David A. Wheeler's Blog

Tue, 08 Jun 2004

UPDATE: Why OSS/FS? Look at the Numbers!

After about half a year, I’ve now made a big new release of my paper Why Open Source Software / Free Software (OSS/FS)? Look at the Numbers! This large paper lists various statistics I’ve found justifying why anyone should consider OSS/FS.

In theory, this shouldn’t be as needed, because much of the IT industry has figured that out already. Look at IBM’s Linux ads, the continuing dominance of Apache and bind in their markets, the take-up by many organizations of Mozilla and Open Office, and so on.

Yet I still encounter far too many software users who never consider their alternatives (even if the alternatives might be much better), and organizations that still create acquisition processes that unnecessarily exclude OSS/FS. These decisions aren’t good for the users and organizations; without considering their options, they’re unlikely to always make the best choices. But worse still, Adam Smith’s invisible hand cannot work without competition. If too many users fail to consider their alternatives, they can encourage vendors and projects to stop working to get better (since the users won’t change anyway).

There’s a certain large software vendor who has paid for a number of reports to “prove” that their proprietary stuff is better, with an implication that their proprietary products are always better. As I mentioned in my interview with NewsForge, you should consider suspect any study of a vendor that’s funded by that vendor. Organizations self-fund public studies to give them good press, not necessarily to give customers a full understanding. I’m actually not out to get the vendor, but it does bug me when people use vendor-funded studies as though they were really independent research. Yet if you don’t know about alternative analyses, it’s hard to respond to the vendor-funded analyses. My paper is a useful antidote.

So… what’s in the updated paper? A lot of little improvements and new references to studies. In market share, I’ve updated the Netcraft and Security Space surveys of web servers (Apache still dominates), and added a new DNS survey showing (in yet another way) OSS/FS’s massive dominance in that market. In security, I noted that SuSE and Red Hat have received Common Criteria evaluations, and noted some of the serious security problems that seem unique to proprietary operating systems (91% of broadband users have spyware; 80% of spam is now sent from infected Windows machines). For reliability, I’ve added the May 2004 Netcraft survey of reliable servers (80% of the top ten most reliable hosting providers are OSS/FS based), and an IBM study showing that Linux is quite reliable under continuous high stress for 30 and 60 days. For scaleability, I added a reference to “Thunder”, the fastest computer in North America (it runs on Linux). I added a lot of text noting that OSS/FS isn’t just pirated source code (a new claim being heard occasionally, though so far only from people who appear to be paid to say it). I added much more text about SCO and ADTI, including some of Tanenbaum’s statements and a Minux/Linux code study refuting ADTI’s claims. In the discussion about forking, I noted the XFree86 vs. X.org fork to show that forking can be about licensing, not just about project speed. Various typos fixed, too; my thanks to the many who sent suggestions.

Overall, this should be an even better starting point for those of you who need specific quantitative data on OSS/FS, and don’t know where to start. It won’t prove OSS/FS is always better — I use proprietary software myself — but shows that OSS/FS needs to be considered.

path: /oss | Current Weblog | permanent link to this entry