In tough economic times, it's always a good idea to know what you're worth -- especially for a product potentially worth billions, even if it's often given away for free.
In a study that aims to take stock of the value of the Linux kernel and a typical Linux distribution, the Linux Foundation estimates the Linux kernel to currently be worth $1.4 billion, based on a metric that calculates lines of code, effort and developer costs per hour.
The kernel itself is just one part of a larger collection of software that makes up a Linux distribution, however. Using Red Hat's Fedora 9 and its application code as the base for estimating the value of a typical Linux distribution, the Linux Foundation found that it would take $10.8 billion to develop Fedora 9.
The new report from the Linux Foundation builds on 2002 report of David Wheeler, who estimated the value of Linux using the same metrics. Wheeler's report had pegged the value of a Linux distribution at $1.2 billion.
"Honestly, I'm surprised the number has grown so much," Linux Foundation report co-author Amanda McPherson told InternetNews.com. "We closely followed Wheeler's methodology, so I believe the number is right, but who would have thought that within a few years the code in a Linux distribution would have grown that much?"
McPherson noted that there is some subjectivity in the analysis, since each distribution's team decides which components make it into the release. Still, McPherson explained that the goal of the report was not to define what made up a Linux distribution. Instead, the researchers followed the Fedora team's lead on what constituted a distribution, basing the Linux Foundation analysis on all of the packages they include as part of the Fedora distribution.
As a result, the Linux Foundation included a wide swath of applications that are in Fedora, such as the Firefox Web browser and the OpenOffice.org office suite. The analysis also wasn't limited to desktop elements, so it includes server elements as well.
Fedora is neither the smallest nor the largest Linux distribution. It is, however, of the same base that Wheeler used in his 2002 study, which makes for a more direct correlation in analysis.
The basis for Wheeler's original study, as well as the new Linux Foundation study, was on what's known as a Software Lines of Code Count (SLOCcount) tool, which he used with a model of estimating the expense and effort of building software, called the COnstructive COst Model (COCOMO).
The cost in the latest study is based on U.S. Bureau of Labor Statistics, which estimates the average salary for a U.S. programmer at $75,662.08.
Using SLOCcount, the Linux Foundation measured 204,500,946 lines of code in Fedora 9, and estimated using COCOMO that it would take 60,000 person-years of effort to build the distribution. For the 2.6.25 Linux kernel, which is included in Fedora 9, the Linux Foundation counted 6,772,902 lines of code, which it estimated would take 7,500 person-years to build.
While SLOC is a valuable metric, it values quantity above all else -- rather than quality and other issues.
However, those other criteria are also important to take into consideration when figuring out the value of the whole Linux ecosystem. McPherson noted that industry researcher IDC's data estimates the 2008 value of the Linux ecosystem at $25 billion, on its way to predicted that the growing to $49 billion by 2011, according to an April report from IDC and the Linux Foundation.
The IDC data only takes into account the server side of things, however. In McPherson's view, it would be interesting to measure the impact of Linux development in the embedded, mobile and desktop areas as well.
"A qualitative way to measure it is the worth of the ecosystems that are built around [Linux] or would certain products/ecosystems be even possible without the software?" McPherson asked. "Software is time more than money. How much would it set back a consumer electronics manufacturer like Garmin or Sony or Motorola if they had to build this software themselves? Or could Google have survived if they would have paid a per-server fee to Microsoft or Sun as they scaled their business? That is one way to measure the value of the software."