Paper 3: Patents
Due: Dec 16, 2011 Comp 317/417,
Dordal
Are software patents a special case?
There are actually two issues; you can focus on both or else either one alone:
- Are software patents a special case?
- Are software patents beneficial, on the whole, to the economy?
If the answer to the second question is "yes", then one might be inclined to support software patents even if they were
a different case from physical-machine patents. If the answer to the
second question is "no" or even "maybe not", however, then the first
question warrants careful consideration.
Paul Graham wrote
There's nothing special about physical [machines] that should make them patentable, and the software equivalent not.
Argue your position, and come to some sort of
conclusion. Your conclusion should suggest some broad position on the
usefulness of software patents themselves.
The purpose of patents is to encourage innovation and investment in
innovation, for social progress. Does this principle apply to software
as well as machinery and pharmaceuticals? Are software patents, on the whole, of benefit to
society? Do they foster innovation, leading to more
software ideas for everyone? For the purposes of patent law, should software be included?
Here are a few justifications advanced for considering software patents to be fundamentally different:
- Software is often about abstraction, and abstract ideas
- Software has greater generality than physical machines
- Algorithms are fundamentally mathematical factsrather than inventions
- Many software "inventions" are straightforward
implementations of broad ideas
- Often the "creativity" in software is simply realizing that, yes, there is a market for the product
Others argue that the only specialness of software patents is that the
patent office allegedly has difficulty conducting patent examinations
for software. Still others argue that open source makes software patents unique.
In the I4I v Microsoft case, an obscure patent about a data structure
for holding XML tags became very expensive for Microsoft, once the data
structure in question was interpreted so broadly as to cover any method
of manipulating XML structure separately from data. Does this represent a fundamental problem with the use
of abstraction (ie in describing data structures) in computer science?
Or was this just a quirk of the judicial system?
The Supreme Court decided the Bilski
case in June 2010 and agreed that some abstract ideas are not patentable. Does this affect software patents?
Yet another way to address the two questions above is that, yes, Graham was right, but that software should be
subject to different patent rules. If this is what you feel, propose different rules and argue for them. If your main argument
is that software is indeed a special case, you may either
make the case that it is so different that patents should not apply, or
make the weaker argument that special patent rules should apply. If
you are arguing that software is not a special case, you do not
necessarily have to argue that all the rules should be the same.
Discuss both sides, and come to a conclusion.
Your conclusion should either support one side or the other (perhaps
with qualifications), or else it should outline some sort of
"compromise" position.
If you are arguing against
patents generally, please make that clear (in that case, you are probably not
arguing that software is different, though you might argue that the
case against software patents is stronger). If you argue against all
patents, you should be sure you understand your argument's implications
for, say, the
pharmaceuticals industry.
Keep in mind the following points (though you don't have to address them all):
- Incentives to innovation and development
- Litigation costs as a negative social side-effect
- Inventor's rights
- Public's rights to shared ideas
- Conflicts between patented software and open source
- Whether software is "too abstract" to be patentable in general (Bilski's process was held to be too abstract)
- Many claimed software inventions are straightforward implementations using existing components
Expected length: 3-5 pages, as usual