Research Freedom And Software Freedom

There's a book on the biotech industry I was reading recently (Prescription for Profits -- dumb title, obviously tacked on by marketing) that tricked me into thinking a little differently about SoftwareDevelopment.

Basically, the author describes the genesis of the modern BioTechnology industry by telling the story of gene splicing, Genentech, and the NIH's AIDS research effort.

One of the drums she beats is that, with the coming of commercialization, researchers stopped sharing their results (not just information, but reagents) with each other (as they had, fruitfully, for a hundred years). They intentionally aimed for personal financial gain (and gain for their companies), which slowed progress in basic research. In fact, certain avenues of research were closed off entirely: if company A had a patent on some discovery, they would only license that patent to labs that weren't going to work on something that would diminish the value of that patent (e.g., by devising a better treatment).

On to software:

We don't have a hundred-year tradition of sharing results behind us. And most of the growth in the field has been in the wake of the tremendous success of commercializing software (even in contrast to biotech, which is a lot more volatile).

But we do have a chorus of voices that call to us to share our results: RichardStallman, OpenSource proponents, and (of course) legions of developers who give us things like TheGimp, GnomeDesktopEnvironment, and even new useful features for EmacsEditor.

I've thought for a while that there's a huge gap between what we know how to do with software and what we actually do. In my eyes, this shows up most strikingly in OperatingSystems, where dominance by two families has resulted in OSes that don't do anything significant that hadn't been thought of in 1985 (and, in many cases, already implemented). Seriously: why is a process always stuck on a single machine? Why do ApplicationProgrammers still have to treat memory in RAM differently than memory on disk? This is stuff academics worked out while I was still an undergrad.

Ditto for UserInterfaces: we have so much more computing power, network connectivity, and clever, low-cost input-output devices now than we did in 1983 (the year AppleLisa debuted). But your fingers are either on a keyboard or a pointing device right now (and 9 out of 10 it's a mouse) -- we're stuck in WimpInterface, and the nagging feeling that something better is available grows year by year.

I've never been able to completely dismiss RMS's arguments for software freedom (both FreeAsInBeer and FreeAsInSpeech); mostly I've told myself that we need ownership in at least part of the software field to provide incentives for progress.

But now, with biotech as a persuasive analogy, I think the software field (as opposed to the software industry, which it contains) is so hung up on how to make money from what we do that we're retarding our progress immensely. I'm doing stuff with GeneticAlgorithms at work, at it's viewed as novel -- but this is stuff academics did at least fifteen years ago!

My diagnosis is that, as with biotech, we've burned through our basic research inheritance from the 1960s and 1970s and haven't successfully advanced it since. Commercial basic research hasn't come up with any software breakthroughs since XeroxParc's work twenty years ago. Government research in universities seems to be either very mathematical, or strongly guided by industry (e.g., most of the GeneticAlgorithm papers I read focused on circuit layout) -- making it more like applied research than like basic research.

Finally, I think the economics of software are so radically different from that of any other field I can name that analogies break down, and we might get some truly counter-intuitive results from wholesale sharing. In other words, the efficiencies introduced to the economy from a staggering leap forward in IT might more than make up for any individual gain we give up by sharing (instead of owning) the innovations each of us make in software. To make this concrete: if IT improvements slash the cost of everything you buy by 50%, that's as good for you as a 100% raise!

But even without this personal benefit, think about what a 100% raise for everyone means, especially for the poor. Is it really better for society for us to own most of what we create? If it isn't, how do we defend what we're doing?

-- GeorgePaci

The problem is with the clash between research and commercial interests. For research to succeed, you need to share IntellectualProperty with others. For commercial interests to succeed, you need a way of protecting your intellectual property rights.

With research, the only way to establish IntellectualProperty rights is to share the intellectual property. If you come up with a neat idea, and you don't publish, but Person A does, then people recognise it as Person A's idea, not yours.

In the commercial world, copyright and "trade secrets" provide ways of establishing IntellectualProperty rights in secrecy. Secrecy in turn makes it hard to protect the intellectual property rights; if you publish/share your intellectual property, it's hard to see who is abusing your rights. This motivates companies to protect their rights by keeping the intellectual property secret in the first place.

Curiously, the main mechanism in place for enforcing IntellectualProperty rights in a collaborative environments (patents) have a bad rep for doing so. Putting a patent on a technique is bad. Keeping the technique secret is not.

A legislative approach could help. For example, when you build a building, blueprints have to be registered. Anyone can come along and view those blueprints. Similarly, legislation could be established that for any commercial product to be sold, the full and complete SourceCode for that product needs to be registered in a public database. You'll have screams of complaints by the IT industry, of course. However, I suspect it would still be possible to make money in such an environment. It would be necessary, however, to improve the mechanisms for enforcing IntellectualProperty rights as they apply to software.

-- RobertWatkins

Is it possible to have GnuPublicLicense-style patents in biotech?

CategoryResearch