On Tuesday, the United States Court of Appeals for the Federal Circuit rejected a patent on a method of detecting credit card fraud. The result was unsurprising, but the court broke new ground with its reasoning. Citing the Supreme Court’s famous rulings against software patents from the 1970s, the court ruled that you can’t patent mental processes—even if they are carried out by a computer program.
Of course, all computer programs implement mathematical algorithms that could, in principle, be implemented with a pencil and paper. So is this the end of software patents? Unfortunately not. The court ruled that the no-patenting-math rule doesn’t apply if the math in question complicated enough that "as a practical matter, the use of a computer is required" to perform the calculations.
In order to justify this result, the court gives the most thorough defense of software patents that we’ve ever seen from the judiciary. We don’t think the line they draw—between ordinary math and math that requires a computer—makes much sense from either a legal or policy perspective. But the ruling at least signals that, for the first time in over a decade, the courts are thinking hard about how to apply the Supreme Court’s old software patent cases in the modern world. We’re hopeful that as the confusion in this week’s decision becomes more obvious, we’ll see further progress.
via Ars Technica – Does not compute: court says only hard math is patentable. It’s nice to see the courts limit patents somewhat, but the logic still has a real problem which is that practically speaking all math can be performed by a human being it may just either be tedious or time-consuming. The other question becomes raised if the math is complicated enough that a computer becomes required does that mean that the math itself is patentable (which the courts have said no math isn’t patentable)? The court seems to be trying to not rule against all software patents while acknowledging they are broken and need to be reformed.
The core of the legal problem with software patents is that they are just algorithms, logic and math, neither of which is patentable. Combine the two and describe a possible computer program and bam, that logic and math is now patentable. Ignoring all practical aspects of patents and software patents in particular, legally speaking software patents seem to me to be indefensible.