S.32:
Friedman (1998) observes that "a single program incorporates many different
algorithms, some invented by the programmer, some borrowed from general practice
and some perhaps deliberately copied from a known originator. If algorithms are 
private property the costs of figuring out which of the ones you are using 
belong to whom and negotiating the necessary licences may be high. So the 
argument against making algorithms private property is similar to the argument 
against making words private property", (Friedman p.379). None of these effects 
is unique to the software industry as Varian and Schapiro (1999) point out. 
However they would appear to be particularly pronounced in this area. Many 
industries experience incremental innovation but the effect noted by Friedman is
thought to be especially pronounced in this area. Friedman also argues that
"programming exhibits extreme diseconomies of scale" and argues that these two 
factors alone "argue against patent protection" for software. If it is true that 
much software can be developed or at least incrementally improved in socially 
useful ways at relatively low cost, the force of the argument that we need 
strong protection to ensure innovation is diminished. The fact remains that even 
if this is true the costs of bringing products to market remain high. 
Nevertheless, many authors stress the fact that much of the basis for current 
generations of software was laid in a period before US firms became accustomed 
to patenting software, and indeed the high valuations placed on "dot.com" share 
offerings currently suggest that firms can sometimes raise money without being 
able to patent business methods.

*FRIEDMAN, D. "Computer Law", New Palgrave Dictionary of Economics and the Law,
ed. P Newman, Macmillan, 1998