Jimmy's weblog

Just back from listening to a two-hour talk by Dr Richard Stallman on the danger of software patents. RMS, as he is more commonly known, is one of the fathers of the Free Software Foundation and author of many of the GNU tools including the Emacs editor.

Below is my summary of his talk. I will be posting more detailed notes when I get the chance.

Main talk

RMS is fairly impatient. During the over-long introduction he was making George Bush-debate like faces.

It’s important to disentangle patents from copyright - they are completely separate but people like to combine them and talk about “IP”.

Copyright protects a concrete example of a work. (Software) Patents cover general ideas, not the actual expression of those ideas.

Software patents are bad because software products are far more complicated and contain more ideas than physical products. A physical product is often covered by one patent while a software application may well infringe on twenty or more. The Linux kernel alone is estimated to infringe on 286 US patents.

Problems occur when standards (de facto or official) turn out to be covered by submarine patents which aren’t revealed for years. It’s almost impossible to accurately search for patents which may affect your software product.

Despite popular wisdom, software patents don’t help the small inventor. He may patent a particular new idea but the rest of his software is likely to infringe on a number of patents from large companies. These companies will force cross-licensing and them compete against him anyway.

Around 1995, IBM estimated that the benefits of cross licensing their patent portfolio were worth an order of magnitude more than actual licensing revenue. That is, the ‘cost’ of their patents was roughly ten times more than the ‘benefits’.

If you encounter a patent which covers software you are developing your alternatives are to: Avoid, cross-license, or try to invalidate. Avoidance is difficult and cross-licensing and (attempted) invalidation costs LOTS of money.

Patents actually appear to be slowing the progress of the software industry rather than promoting it which is the reason patents are granted.

Why shouldn’t software be allowed to be patented when everything else can be? Largely because software is far more complex and even the simplest product covers lots of ideas. The complexity comes not due to the fact that software engineers are smarter than chemical / physical / electrical engineers but because software is a perfect mathematical representation, not subject to pesky physical laws.

Useful analogy - software as a symphony. If symphony ideas had been patented, Beethoven wouldn’t have been able to produce works based largely on existing ideas with just a few new ones thrown in. Even he couldn’t have completely developed his symphonies from scratch.

Questions

Q - Introducing the economic term ‘Rent revenue’, where income is derived with no effort. (re: patent parasite companies *cough*SCO*cough*)

A - Don’t try to talk economics. This forces the discussion to be about relative costs, instead of about absolute issues of what is possible or impossible under the Software patent regime. Most people don’t understand economic arguments anyway.

Q - What about copyright issues where software is stolen, rebadged and sold?

A - That has nothing to do with patents! That is a copyright issue.

Q - Are there countries where there are no software patents, and has this helped their economies?

A - Yes there are, such as India and Brazil. As to whether it has helped their economies, that is almost impossible to determine, as it is tied in to a lot of other detail which has nothing to do with patent law.

Q - But if a process has been published in a journal (as was the case for the LZW compression algorithm), doesn’t that invalidate the patent by serving as its own prior art?

A - In the US there is a one-year grace period during which published work can be patented even though it has been published. Besides, the LZW algorithm had already had its patent applied for at the time its journal publication came out. It’s just that no-one knew about the patent application, so European/Commonwealth patent law (that only trade secrets can be patented, and publication before patent invalidates all attempts to patent subsequently) wouldn’t have helped.

Q - You’re preaching to the choir here, how can we convince the wider population and politicians about the evils of software patents?

A - I’ve already told you! Do I need to repeat my talk? That’s what I’ve been talking about! People don’t need to be lefties, or technically minded to agree with us, as evidenced by the conversion of the vice-chancellor of Austria (not known to be politically left or a tech-head). What we must do is tell people that the simplistic view of patents they have is wrong, and how much software patents hurt innovation and free development. Most will change to our side when they know what the truth is.