Read the article here: What is a patent? Join the debate

Here's my opinion as a non-lawyer. The US Constitution says Congress has the power "To promote the progress of science and useful arts, by securing for limited times to authors and inventors the exclusive right to their respective writings and discoveries". So if a patent "promotes the progress of science and useful arts", it is constitutional. However, if its purpose or misuse holds back progress, in my opinion it is unconstitutional and therefore invalid.

In general, patents do not "promote the progress of science and useful arts" since practitioners are advised by legal staff to never, ever read patents since that could result in damages for "willful infringement". So practitioners just go ahead and design things and hope for the best. They will probably end up infringing out of ignorance because there are only so many ways do to a thing. Unfortunately, independent invention is considered by the courts to be infringement, even though in my opinion it should not be because the patent did not in that case "promote the progress of science and useful arts".

Red Hat's comment that many (if not most) of the great innovations in software occurred before software patents is extremely important. Obviously software patents were not required to "promote the progress of science and the useful arts" in the past. Copyrights alone were sufficient, and still are.

If every software developer needed to know about and consider all the software patents active or in USPTO's queue, software development would be paralyzed. This is hardly "promoting the progress of science and useful arts". What's worse is that practitioners are not even allowed to consider whether they might be infringing: that requires being a patent lawyer.

So here's the test I'd like to see: does a patent and/or its use by its owners "promote the progress of science and useful arts"? In my opinion, software patents never do so.

I think that it's reasonable for Patent Law to disallow, loosely speaking, abstract ideas, and therefore require that an invention have a tangible embodiment. Without that restriction, IP Law will eventually degenerate into not much more than a branch of Philosophy. Perhaps there does need to be a little more liberalization as to what precisely constitutes a tangible embodiment.

A patent's basic function is to preserve the ability to obtain a return on investment in exchange for 'teaching' the community. Inherent in this is some measure of the investment required for reduction to practice. In general, software requires far less investment than hardware to demonstrate functionality and for commercialization (not counting marketing). Business methods are a step further along this same direction. Both are also far more portable than are typical hardware inventions. I am not sure whether software and/or business methods should be patentable but if so they should be classified and regulated separately. For example, they should have far shorter expiration terms than hardware (maybe 5 years).

Here's my take on Bill's question. These are my opinions as a hardware (mostly digital) and software engineer -- not a lawyer.

"Could those hardware engineer or software engineer, for the matter, who claim that only hardware can be patented, but not any software please answer a few questions. Let's limit the technology scope to electronic and computer industry."

What is your definition of hardware that can be patented?

1. Novel physical device (new type of transistor, for example):

Yes, provided it is novel and not an obvious application of physical principles. The test should be whether a great deal of effort was required to discover the device, effort that needs to be protected a patent to "promote the progress of science and useful arts".

However, the physical principles themselves should not be patentable.

2. Novel processing technology

Yes, with the same restrictions as (1).

3. Novel material

Yes, with the same restrictions as (1).

4. Novel chip package.

Yes, with the same restrictions as (1).

5. Novel electronic design (chip, circuitry, board etc) to make transistors do something useful.

Here we have to be very careful. If the circuit is truly novel, e.g., a truly new way to build a content-adressable memory, then yes. However, if the circuit is merely a remapping of an old technology (e.g., relay logic) into a new type of device (e.g., CMOS), then it's not novel. However, a patent examiner may not be aware of Shannon's 1938 paper that describes relay logic and award a patent by mistake. In all cases, (1) applies as well.

Mapping a complex machine (e.g., a computer) into logic or circuits is most likely not patentable, unless some entirely new circuits are invented. The machine itself may be patentable if it performs a novel function. However, in all cases the implementation can be copyrighted and if it's priced attractively people will licence it instead of going to expense and risk of implementing it themselves.

Do you think the following is hardware design:

1. CDMA algorithm

No, it's pure mathematics applied to communications theory. The basic idea behind CDMA is Walsh functions, from 1923. However, a machine that uses the CDMA algorithm for actual communication would be patentable if novel. However, since CDMA derives from work by George Antheil and Hedy Lamarr in 1941, it cannot be considered novel. Indeed, the idea of everybody talking at once has long been the basis behind the New England Town Meeting.

2. CDMA implemented by ASIC (after analog demodulation)

It's hardware design, but except for clever circuits the ASIC cannot be patented. So anyone can take the CDMA algorithm (if indeed the USPTO ruled correctly and determined that it's not patentable) and synthesize their own ASIC provided they use standard circuits and not the clever patented ones. However, all the ASIC implementations are protected by copyright, so if someone has a much better ASIC implementation and prices it well they can make plenty of money through licensing.

3. CDMA implemented by DSP (after analog demodulation)

Totally software design, and should not be patentable. However, the software is protected by copyright and an obviously superior implementation can make plenty of money through licensing.

Do you think CPU design, such as novel cache design, can be patented?

Yes, and it happens all the time. And that's appropriate, provided that the patents "promote the progress of science and useful arts". If it took a great deal of effort to invent the novel cache design, that effort should be protected. However, if others making a modest effort come up with the same design, then the patent is not valid.

Do you think any electronic device based on transistor, circuitry, and etc can be patented, regardless of their functionality?

Yes, provided that the functionality is truly novel and that patenting will "promote the progress of science and useful arts". However, if the functionality is something that others could come up with with a modest amount of effort, then the patent is not necessary to "promote the progress of science and useful arts". If the functionality is simple but takes a great deal of effort to implement, then copyright is more than adequate to protect the inventor from infringement.

Patents should be a last resort, to be used only when an idea is very expensive to develop but is easily reverse-engineered and it's easy to make an independent copy without copyright infringement. Large programs are very expensive to design and debug, and there is no easy way to copy the functionality without copyright infringement. If it's easy to duplicate the software's functionality without copyright infringement, then the original developers do not require patents to protect their development efforts to "promote the progress of science and useful arts".

Again, just my opinions as a non-lawyer.

Betajet:

You certainly have some interest points. For example, a clever CAM circuit design could be patentable but not a clever algorithm, such as search cannot, even though both are making some thing more faster and efficient. Or a new material/circuitry make nonvolatile memory density high but not a new compression algorithm can store digital content using even less nonvolatile storage.

Behind either a novel circuitry or an algorithm is the same human ingenuity and effort. Neither is less in impact on the society. Both are useful arts. Both promote progress in science and technology.

Today, people can afford to enjoy crispy high definition contents at home, thanks to all technologies involved, where video/audio compression algorithm are one of the most important piece, if not the most important one. Tell the MPEG-2 LA, Dolby Digital algorithm are not patentable.

Not all algorithms are pure mathematical expression. They often involve doing something in certain steps. If it could be copyright, it is more harmful (to the general public who want use it for free) since copyright last much longer than than patent. If one published a novel, another one published a novel later with very similar storyline or contains similar storyline, the later is infringed the former even if the words used are differently or it is translated to another language.

Any vote on copyright to protect algorithm?

Patents should not only protect investment but innovation in general. The innovator must be protected against copycats or else originality is insufficiently rewarding. Many patents are not really innovative but just claim an obvious next step as IP. Such patents do not serve the interest of science and technology and may actually hamper progress. The challenge is to legally distinguish 'true' from 'obvious' innovation.

I'm a patent author who has worked at large and small software companies.

Issued patents are of wildly uneven quality. Every piece of software one could imagine creating is in violation of many previously issued patents. Large companies like Intel and Microsoft tell their employees to never look in a patent database, because if anyone thought to attempt to track down all the applicable patent holders for any real product and enter the legal negotiations to license them all, it would be prohibitively expensive and effectively impossible. So it's best to close your eyes and at least avoid treble damages.

And proactive licensing in this way is absolutely impossible for small entrepreneurs.

Rather, large companies rely on "mutually assured destruction" to bring other large companies to the table. Then they pay out hundreds of millions of dollars to small, zombie companies whose products failed and now have nothing to lose by playing the patent "hold up" game.

While small companies that ship product are forced to play Russian roulette, fearing patent action at any time (although the lawyers won't actually knock until they are successful).

And small companies don't have the financial resources to prosecute their own patents. So they simply don't work as protection against the big guy. If they sell anything at all, their greater concern would be triggering the large company to use their larger portfolio against them.

As an entrepreneur who has seen all sides of how patents play out for companies - it’s clear the patent system isn't working as intended in the software space. The system can become completely cynical and dysfunctional.

It's different in other areas (like Pharma). But generally, the patent system is in desperate need of reform to reduce litigation and make for a more rational, predictable system.