On Thursday 23 September 2010 00:05:05 A. Mani wrote: > On Wed, Sep 22, 2010 at 9:00 PM, jtd <[email protected]> wrote: > > an excellent article and subsequent discussion on why software is > > maths. The article is in prepartion for presenting to the USPTO > > waht constitutes an abstract idea. > > > > http://www.groklaw.net/article.php?story=20100915004250723 > > > > The gist of the argument is that math is a method for processing > > symbols (representations of data) and the resultant is always > > symbols (or representations of data). This activity of processing > > symbols can be done simply by paper and pencil. Software does > > exactly the same processing of symbols, replacing paper and > > pencil, with a programmable machine, with the end result always > > being more symbols.
Was hoping you would join in the discussion ;-) > > Your summary does not seem to make things clear. > Mathematics is basically a form of formal reasoning. In fact all > formal reasoning is mathematics. > Computable mathematics forms a part of it. so is there some math which cannot be computed? > Computable mathematics in a practical 'feasible' sense is > representable as software. > The latter would again be a representation of reasoning > irrespective of the process of making it (man made or machine > made). Ok > > Softwares are necessarily representation of reasoning > processes/schemes. This is one very intuitive way to convey the relationship between math and software. > These will always be copies of known things in > AI, day to day life, human interaction etc in mathematical terms. But you can patent devices which do known things eg. device to make sound. Thus if one were to create a device which mimics the larynx and does not use a diaphragm, which includes some hardware and some software, it would be patentable. Although individually none of the hardware components are unique. This particular aspect is exploited by separating the non unique hardware and claiming uniqueness for the software to cause the hardware to do something unique. > Even if people were to do research work using programming language > primitives, it would still be the same. > > Any expert system capable of genuine learning (definition?) would > need different kind of laws. No patents should ever be allowed on > such things. AI is not a well defined concept. That is a scenario in an ever worsening situation. If the current situation of permitting software patents continue, AI (software) systems would indeed become patentable. > > In the world of western law, things are far worse. They patent > everything based on the absurd principle that 'all new physical > /abstract? objects/ideas are patentable'. All patents are evil. Indeed a deeper check on some famous patent holders like Edison, indicate a blatant gaming of the system. Given the current means of creating knowledge using computers, much of the old arguments of return on investments, publishing of knowledge etc. are absurd. > While arguing against s/w patents, nobody should try justifying > other forms of patents. But in the current scenario, we are trying to distinguish between greater and lesser evils. so to rephrase my first post and tighten the noose The gist of the argument is that math is a method for formal reasoning. Presentation of the reasoning requires processing symbols, (representations of data) and the resultant is always symbols (or representations of data). This activity of reasoning and processing symbols can be done simply by paper and pencil. Software follows exactly the same process of reasoning and processing of symbols, replacing paper and pencil, with a programmable machine, with the end result always being more symbols. -- Rgds JTD _______________________________________________ network mailing list [email protected] http://lists.fosscom.in/listinfo.cgi/network-fosscom.in
