Today I wrote a comment on an ipWatchdog article where I said:
To say software does not alter the physical machine is wrong. Software is stored as a specific orientation of magnetic particles on a disc, or a specific orientation of physical electrons stored within a memory matrix (RAM, SSD). Software sends specific computing instructions to a processor in the form of precise physical patterns of electrons. Just in the same way, the distributor in a car sends electrons to spark plugs in a specific sequence enabling the engine to run, and most cars now use a computer to control much of this. Further any machine is nothing more than information temporarily conveyed in the arrangement of protons, neutron, and electrons within the spatial matrix of the "reality" of the universe, which is arguably itself a giant quantum computer that we ourselves are merely software running on (see MIT theorist Seth Lloyd's paper, "Universe as quantum computer"). The fact that genes are patentable belies the validity of arguments against software patents since gene sequences are merely biological software! And biology is merely software running on the universe itself.
Re Bilski was denied because they tried to tack "on a computer" to their ideas just to make it patentable. Citing a "data source" does not require a computer: an autistic guy who memorized the phone book is a data source, too. Software that relies on specific things about computers like network protocols, markup languages, complex data structures that are clearly defined by the patent, etc. are things that remain patentable. Saying "a piece of software that accomplishes goal X" is not patentable without thoroughly explaining how that goal is accomplished by the software.