Medical researchers agree that the best way to see whether drugs are really effective is large-scale double-blind testing, which is expensive. Most such drug testing these days is paid for by drug companies. Naturally, they tend to test their new, expensive, patented drugs. There are relatively few tests run on old, cheap drugs with expired patents or no patents, even when those drugs might have interesting new uses that could help lots of people.

It's pretty obvious that there is a potential conflict of interest between drug manufacturers and public health. It's less obvious how to resolve it. The present resolution is not quite putting the fox in charge of the hen-house - but it's pretty close. Why am I writing this in a magazine about Linux? To remind you that there is a potential conflict of interest between business and society, and to talk about how that conflict can arise in free software.

Free software is important to society because it gives us all a computer environment that we are free to use and to change. When that environment becomes steadily more powerful and easier to use, as has happened and continues to happen with Linux, then we all benefit. However, those improvements don't happen by themselves; they happen because people work on them.

Free software at its best is a collaborative effort in which many developers work together in different ways to produce a harmonious whole. Successful free software development efforts generally require central maintenance - one person or a small group of people who weave the various contributions together while maintaining the overall harmony. It's neither anarchy nor dictatorship.

The interests of society are best served when the central maintainers are free to choose the changes they prefer. When the maintainers make good choices, the project succeeds. If the maintainers make bad choices, that project will fail, or else different maintainers will pick up the sources and start making their own choices, an occurrence known as a fork.

Businesses, of course, are interested in profit. When the maintainers of a free software project work for a company with a direct business interest in the project, their decisions about the project are no longer entirely free; they are constrained by the needs of the business.

The potential problem is not really that business interests will cause bad choices to be made. After all, it's probably not good for the business if the free software project fails or is forked by somebody else. The potential problem is that business interests will cause certain good choices to not be made, either because they are against the interests of business or simply because the maintainers don't have time to focus on particular issues.

This issue is not purely theoretical. In fact, in my years in the free software community, I've seen it frequently. Rather than speculate about other people's motivations, I'll give an example from my personal experience. For a few years I was the GNU binutils maintainer and also an employee of Cygnus Solutions (which was later purchased by Red Hat). During that time I often had to choose to implement features useful only for specific Cygnus customers rather than features useful for a broad range of people. For example, to this day the GNU binutils include a program which can convert an ELF object file into a NetWare Loadable Module. That program is rarely used even on the Intel x86 platform, but in fact several man-months of time were put into supporting other processors - work which I doubt anybody has ever used, as it was intended to support the Processor Independent NetWare project which was, in the end, canceled. That time could surely have been better spent, and in fact the nlmconv program continues to have a minor cost as the current binutils maintainers have to make sure that it continues to compile, and will eventually have to decide to get rid of it.

On the other hand, I don't want to minimize the benefits of having businesses pay the maintainers of free software projects. Most obviously, it gives those maintainers more time and more resources to work on the project.

But we as a society shouldn't let these benefits blind us to the potential conflicts. As with most things, we have to weigh the benefits and the costs.

In fact, I think that as a society we're doing fairly well right now with free software. I think that most free software projects have found a reasonable balance between business interests and other interests. But as free software becomes steadily more popular, business interests become steadily more entangled with the free software world.

I don't want the free software world to wind up like the drug industry. I hope you agree with me and see that although it hasn't happened yet, it still could. Fortunately, as with all things in the free software world, we're all free to choose what will happen.