[Asterisk-Dev] Re: writing a GPL G.729? [OT]

[Steve Underwood]

Schaefer, Mark wrote:

>I'd like to see what you say about software patents when you come up with a truly unique and innovative solution in software.  Patents are designed to protect the innovators.  If software innovation is not protected, then there is little incentive to innovate.  As the US Constitution says, one goal of the state is "To promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries"
>
>Without patents, here's what would happen: 
>1) You spend a year designing a brand new algorithm that represents a quantum leap in switch efficiency and throughput.  
>2) You build that algorithm into a new line of switches that perform ten times better than the competition.
>3) Your competitor reverse-engineers your algorithm and puts it into their switches.
>4) No company invests in switch R&D because its a cost center and there is no financial reward.
>5) Despite what Richard Stallman says, you would not see Universities step up to do all R&D.  Do we really think that Universities innovate without thought of financial gain?  MIT has one of the biggest Patent offices of any University.
>
>The only people I see complaining about software patents are the people who don't create anything new.
>  
>
This kind of argument holds no water whatsoever.

First, 99.99% of the world's software is not protected by patents. It 
was still written, though. Much of it still makes lots of people lots of 
money. Patents are demonstrably not necessary to foster innovation in 
software.

Second, patents have degenerated into a land grab, and don't represent 
the fruits of real labour at all. The subject is G.729. Another name for 
that is CS-ACELP  - conjugate structure algebraic code-excited linear 
prediction. The last part - linear prediction - it pretty old. In the 
1930s people like Durbin and Schur worked out efficient algorithms for 
that without any commercial imperative. That was pretty innovative 
stuff. The next bit - code-excited - refers to a technique invented in 
about 1980 by guys at Bell labs. They never patented it. Perhaps the 
huge compute power it needed at the time made them think it was a waste 
of time. They still did *really* ground breaking work on this. Then came 
a few people who never made much of a splash. They reduced the compute 
quite a bit. No patents, and very little recognition. The GSM started to 
get people's attention. The Europeans were actually going to make 
compressed digital voice a mass market thing. There was feeding frenzy. 
Every dumbass technique you can think of was suddenly patented. The 
algebraic sum, and the conjugate parts of G.729 are where all the 
patents are, and they are the least ground breaking aspects of the whole 
codec. There is no solid relationship between innovation and patents. 
Patents are normally worded by lawyers who know how to achieve the 
biggest land grab, without being so broad they won't get the patent 
through. There might not even be any engineering involved.

Regards,
Steve