To view parent comment, click here.

To read all comments associated with this story, please click here.

A trick question indeed I'm aware that programs can be proven with techniques like induction, axiomatic semantics, etc. Then again, I can also prove properties of physics using mathematics, and physics are at the heart of many "solid" inventions.

I will dodge your question slightly with my motivation for such an underlying "feeling": I have some discomfort with the notion that if I write down an algorithm in software it's not patentable, but if I write it down in VHDL and drop it onto a circuit, it is.

An analogue filter using resistors and wires may be patented (let's assume), but a software based one cannot? To me, this would seem to be a rather arbitrary distinction, the key innovations are in the clever arrangements!

I don't have an answer for you about the fundamental difference, unfortunately, but maybe if I thought on it a while longer

saynte,

"I'm aware that programs can be proven with techniques like induction, axiomatic semantics, etc. Then again, I can also prove properties of physics using mathematics..."

Really? I'm still under the impression that no single mathematical model has ever addressed all observable phenomena. A hypothesis on the properties of physics lives or dies by whether or not it stands up to physical evidence, not whether or not it's mathematically correct.

For example, newtonian physics is mathematically sound, but it doesn't fit the physical world at the extremes.

But maybe I'm wrong and I missed some big new developments in the theory of everything, in which case I'd certainly appreciate a link to more information.

Let me know if I misinterpreted what you were trying to say.

"I have some discomfort with the notion that if I write down an algorithm in software it's not patentable, but if I write it down in VHDL and drop it onto a circuit, it is."

I can understand that.

"An analogue filter using resistors and wires may be patented (let's assume), but a software based one cannot? To me, this would seem to be a rather arbitrary distinction, the key innovations are in the clever arrangements!"

I would have expected that any R/C network patents have long passed their expiration dates, but it is not really my domain at all. I don't know how much work it is to establish a fabrication plant, or the effort required to minimize interference and leakage between components, etc... I'm just not comfortable making claims about electronic patents.

The reason I focus on software patents is because that's my expertise. People sometimes read into it too much.

Member since:

2011-01-28

saynte,

"Software is an arrangement of 0s and 1s; a car is just an arrangement of metal and rubber. I think the question should be: was the arrangement found, or built? Proper mathematics is found, but software is most definitely built."

I don't agree with your distinction.

You may not be aware of this, but in computer science discrete mathematics courses, we do study how to apply mathematical concepts like induction towards computer algorithms.

If an algorithm is patented, and a developer can prove that the algorithm is mathematically derivable, then would you say the algorithm patent should be invalidated? It is sort of a trick question, since every algorithm is mathematically derivable, given adequate specification.

I'd go as far as to say a genuine distinction between mathematical and computer algorithms is blurred to the point of non-existence seeing as one can clearly be translated to the other (within physical constraints of the machine).

Could you elaborate on a fundamental difference?

Given the same problem, many developers will come up with overlapping algorithms. You may say "oh just use a different algorithm", but now developers are wasting their time in search for algorithms with less desirable properties in order to satisfy a patent holder's monopoly. Considering that many patent holders deliberately file dozens of variations on the same idea to deliberately block other implementations, it's no wonder honest developers are pissed off with software patents.

Edited 2011-07-19 06:24 UTC