Artefacts

The concept of "software patent" fails under certain circumstances for the same reasons as the reasons which make thoughts or social behaviours morally difficult to patent.

Thanks to "artificial life" technology, some computer methods or techniques can be invented by computers with no human interaction. Users for example specify a functionality, such as adding two numbers, and computers will generate an algorithm which allows to add two numbers (this experiment has been successfully at ATR labs in Kyoto). This generated algorithm is invented by virtual agents growing in a massively parallel calculator. It has all the characteristics of an invention. It can be described, reproduced, etc. Who does it belong to then, to the man or the machine? And, in the case a computer invents some kind of technique at run time, no human beings will ever check or try to understand the fundamental principles of this invention, thus making patent infringement impossible to prove for exactly the same reason as it impossible to read in someone's mind.

A similar problem happens on the Internet. It is possible to patent, as a business method or as a technique, a combination of existing services on the Internet. One may then wonder whether the use in combination of various Internet services, something everyone does everyday, may be sometimes a patent infringement. And in case it is, who will ever be able to prove it ? And, does it change something if this use is achieved by a human operator or by a software written in house to automate what human operators do?

Artificial Life Workshop at ATR, Kansai Japan

This Kahaner report (Dr. Kahaner used be paid by the US Army to travel in Japan and write nice public reports on Japanese R&D. Such reports mostly contain valuable information and, sometimes, also contain misleading information) includes a short presentation of Artificial Life experiments in which computers invent new things.

http://www.atip.or.jp/ATIP/public/atip.reports.93/al.93.html

Surviving a War with Patents. Kevin G. Rivette

This article, written by a patent attorney, is much in favour of software patents, includes many examples and, at the end, shows that networking Internet services may generate contributory patent infringements which can not be detected.

http://www.upside.com/texis/mvm/opinion/story?id=382b10570