The sewing machine analogy is quite good: Although it's not possible (or
reasonable) to patent the sewing of straight lines it is possible
to patent a novel method of sewing waterproof seams. A good patent specification
for this would keep the type of sewing machine, thread and so on as general as
possible while describing the essential features that made this a new method.
Switching back to software this corresponds to patenting a new algorithm without
describing the computer or listing the code.
'The software is Maths'
discussion is great because it opens up the whole issue of invention and
discovery:
What about a 'mechanical' computing machine such as Babbages?
Surely everyone agrees that the physical implementation of the computer is
irrelevant to this argument, we are interested in 'abstract' software, and
that's still Maths
So what about controllers in the days before
microprocessors? These would often consist of electrical or pneumatic switches,
sensors and valves. Sometimes separate components joined by wire or tubes but
sometimes all built into a single device. A good example would be an old
fashioned carburettor.
I doubt they get taught any more but when I
studied engineering we learnt about 'analogue' computers. These were devices
that used some physical properties or behaviour to model an otherwise
intractable calculation (actually, a modern equivalent would be the use of
quantum properties to calculate prime factors of very large numbers)
Now
imagine that two people (independently) invent a new carburettor that is a
dramatic improvement over anything in existence. It's main feature is responding
intelligently to various inputs and adjusting air and fuel supply to the engine.
One person devises a clever, complicated, analogue method for doing this and
files a patent. The other sees that the same result can be obtained using the
existing sensors and engine management computer but with an exquisitely
sophisticated modification to the software. Surely everyone would agree that the
'black box' principle should apply, both inventions are essentially getting the
same result so if the first one is patentable the second one ought to
be.
However just before she files her patent the second inventor realizes
that the software contains a brilliant new algorithm that represents a
fundamental mathematical breakthrough and as a byproduct proves that p equals
np. The commercial value, and benefit to mankind, of this new 'invention' make
existing audio and video codex look like chicken feed. However it is simply the
discovery of a mathematical proof so surely not patentable. The only thing she
can do is go through all the possible practical uses of her algorithm and file
patents for each. [ Reply to This | Parent | # ]
| 
