🚀 go-pugleaf

In article <9008021406.AA24725@stc06.CTD.ORNL.GOV> de5@STC06.CTD.ORNL.GOV (SILL D E) writes:
>In article <1990Aug1.223920.21363@ico.isc.com> rcd@ico.isc.com writes:
>>
>>I think that escher is saying that it may not be easy to anticipate what an
>>algorithm can be used for.
>
>What does being able to anticipate possible end-uses of an algorithm
>have to do with patenting them?  ...

Patents are supposed to be written as specifically as possible.  There is
something of a game one plays with the patent office, in which one tries
to write as broad a patent as possible, and the patent office makes you
narrow the scope of the patent.  The reason for the game is that if a
patent is too specific, one can make a trivial change to it and not violate
it.  If the patent is too broad, the Patent Office would be granting you
a monopoly over things you did not invent, and be stifling innovation.

Software patents are specifically supposed to be given only to those 
algorithms which are directly related to the control of a machine.  Something
like a spreadsheet calculating algorithm, in my understanding, could not
be calculated.  Now suppose one invented the topological sort while developing
a fishing net weaving machine.  One would appropriately get the patent for that
algorithm.  Would using that algorithm for object-library directories
be a violation?  It would depend on how that particular patent was written.
However, given that algorithms are patentable only to the extent they
control the function of physical machinery, is the use of patented algorithms
purely for information processing a violation.

>precisely specified and predictable ways; that manipulation is the
>function of the algorithm.  A sorting algorithm orders data.  Whether
>the data is personnel records or pixels, the algorithm is performing
>the same function. 

The courts don't see that algorithms don't manipulate data; they see it
as manipulating concrete devices.  I think a fundamental problem with the
courts and the patent office is that they don't understand the abstract
concept of data.  They see gears and levers and electrons; we see data
structures.

I think it also does not take much work to devise a machine for a given
algorithm to control, if one is more interested in protecting the algorithm
than the machine.

>well-defined, and it doesn't change the basic function.  BTW, what's a
>topological sort?

I don't know, but the unix program 'tsort' does it.  When used for 
ordering object modules in a library, it arranges them so that those 
who reference are placed after those who define.
-- 
Michael D. Crawford
Oddball Enterprises		Consulting for Apple Computer Inc.
606 Modesto Avenue		escher@apple.com
Santa Cruz, CA 95060		Applelink: escher@apple.com@INTERNET#
oddball!mike@ucscc.ucsc.edu	The opinions expressed here are solely my own.

		alias make '/bin/make & rn'