Weblog entry #200 for simonw
See all weblog entries made by simonw.
In his blog entry Jeff Jaffe starts by giving publicity to an unwarranted article that creates a "non story" out of this years IDC report.
The story claims that Linux is losing share of the x86 market, but the IDC press release actually states that Linux is gain market share faster than Windows in the server market, it just isn't doing it on what IDC define as "x86" platform. IDC's definition of x86 also seems quite narrow. Sorry x86 may be a big platform, but it is no longer the only game in town, and it is showing its age, propped up by software platforms with limited support for other architectures. There are other debates about whether the rate of growth of market share is slowing, but I think that is almost irrelevant debate about the differential rates of growth.
Then Jeff goes on to say we need a more "standard" Linux, with one way of doing things, one solution for each problem. But I think this is simplistic thinking, a natural human tendency to shy from complexity in intellectual endeavour.
Whilst I think having two major competing desktops (if I might loosely call GNOME and KDE desktops) on GNU/Linux (and other free software operating systems!), divides efforts, increases cost and complexity. It seems natural to assume that it would be better if we could all agree on one, and work towards the one true desktop (KDE 5?).
But I think the "one true goal", is a myth like; bug free software. Diversity in an ecosystem leads to specialization and thus to other efficiencies in how needs are met, needs of both vendors and users of software.
Sure it may be right for Novell to support one desktop (GNOME not KDE - although Novell Enterprise Desktop supports both?!), one way of securing applications (App Armour not SELINUX), but for the end user it may be better to have a diverse and competing supplier base that do these things in different ways, so they can pick the best fit for them.
Where the split happens of course is in proprietary closed source software, which can't cope with the diversity, because it can't easily be adapted to live with the diversity that exists by the people who are responsible for the diversity.
Whilst Linus may claim not to be a free software radical, his insistence on not supporting a stable binary API, is one of the guarantees of diversity (and rapid development), and also the reason that closed source (driver) vendors will struggle to support a wide range of GNU/Linux platforms.
Let a hundred flowers bloom, let a hundred schools of thought contend. If this will not do, I note Debian derivatives (cough mainly Ubuntu cough) hold nearly 50% of the desktop GNU/Linux market, he is in a position to suggest Novell switch to a Debian based distro on the desktop, and eliminate in a single decision a whole raft of this pointless variation that limits ISVs on the desktop, and presumably also limits who Novell can sell their nice admin tools for GNU/Linux to - (directory services anyone?). But if market differentiation is more important than standardization to Novell it won't happen, and if it isn't good enough for Novell why should anyone else do it?
Comments on this Entry
Evolution then takes the best adapted at a point in time and tends to favour them plus/minus some random activity. It's not perfect you can't evolve for the future and you are always constrained by your history but it's still very effective.
In Linux this means you need lots of distros made up of lots of packages. Lots of them will fail (because they are rubbish) and at the same time some good ones will also fail (random events). Overall over time you will tend to find the better ones being more popular.
Linux distributions are open-source they reproduce sexually and good ideas in one distro quickly cross to other ones - you can argue that closed source Unix and Windows is asexual and can't take advantage of other ideas as quickly. This is one of the strengths of open-source and the reason why forking is more of a problem in the closed source (asexual) community.
Diversity is important and you just have to tolerate the short term waste and duplicated effort for the longer term gain.
I'm sure if everyone were to concentrate on one distro with only one set of apps we would get a really good distro really soon, only it wouldn't be able to change as the market place around it changes. It's much better to put our eggs in several baskets and accept that we pay for that in the apparently wasted effort with future flexibility.
Remember even though we are consciously designing, just like evolution we can't see into the future we are building based on past events. Biology is littered with examples where monoculture/lack of diversity gives short term gains and long term catastrophes.
--
"It's Not Magic, It's Work"
Adam
[ Parent | Reply to this comment ]