I find it amusing when the same human beings who rationally promote GNU/Linux on the server get hung up on small details when the subject changes to the Desktop. Sam Varghese points this out:
“Both Weber and Behlert were at pains to point out that the SUSE enterprise desktop would cost less, provide inter-operability, ease of use, desktop security and ease of management far in excess of other operating systems. But still, they said, it would be incorrect to position SLED as a full replacement for Windows.
Asked whether this was due to the licensing agreement which Novell had signed in 2006 with Microsoft, both replied in the negative.”
The fact is the same characteristics that make GNU/Linux work on the server are available on the desktop and they are more than sufficient to choose GNU/Linux desktops. One does not “need” a particular M$-only application any more than the world needs only one brand of car. We need generic software for IT and we should shop around. Many can and do work without M$’s OS on the desktop. The performance, reliability and security are great as is the low cost. What’s not to love about IT based on GNU/Linux?
In fact, RedHat which is doing very well on the server has lots of clients using GNU/Linux and OEMs are shipping millions of units. They just are not as tied to M$’s apron strings as Suse.
“What do Vanderbilt University, the University of Stirling in Scotland, and Yuba K12 School District in Northern California have in common?
Along with thousands of other higher-education institutions and K12 school districts globally, they’ve chosen Red Hat® to enable their IT infrastructures and encourage students to learn in new and exciting ways.
Desktop or server. To teach. Learn. Manage networks. Or develop and conduct research. Red Hat can empower you with affordable, reliable open source technology.”
The quality of FLOSS is superior to non-FREE software in many ways but measures of cost and quality are undeniable:
“The democratization of innovation has also demonstrated a remarkable solution to the problem of The Mythical Man Month, thereby transcending the limits of conventional industrialization. For example, sourceforge.net is an open source development resource that hosts over 180,000 projects and has more than 1.9M registered users as of December 2008. Extrapolating from the extensive FLOSS (Free, Libre, and Open Source Software) survey of 2002 (and updated in 2005) there were over 490,000 sourceforge.net developers in 2006 [when the thesis of this section was first developed--tiemann] who spend more than 10 hours a week or more tending their open source projects —an aggregate effort of some 5 million person-hours per week. The three top reasons they list for their involvement is:
1. Because it’s fun
2. Because it improves their skills
3. Because it is good for society
Note that this does not include Linux developers (who use kernel.org, not sourceforge.net), nor Apache, nor the GNU project, nor many of the other larger and more heavily commercialized open source projects. To put these 5 million joy-filled person-hours per week into perspective (again, this does not include Linux, Apache, GNU, or many of the other “large” projects), let’s look at the productivity potential of the most successful proprietary software company, Microsoft, in two ways (using numbers that were contemporaneous with the FLOSS survey data, October 2006):
1. If all 61,000 employees wrote code, they would have to work over 80 hours/week
2. If Microsoft’s $6.6B/year R&D budget were spent on programmers averaging just $25/hour, they could pay for about 5 million person-hours of work per week
Thus, the sourceforge.net website has equaled or exceeded Microsoft’s productive potential using a social, not an industrial model. When we consider all the open source developers not included in the sourceforge.net numbers (numbers that are increasing exponentially), we see the clear emergence of a new software production capacity entirely outside the conventional limits of the industrial model. Moreover, we find precisely the kind of improvements that Deming would have predicted by taking a transformative approach: according to findings published by Coverity, typical proprietary software has a defect density of 20-30 defects per 1,000 lines of code (KLOC), a number relatively unchanged since the 1960s. When they measured the quality of the Linux kernel (and later, other open source software) they found the following results:
2004: 985 defects in 5.7 MLOC of Linux kernel source code, or 99.3% lower defect density than average (compare to 114,000 to 171,000 defects in same amount of code)
2005: While the Linux kernel grew 4.7% in overall code size, defect density decreased by 2.2%. Moreover, 100% of all “serious” defects identified were fixed within 6
2006: The survey was expanded to entire LAMP stack and an additional 32 OSS programs. No correlation found between size and defect density, implying OSS development methodology is not limited by scale (nor restricted to just Linux developers)
What the top industrialists could not achieve with proprietary software and financial capital, free software has demonstrated with community development and intellectual capital.
That’s not too different from my oft-repeated expression, “FLOSS is a cooperative project of the world.”. The world can and does produce its own software for server and desktop. Not using GNU/Linux without any reason is questionable. Dependence on particular applications is not an excuse for poor design of IT systems. The mistake was often choosing to be locked-in to the non-FREE model of software development long ago.
Ignoring those kinds of savings on the desktop due to imaginary problems is foolish. RedHat and others have a hard time making money with GNU/Linux on the desktop because so many are locked in but that doesn’t mean the rest of us cannot save big money by using it. Canonical, Dell and many other OEMs do make money shipping PCs with GNU/Linux but we can install it ourselves customized as we wish at very little cost. I recommend Debian GNU/Linux on client or server because it works. GNU/Linux works for Munich, School Divsion 73 in BC, Canada, the city of Largo and Google on desktops and servers. Why not you?