I would not be unhappy if software, or software+computer, or adding
software+computer to something else, was legislated or ruled to be not subject
to patents, for the simple reason that most software patents are blindingly
obvious and are obviously a drag on the economy.
But I really don't believe you get there by arguing that software+computer does
not make machine equivalent to dedicated hardware, or cannot perform the same
process as dedicated hardware. You might nibble around the edges on the
business or screen bounce patents, but most of those should be covered by prior
art anyway.
Absent some stunningly surprising ruling by the Supremes that software patents
obviously don't promote progress, and absent some stunningly surprising informed
legislation, I think the best bets are:
- Focus on insuring that the thing that is patented is a particular
implementation.
- Focus on insuring that patents are properly categorized, and are readable by
engineers, and that there is no penalty for engineers reading them.
- Focus on obviousness, starting with "+ a computer" was obvious to
anyone 20 years ago.
- Focus on prior art.
- Focus on methods that can help to hobble patent trolls, e.g. reduction to
practice, loser pays, etc.
There are more bad patents than just software, and this will help them all out.
[ Reply to This | Parent | # ]
|
But some of us believe that a combination of hardware and software
can create a new machine, or perform a new process. I know you do not, and
believe it is a fact that this is not true, but to me, your "fact" is still an
opinion, and one that is not actually backed by solid logic. If a
computer+program logically performs the same task in the same fashion as
dedicated hardware, then if one is a patentable new machine, the other is an
equivalent machine.
Just to be clear, I have not said a
"combination of hardware and software" can't make a new machine or a new
process. You can make a new anti-lock brake system or a new process for curing
rubber in this manner. My point is narrower. I say programming a computer
doesn't make a new machine. This is based on sound logic and on the principles
of computer science but I have not developed the argument in this thread. I have
provided a link where the evidence is presented.
Your equivalent machine
argument is not sound logic because it discounts the possibility that the same
functions could be achieved by means other than making a new machine.
I don't see any problem with leaving people free
from performing the same functions using other means. Please do not conflate the
circuit with the algorithm which could be read in the source code.
I do not. But I do conflate the circuit with an equivalent
execution running on a computer, and so (so far) do most of the
courts.
Where is the difference? You have argued an expression in
Verilog could easily be translated in a programming language. Isn't this the
expression of the algorithm computed by the circuit? When this algorithm is
given to Verilog a circuit is made. When this same algorithm is given to a
compiler software is made. This conflation you admit making is not different
from equating the algorithm with the circuit. There is no difference that I can
see.
But (IMO) a program running on a computer that executes the
same function as otherwise patentable hardware, should equally be patentable.
Why? The function is not patentable. Why the possibility of
making one patentable implementation imply that all implementations are
patentable?
Nobody ever claims a new computer has been invented.
What they claim is that a special purpose machine has been
invented.
Where is the difference? The words you use are
different, their meaning is the same. The courts have said a programmed computer
is structurally different from one which is not programmed. That special purpose
machine is a programmed computer. This is a new computer different from the
unprogrammed computer. So yes, people are arguing a new computer is being
made.
Which the Supreme Court
backtracked on in Diamond (foreshadowed by the dissent in Flook).
No they didn't backtrack. They reaffirmed Benson and Flook in
Diehr. They reaffirmed these cases again in Bilski and Mayo. The rubber curing
patent was allowed precisely because there was more than just the calculations.
It included the step of curing the rubber.
There were physical
steps in the earlier cases too, though perhaps not pled well enough.
This is precisely the point I was making. Please recall, I have
used these cases as examples of mathematical calculations that were held patent
ineligible abstract mathematical ideas even though they were physically
executed. This was meant to be a pair of counterexamples to your argument that
actually doing the calculation is patentable "applied math".
This happens only if you insist that actually doing the
calculation is "applied math". If you just accept the mathematical truth, that
doing the calculation is just math and the application is whatever the number
means in the real world then the "on a computer" magical pixie dust doesn't
happen.
Exactly. Which is why it will be replaced wih the "and
for any application that could make use of a similar calculation" pixie
dust.
Is there really such a pixie dust? Putting this language is
equivalent to admitting the claim is not on a specific application but on the
mathematical calculation itself.
Then there is a long section where you
seem lost by my comparison of software with a book. I think this next quote goes
to the heart of the miscommunication.
In a
configured printing press and a programmed computer. the invention is defined in
terms of symbols and their meanings. This is not the case with an injection
molding machine.
Again, which are you talking about? The book is
not an invention. Symbols are used in describing the patent for curing rubber,
as they are in all patents.
The symbols I talk about are not in
the text of the patent. They are the bits in the computer. If you read the
textbook on the math of software you will find that algorithms are procedure for
manipulating symbols. They solve problems associated with the meanings of the
symbols. If you read textbooks of computer architecture you will find how the
bits are used as symbols to represent numbers, boolean values and carry out
operations of arithmetic and logic. And if you read textbooks on the foundations
of mathematics, you will find the definition of the language of mathematics, its
syntax and its semantics. The math of computing are derived from this basis.
The functions of a computer is to manipulate symbols, the bits, according
to the rules of mathematics. The problems which are solved are associated with
the meanings of the numbers, boolean values etc. I compare a computer with a
printing press because they are both devices which manipulate symbols with
meanings.
The point of the printing press analogy is to explain how
semantical relationships work and how they relate to the functions of software.
In a book, the normal semantical relationship is the ink represent letters and
the letters have meaning. But a typographer can reverse this relationship. He
takes the novel as the series of letter that must be printed and this described
how the ink must be laid out on paper.
Something similar occurs with
software. The normal semantical relationship is that the voltages represent the
symbols, the bits, and the bits represent numbers. Finally the numbers mean
stuff like payroll data or space shuttle speed and position. But in a patent
this relationship is reversed, like for a typographer. The functions of software
are disclosed and claimed. This is a recitation of the meaning of the data and
the mathematical or logical operations which must be applied. This is the
meaning of the numbers and boolean values. The programmer is supposed to write
the corresponding code from this description. Then the compiler will generate
binary code. All this is supposed to be a description of the process by which
transistors are turned on and off to manipulate voltages. This whole thing is
the digital equivalent of saying a mathematical formula is the description of a
process for pushing a pencil on paper because the calculating procedure can be
inferred from the formula. This reverses the normal semantical
relationship.
Why does that matter? It is because this reversal of the
semantics is not acknowledged by the law. If we transpose to a printing press to
arguments made about computers, it it clear they are bogus. People react like
you do, saying the book is not an invention etc. But when it comes to a
computer, the reversed relationship is treated like the normal one. The
algorithm is treated like a description of the computer similar to the laws of
physics. But it is not similar to the laws of physics. A physicist uses the
normal semantical relationship, from the ink to the symbols, from symbols to the
numbers and from the numbers to the real world. A patent lawyer uses the reverse
semantical relationship from the real world to the numbers, then from the
numbers to the bits and from the bits to the voltages. This is why they are not
the same. A programmer will point to that difference by saying the computer
manipulates the symbols. They say software is math because this manipulation is
math. It is not something described by math like the laws of physics.
The
point of the printing press parody is to show that when we reverse the
semantical relationship of a book, the same new machine argument can be made.
And it is bogus for both the printing press and the computer for the same
reason: the normal semantical relationship is not taken into account. If we do
take it into account in both cases we find there is no patentable invention
because the innovation is in the meaning of the symbols. It is not in their
physical representation. I argue that this is a problem with the current case
law. A lot of my points are much easier to understand if you keep in mind that
in my view the normal semantical relationship is the normal one and patent law
uses the reverse one. Logic flows very differently when this dichotomy is
acknowledged.
As a final note I have seen you other post where you clarify
your beliefs. It is quite possible that there are more problems with patent law
than those I denounce. I am fine with people trying to fix these problems. I
just insist that the problem that I see are addressed as well.
[ Reply to This | Parent | # ]
|