Patent lawyers don't understand software

[DaBees-Knees]

http://www.theinquirer.net/inquirer/feature/1562231/patent-lawyers-understand-software

AN EXPERT IN COMPUTING has suggested that most patent lawyers can't locate their buttocks with both hands when it comes to the nitty gritty of computing, and that computers quite simply don't work the way that most members of the legal profession presume that they do.

Writing for the technology legal champion Groklaw, a 25-year veteran of the computing industry with a masters degree in computer science known only as PO1R reckons that most legal decisions connected to computing software fail to accord with reality. To explain that he (or she) has written a fascinating and informative paper on the fundamentals of computing and what's wrong with trying to patent software.

The 18,000 word submission is couched in such dense language that most INQ readers will be gnawing on their knuckles by the time they get half way through it, but the arguments put forward by PO1R have far-reaching implications for the future of the software industry in general and one long-running patent case in particular.

In 1997 Bernard L Bilski filed a patent application for a business method of calculating hedging risks in commodities trading. Basically the method was just a mathematical algorithm for predicting the supply of and demand for various commodities and adjusting some predictions about the future prices of those commodities based on the maths.

The patent application was rejected on the grounds that "the invention is not implemented on a specific apparatus and merely manipulates [an] abstract idea and solves a purely mathematical problem without any limitation to a practical application, therefore, the invention is not directed to the technological arts."

That judgment, that all software is an abstract idea and therefore can not be patentable, could be the single most consequential ruling in the history of computing. If that is upheld on appeal before the US Supreme Court, it could invalidate all software patents and create a world where software can no longer be patented, but we are getting ahead of ourselves here.

The ruling, according to PO1R, rests on a few simple principles, although he does then go on to explain these in detail, which some might think makes them far more complicated than they really need to be. The basic rules are as follows:

All software is data

All software is discovered and not invented

All software is abstract

All software is mathematics.
We think what he is saying is that, as mathematics is, by its very nature, pre-existing - in other words, you can't invent the mechanics of mathematics, you can only discover them - and as software is simply, at the end of the day, just mathematics, it is or at least should be totally impossible to patent software.

Which is a bit like telling a best selling author that, as he didn't discover all of the words in his latest book, and that all he really did was put them in a particular order, he has the right to get paid for writing it, but he can't prohibit anyone else from using any of those same words. When put like that, this is just common sense really, innit.

This concept only relates to software at its most basic - or complex, depending upon how you look at it - level. We're talking about the underlying mathematical structure of a piece of software here folks, not the shiny stuff you see on your screen.

But the concepts in the Bilski case are far more fundamental, and it's the fundamentals of mathematics and how they relate to the way computers work that concerns Groklaw member PO1R. He offers up an in-depth guide to computational theory that is almost guaranteed to make your brain hurt, explaining everything from effective methods through formal systems via computable functions, Gödel Numbers and the Church Turing Thesis.

PO1R maintains that in order for lawyers and judges to properly execute their legal duties, they first need to understand the relationship between mathematics and computing. What it all boils down to is that the only difference between a computer and a mathematician with a pencil and a piece of paper is the speed at which they can carry out calculations.

A computer is nothing more than a box of bits that is very quick at mathematical mechanics. And since you can't patent abstract concepts like all those of mathematics, then you can't patent software. Simple.

Now we can watch the patent lawyers jump through all sorts of hoops trying to make software all so very complicated - and patentable, of course - all over again.

An interesting thought there. Let's see if that arguement wins the day in court.   

[Bluey_412]

A computer is nothing more than a device that knows how to count from 0 to 1....

and back again.

But it CAN do it very, very, very quickly...