The true costs of Linux development

A Register reader challenges the Microsoft 'one man' theory of Linux

Microsoft's claim that any talented individual can compete effectively with it has been challenged, as one might expect. Yesterday a Microsoft attorney (see earlier story) cited Linus Torvalds in court as an example of how one person can produce a sophisticated OS, but Register reader Eric Green of Linux Hardware Solutions has some numbers that make interesting reading. "You might be interested in my current estimates of the resources involved in the Linux project. "Start of project: The formation of the Free Software Foundation in 1983 (most of the core libraries and utilities of GNU/Linux are owned by the FSF, as well as the core C compiler used to compile the beast). Meaning: 15 years since start of project, GNU/Linux is finally a threat to NT Server. "Estimates of programmer time: Linux kernel: project began in 1993. In the five years since, an average of 200 programmers has been hard at work part-time, approximately 500 man-years. "Meaning: approximately $50,000,000 investment in the kernel. "The core OS itself (kernel + API libraries + core OS utilities) has taken over 1000 programmers close to 15 years to complete (to be fair, the first few years were done with only 20-30 programmers). Figure 10,000 man-years involved in the core operating system. Meaning, approximately $1,000,000,000 investment in the core GNU/Linux operating system. "Add in the other software bundled with the typical Linux distribution and you have an approximately $4,000,000,000 investment contained on that one single CD-ROM. My current estimate is that it will take another $200,000,000 worth of programming time over the next three years focused on ease-of-use and ease-of-administration issues in order to make Linux competitive with Microsoft's core Windows 98 and Windows NT Workstation products (currently it only competes with Microsoft's Windows NT Server product, which accounts for a very small percentage of Microsoft's OS shipments). The fact that the programming time is mostly donated does not reduce the magnitude of the investment needed to write a modern OS. The kernel (which has been written by a team of hundreds since a few months after the start) is the smallest part of the investment. "The sad part is that most of the general public is ignorant enough to buy Microsoft's argument." If Eric's numbers are out by even a factor of two or three, we could still note that effective Linux R&D development spend is of an order to match Microsoft's annual development spend, which kind of puts things in context. But the point Microsoft's attorney was failing to make, and should have made, is that it can be seen to be possible, from the Linux experience, for a single individual to start a massive ball rolling, ultimately placing that individual's innovation in a position to challenge Microsoft. QED Microsoft is not a monopoly. Yes, we realise it's a tenuous argument and Linux is a special-case phenomenon, but it's a better shot than the one they fired. The funny part, we reckon, is that Microsoft can't comprehend an explanation that doesn't involve just one person driving. Complete Register trial coverage Click for more stories

Sponsored: How to determine if cloud backup is right for your servers