October 18th, 2004

Software Patents

» , ,

The news that Linux potentially infringes 283 patents initiate a strong reaction from the OSS community. Actually, just look at some examples of software patents and you will see the most common algorithms like Quicksort (#5,175,857) and Random Number Generator (#5,251,165). So pardon me while I marvel : “Only 283 patents?”

Some folks, like Richard Stallman, may argue that this indicates a failure in the software patents and hence we should fight against it. Such arguments equates software patents as “Anti-Commons” (coined by Michael Heller).

On the other side, the traditional view on patents is that it provides incentives to inventors, and hence create more innovation. Without protection of patents law, a small but innovative company has no chance against a big company with marketing power.

Both are right; The trick is in achieving the fine balance between “protecting inventors” and “patents abuse” so as to create an innovative environment for everyone equally. And sometimes, we swing a bit too much to the left and sometimes too much to the right but neither extreme is good for all. This is well articulated in Tim Bray’s essay on Patent Theory.The problems with (US) software patent are real: That software patents aren’t subjected to disclosual (whereas all other patents are); That USPTO have demostrate they will granted almost anything, novel or otherwise; That USPTO did not investigate sufficient prior arts before granting a patent; That software patents granted often hurts innovations then help innovations (which defends the goal of patent law); etc etc

As Lawrence Lessig wrote in his book ‘The Future of Ideas‘:

Software patents are (relatively speaking) new. At least, they are newer than software itself. For many years, software could not be patented, which means that, for many years, the US Patent Office did not collect data about prior art in patents.

But the failure of USPTO is a failure of USPTO. They have a lot of difficulties: under-budgetted, overworked patent reviewers (who have little software experiences initially); 1,500 applications per year, with persistent and political strong filer; etc etc.

So fix the patent system (or fix the office) and restore the balance.

But still, this does not answer how we going to deal with the ‘283 patents’ that Linux potentially infringes. Well, first of all, many of the 283 patents are held by ‘Linux-friendly’ company like Novell and IBM. The rest can be divided into ‘bogus’ patent and ‘non-bogus’ patent;

Dealing with ‘bogus’ patent is simple (but expensive); Wait for it to be challenge and invalidate it in courts; Such process would potentially cost more then US$1M each but hey, we cleaning up mess here. This is where Public Patent Foundation comes in (hopefully).

Dealing with ‘non-bogus’ patent are tough, especially if it is held by non-OSS-friendly company, like Microsoft. (Microsoft have huge patent portfolios btw, but so far, they have not use it aggressive against competition yet). This is where we need a ‘Cold War’ in software patents – Where two or more giants, each with huge patent portfolios, and no one wants to be the first to strike.

And Novell vows to defend Linux with its patent portfolio is admirable (They already done so by voicing out in the SCO facade). I hope others like IBM will do so soon.

Disclaimer: IANAL (I Am Not A Lawyer).

Update 11th Jan 2005: IBM announced that they will allow open source developers to share 500 of their patents to establish a patent commons. Wohoo!

Comments are closed.