Is it better to have multiple flavors of a thing to suit the tastes of multiple users, or do the benefits of standardization outweigh the value added by all those variations?
That question was at the heart of a contentious discussion going on in recent days throughout the Linux blogs.
It all started when DistroWatch published an interview with none other than Linus Torvalds early last week.
‘Absolutely Required’
For those who missed it, Torvalds also featured prominently in the headlines late last month for revealing that he had switched from KDE to Gnome, as Computerworld and others reported. Sure enough, Torvalds confirmed that bit of news in the DistroWatch interview, and then went on to assert that he believes it’s a good thing that there are so many Linux distributions.
“I think multiple distributions aren’t just a good thing, I think it’s something absolutely required!” Torvalds told DistroWatch. “We have hundreds of distros, and a lot of them are really for niche markets. And you need that — simply because different markets simply have different requirements, and no single distro will take care of them all.”
Having multiple players also “just keeps everybody honest, and allows you to compare them,” Torvalds added. “It may all look a bit messy and complex, but I’d much rather have a multi-party system over a single-party one. Even if it’s more complicated.”
Contentious Idea
Now, vocal members of the FOSS community have long argued that developers should unite behind a single distribution, reasoning that standardizing would give Linux a better chance at a bigger piece of the market while also minimizing duplicated effort.
But the question is a controversial one, and Linux bloggers weren’t shy about making their opinions known.
“Calls for ‘less fragmentation’ are vacuous without a call to unite behind something specific; then we can debate the pros and cons of what would be gained and what would be lost,” wrote DragonWriter on Slashdot, where the conversation had unleashed close to 800 comments by Friday. “Of course, the people you really have to convince are the people working on whatever would be axed, since it’s an open source community and the only way to make that happen is to convince those people to stop working on what they’ve been working on and start working on something else.”
‘Absolutely the Wrong Thinking’
“We need a main, reliable, one size fits all DESKTOP distro,” responded unity100.
“Trying to make one big distro is absolutely the wrong thinking, it would be impossible to decide on anything first of all, and it’s been proved this concept doesn’t work already — by a company called Microsoft,” asserted kcbanner.
“How about Linux Starter, Linux Home Basic, Linux Home Premium, Linux Business, Linux Enterprise, Linux Ultimate,” quipped Emperor Skull.
‘I Agree 100 Percent’
The topic also appeared not once but twice on Digg as well — generating plenty more comments there — so we here at LinuxInsider decided it was time to hit the streets of the blogosphere and ask around.
“It’s odd — normally I absolutely despise Linus,” Monochrome Mentality blogger Kevin Dean told LinuxInsider. “More specifically, I usually object to the way that he seems to inspire hero-worship: ‘Linus is using Gnome now, so I should think KDE sucks!'”
This interview, however, “is the first I’ve read where I agree with him 100 percent,” Dean added.
‘Fatally Stupid’
“The simple fact of the matter is that most people don’t give too much critical thought to themselves — to think about themselves is seen as ‘selfish,’ as if pursuing one’s self-interest is somehow bad,” Dean explained. “Because of this, people are willing to overlook how they’re inconvenienced, annoyed and hindered by their operating system.”
This “‘settle with what you’re given’ mentality is exactly why people struggle to cope with Windows when it doesn’t meet their needs,” he added. “It would be fatally stupid to try to do the exact same thing on Linux.”
There is no “one-size-fits-all distro,” Montreal consultant and Slashdot blogger Gerhard Mack told LinuxInsider. “Personally, I prefer the simplicity and cautious stability of Debian, while most new users prefer Ubuntu,” he noted.
Easing Migration
Users benefit from having multiple distributions “just like they benefit by having multiple sources for widgets — it keeps the costs down and prevents lock-in,” blogger Robert Pogson told LinuxInsider.
What should be avoided, however, is the creation of “distros so divergent that migration between them is difficult,” he warned. “Developers would benefit if they had an easier way to make an application that would run seamlessly on any distro.”
Torvalds is “the Hari Seldon of the computer world, distributing responsibility widely because he knows that’s how to up the odds of his system’s survival,” Slashdot editor Timothy Lord told LinuxInsider. “I’m really glad to have Linus himself say this, because one of the constant annoyances of the Free Software world is the constant, ongoing, petulant call to ‘unify’ this or that, and to ‘stop wasting time with duplicated effort’.”
‘Impossible’ Problem
The more ways people approach packaging Linux “plus other stuff, the better,” Lord added. “Some of the ideas may be bad (or at least less popular than others), but keeping a wildly diverse set of likely starting points has been a good thing for the last decade-plus of Linux development. The invisible hand needs some room to swing!”
There’s also the question of who’s supposed to do the consolidating, Lord noted.
“Anyone who believes that the number of options available in a marketplace of ideas ‘should’ be combined or trimmed down is perfectly free to do the combining as an actual project, make it available, and see whether others like the result,” he explained. “But human desires are complex, evolving, and contingent; it’s not just a hard problem to predict the ‘ideal’ outcome in a marketplace, it is (literally) impossible.”
‘You’re Never Really Done’
Indeed, whether there can ever exist a single, ideal solution is an open question.
“I think the simplest argument in favor of multiple distributions (and maybe the only one you need) is that the different distributions give authors a chance to try different things,” Slashdot blogger drinkypoo told LinuxInsider.
“I believe in the end that there really is one right way to do any given task, but you don’t find it by finding one way that works and sticking with it forever,” he concluded. “You find it by trying different things and seeing which one works best, and you’re never really done.”
Loading ...
Good discussion. I absolutely believe that from an overall Linux standpoint, the more alternatives, whether fragmented or not, the more ideas, the more variations, the more thinking… you get the idea, the stronger the overall effort becomes.
I have no problem at all if someone wants to group a set of features together – and make their own distribution. I do have a problem with a call to get rid of a and create b. Let people decide that for themselves in what they choose – and use.
I have three distributions that I happen to use quite a bit. They do have one thing in common: a Debian packaging mechanism. One of them, sidux, is a cutting edge, rapidly changing, volatile desktop system, based on Debian Sid repositories. sidux aims to create its own Linux kernels frequently, which constantly update and add the latest kernel improvements and new hardware support. sidux also attempts to carefully control the volatility of the Debian Sid packages, either correcting inconsistencies or putting packages on hold until there are correctly matching packages available. This really helps a lot in making Debian Sid usable on an every day basis.
Another system that I like to use, antiX, is a relatively small, lean, fast distribution which uses Debian Testing repositories, installation features from SimplyMEPIS, graphical appearance tools from TinyME and Absolute Linux, and system administration tools from sidux, smoothly blended together with Debian Testing. It is a very flexible system, one which can be used as is or transformed quite easily into whatever you want it to be. Quite handy, yet quite different from sidux.
A third one I like is SimplyMEPIS. On the surface, antiX and SimplyMEPIS appear to have a lot in common. In reality, SimplyMEPIS is a full featured, conservative, stable, easy to use system. What it has in common with antiX are the installation tools, a common base forum, and many users who use both systems. Other than that, they diverge quite a bit in characteristics and interests. SimplyMEPIS always favors stability and what will make the user experience simple.
I have uses for all three of these systems. Why would I want to limit my own choices?
I also enjoy testing many other systems. Though I do not use all of them on a regular basis, they still form an important part of the personal system base that I use. Why would I want to reduce or limit the different kind of systems that I can test?
I enjoy installing, configuring, testing, and writing about my experiences with different systems. My contributions to the community are to provide defect reports when I run into something that I believe to be functioning incorrectly. I report what I was doing, what behavior I encountered, what happened, and what I think should be the correct behavior.
I also write a lot of forum notes to promote Linux software and help relative novices decide what they can do to become a part of the community. I recommend systems, but I primarily categorize systems, so that others can make their own choices about what may work best for them – and that may NOT be what works best for me.