Column: Polemic research
What may we expect from a research report? Is it possible to be objective, neutral, when doing research? While I was in university majoring in societal history this was one of the fundamental questions. Can any research into social phenomena be without bias? As a blogger and columnist I present opinions, mine. But what if I wanted to report seriously about the strengths and weaknesses of developing open source software. I strongly believe in the classical scientific method: formulate a hypothesis, determine the instruments, collect data, analyze them and see what the results lead to. No matter whether I like what comes up or not.
Unfortunately there is too little research of this type in IT. Commercial and/or ideological interests play too great a part that the conclusions are written beforehand. I do think it is worse when a blogger/columnist gives personal opinion the odium of research. Like ‘The sorry state of open source’ which was published at PlanÃ¨te BÃ©ranger earlier this week.
The 25 html-page piece is more like a j’accuse, a polemic instead of clear and objective analysis. Which is a shame since the issues mentioned do deserve a closer inspection.
They key argument is that the development of Linux has deviated from solid Unix traditions. As a result there is more attention for features instead of functionality, for innovation instead of stability and for quick hacks instead of structural solutions. PB’s examples to support the argument are quite disturbing. It reminds me of my own thoughts when I read about the inclusion of KVM in the Linux kernel, a new technology by a complete outsider.
The drive to become userfriendly is blamed for the deviation from Unix traditions. PB pushes for education over userfriendliness. The lack of proper Unix/Linux training shows itself in the dramatically reduced quality of help and advice in forums and IRC. The last argument I have been heard before, but is it really necessary to go all the way back to the man pages? I agree that online documentation should improve, but the man pages are simply not accessible enough for most endusers.
It seems like -from the viewpoint of PB- that userfriendliness and quality are mutually exclusive. FreeBSD fell into that trap and apart from Slackware most Linux distributions as well. Some releases are okay, but mostly PB encourages to take the example of *BSD.
And this is where the article looses it. OpenBSD and NetBSD are not for the desktopuser. Different target group, different philosophy of development. To start here and judge the (non-)value of Beryl/Compiz and the loss performance because of it is nonsense. PB is full of dÃ©dain about Gnome versus KDE debates. You won’t find such wastefull debates among ‘serious’ users of Flux and windowmaker. Yep.
I can imagine -and the article makes a solid case for it- that there is a tension when releasing distributions. There are benefits to working with predetermined release moments, but this could also encourage using quick hacks as last minute solutions. The iso’s have to be on the mirrors by tomorrow, so… There isn’t much wrong with this approach, but one wonders whether the quick hack will be replaced by more permanent solid solutions. The new releasedate is rapidly approaching and -be honest- it is way more fun to work on new feature. ‘Bleeding edge’ does sound so much better in the Blogosphere then ‘stable’.
When we look at the large group of Windows 2 Linux migrators (W2L) the release cycle doesn’t have to be that fast. Windows users are accustomed to working with a version of an operating system that lasts three to five years, with a service pack or two to fix the holes. The same goes for most of the desktop software they use. The largest and most important group of Linux users keeps all kinds of servers up and running and from a professional perspective they also don’t need 6 monthly releases. So -it seems to me- there won’t be much complaint when a release is one or two months late, especially when this visibly benefits the quality of the release. Sure, there will be a cry of complaint from bloggers and online tech-reporters, but who do they write for anyway.
‘The sorry state of open source’ is a polemic wrapped around a number of very valid arguments. But it is not research and the whole article doesn’t contain anything that justifies the weird copyright notice that comes with it. Promoting a more solid method of developing Linux and better documentation, both following the best Unix traditions is fine, as is impressing the need for better enduser education. Sadly, the overall presentation might the factor that prevents most from accepting that.
This column is part of a series I write for Digiplace.nl, a Dutch website where Ubuntu users meet and learn.
Tags: Linux, Open Source