Lora Bentley asked:


Lora Bentley spoke with Amanda McPherson, marketing and developer programs VP at the Linux Foundation. She and two colleagues recently released a new paper, “Estimating the Total Development Cost of a Linux Distribution.”

Bentley: Your study found that it would cost $1.4 billion for a company to build the Linux kernel from scratch today, and $10.8 billion to build an entire Linux distribution similar to Fedora 9. Can you explain how you reached those figures?

McPherson: The conclusions were reached by using David Wheeler’s well-known SLOC tool, SLOCCount, which makes use of the industry standard COnstructive COst MOdel (COCOMO). This methodology takes into account lines of code written, the appropriate number of labor years, and salary adjustments for inflation. We wanted to come up with a real number based on the one thing you can quantify in open source — code. We used a well-regarded methodology and tool that had been used before. Instead of making random projects, we thought this was the best way to approach it.

Bentley: Why the Fedora community distribution and not another?

McPherson: Fedora is the basis for Red Hat Enterprise Linux, which represents a large percentage of the Linux market. This provided us with a very relevant model to assess. Also, David A. Wheeler had used Red Hat for his study in 2002. OpenSuse and Debian/Ubuntu would, of course, also be great targets for this study. We may do that at a later date. We also would like to use an embedded distribution.
Bentley: What do or should these findings mean to proprietary software vendors?

McPherson: I think it means the future of software development is collaborative. These systems have grown so powerful and so important that for any one company to fund the development on its own would be a foolish and possibly financially untenable decision. Software development today actually requires collaboration in order to innovate at the pace the market demands. Consider devices like the Kindle and Gphone. They wouldn’t likely be available today were it not for the billions of dollars worth of R&D that they can use from the Linux kernel. You see companies like Intel using Linux and open source components in the Moblin project to expand the use of netbooks running its products. Intel could instead develop proprietary software in-house to meet this need, but why would they when they can make use of billions of dollars of free R&D? Things have changed since the desktop computer revolution.

Bentley: Don’t many proprietary software vendors recognize the value of open source now given that so many use open source in some way or another?

McPherson: Absolutely! You can also look at our “Who Writes Linux” report to see that hundreds of companies support Linux development directly. This study shows that those companies (such as IBM, Intel, Red Hat, Novell and HP) have made a very smart decision. They can fully participate in a large ecosystem and make use of free R&D without having to shoulder the burden all alone.

Bentley: So why is a study like this one helpful?

McPherson: Sometimes it’s easy to take a ubiquitous piece of technology for granted, especially one you can use for free. I think it’s not just Linux we take for granted: Just imagine the R&D value of the Internet itself and what that means for our economy. Compared to that, Linux seems small, yet when you think about all the innovation it’s powered or is powering, you start to get the idea. I honestly can’t imagine where we would be if Google had had to pay a company a per-server fee for its servers. I do not believe the economics would have been there to build out the powerful search network that we all use everyday. This study makes us appreciate the sometimes-unheralded piece of software and the license that has powered this innovation.

Bentley: Do the findings have added significance in light of the current economic climate?

McPherson: I think so. Linux has always been a lower-cost alternative to Windows, but this report illustrates its economic impact on technology innovation. It’s exciting to see how the collaborative development model is fueling a new category of devices and technologies that would be at least a decade into the future if it weren’t for Linux. Let’s remember that in software, time is money; oftentimes time is more important than money. For a company like Google or Intel to be able to make use of this code that has taken years to develop, drives innovation and keeps costs low for consumers.