
Hi all, and thank you, to everyone who attended the develper gathering in Bergen.

And a special thanks to St. Paul's catholic school. The schools location, in the city center was a great time saver... And also the accommodation at the school made the developers time for travel, to and from, non existent. :)

The school created user accounts for our translators on their production machine, so it took the translators minutes to get up and running. The network was a standard skolelinux net, and ssh, ftp, irc ++ and other protocoles we needed was up from the first minute.

So on behalf of skolelinux/debian-edu comunity, thank you St. Paul's catholic school, this was a very effective gatherings, with great accommodations  :)

(Even thou we had a lot of meetings that, in them self was very interesting, but they stole precious development time. But more about that in another mail. :) )
Next event, to me, is the much bigger "gathering", or conference, DebConf05. See you in Hel...