I think you intended the last item on your list to be pure software, but it's not.
You have hit the crux of the discussion. I am defining software as a tangible thing, existing as a pattern of matter and fields. It is not abstract, virtual, or ineffable. What you are talking about ("pure software") are ideas about the software's principles of operation. The mathematical and logical ideas by which we understand a process are not patentable, but their tangible embodiment (software) is patentable.
Consider an example of the difference. Prime numbers are abstract ideas that can be used to create difficult-to-solve problems, and thereby used for cryptography. As abstract ideas, they are not patentable. Someone could embody those ideas in a logic machine (like the RSA algorithm) and patent that. But somebody else could embody those ideas in an analog laser interferometer and patent that, without conflict with the other patent.
No one is arguing that a complete electronic timer should not be patentable.
And that is exactly what is covered by a valid software patent. Such patents claim something like "Claim 1: A stored-program logic machine, configured to provide the process described in claim 2." The configuration of the logic machine is the software.
A well-written patent will cover all variations of the software that are "obvious" to "a person skilled in the art". This generality is why all patents are so painful to evade, not just software patents.
> "I am defining software as a tangible thing, existing as a pattern of matter and fields."
Ok, but that means you're not talking about software patents any more. The U.S. patent office currently cares nothing about patterns of matter and fields.
> "And that [a complete physical timer] is exactly what is covered by a valid software patent."
No, wrong. That's how it should be, but it's not, and that's the problem. Officially, patent law isn't supposed to allow patents on "abstract ideas", but due to incomplete guidance from the courts, the requirements have been diluted to nothing. Right now in the U.S., a software patent application can satisfy the requirements by specifying any trivial physical step. Most software patents specify only a "computing device" on which the software is used. Well gee whiz, guess what that covers? You got it, everything! It's useless language; there's no narrowing, no added specificity. It would be just as usefull to specify "software in Her Majesty the Queen's service". As a result, for the past 20 years the USPTO has essentially been granting patents on pure software.
A patent on a laser interferometer that encodes instructions is no longer a software patent; that would be covered by a mechanical patent. And this patent certainly would be in conflict with any software patents that covered the software encoded thereby, if the interferometer were ever put to use in a product.
The U.S. patent office currently cares nothing about patterns of matter and fields.
Right now in the U.S., a software patent application can satisfy the requirements by specifying any trivial physical step. Most software patents specify only a "computing device" on which the software is used.
So which is it? Physical embodiment = not abstract.
A patent on a laser interferometer that encodes instructions is no longer a software patent; ...
I was talking about interference machines that use the wave properties of light to do computation without the use of digital logic. With these machines, abstract mathematical approaches are reduced to practice by means other than software. The software patent does not somehow magically fence off an area of abstract thought. What it fences off is a particularly convenient and valuable area of concrete machinery.
> "So which is it? Physical embodiment = not abstract."
No, the point is that there effectively is no "physical embodiment" requirement. Adding magic words like "computing device" or "system and method" to a patent application has no real effect, because ALL software is run on a "computing device" or a "system". This does nothing to narrow the scope of a patent, or to tie it to any distinct physical implementation. The ONLY distinct entity in such a patent is the algorithm.
This will become immediately clear to you if you read any number of software or business method patents. You seem to have created your own alternate reality here, where software patents are necessarily like mechanical patents. They are not.
> "I was talking about interference machines that use the wave properties of light to do computation without the use of digital logic."
Fine, whatever. Any patent that covers your thingamajig in its physical manifestation will have to be a proper mechanical patent, complete with design drawings. As a separate matter, the computation performed by the machine can be expressed algorithmically, and that algorithm is what could be covered by a software patent, completely apart from whatever machine implements it. That patent can then be asserted against any other use of that algorithm regardless of the physical system that uses it.
> "The software patent does not somehow magically fence off an area of abstract thought. What it fences off is a particularly convenient and valuable area of concrete machinery."
No, again, that would be a mechanical patent. Have you really not ever read a software patent? Go look one up; maybe start with one of the patents on LZW, or maybe amazon's 1-click patent.
You have hit the crux of the discussion. I am defining software as a tangible thing, existing as a pattern of matter and fields. It is not abstract, virtual, or ineffable. What you are talking about ("pure software") are ideas about the software's principles of operation. The mathematical and logical ideas by which we understand a process are not patentable, but their tangible embodiment (software) is patentable.
Consider an example of the difference. Prime numbers are abstract ideas that can be used to create difficult-to-solve problems, and thereby used for cryptography. As abstract ideas, they are not patentable. Someone could embody those ideas in a logic machine (like the RSA algorithm) and patent that. But somebody else could embody those ideas in an analog laser interferometer and patent that, without conflict with the other patent.
No one is arguing that a complete electronic timer should not be patentable.
And that is exactly what is covered by a valid software patent. Such patents claim something like "Claim 1: A stored-program logic machine, configured to provide the process described in claim 2." The configuration of the logic machine is the software.
A well-written patent will cover all variations of the software that are "obvious" to "a person skilled in the art". This generality is why all patents are so painful to evade, not just software patents.