The Software Patent Debate Is Incorrectly Framed

An anonymous reader writes "It doesn't matter whether a true invention is implemented in hardware or software, it should still be patentable, argues Marty Goetz — the man who was granted the first software patent in 1965." The crux of the argument, according to the author: "Has there ever been a controversy about whether an invention using hardware circuits is patentable? I doubt it, even though hardware circuits are much like software in that they can be reduced to a mathematical algorithm."

Comments

Mask Work Law and Why the Heavy Process?

[eldavojohn]

I found it odd that the author didn't cover mask work rights in the United States. They only last 10 years in semiconductor mask works and are also reducible to mathematical equations despite being amazingly complex layouts of basic transistors and connections. I would argue these are closer to mathematical equations than Goetz's hardware circuit board example.

I'm not sure if software patents should be completely abolished, just reduced. Maybe five years? I mean, how long in the software industry until something is considered old news or common knowledge? For them to last 20 years just seems to be nothing but inhibiting of innovation to me -- and I'm a software developer! I'm one of the guys that should be benefiting from a longer term. But so far, it's only been a major pain in the ass. I'm sure Goetz could argue I'm just not "inventive" enough to hold software patents. I'd wager I'm just not up to the task of working with an army of lawyers.

I actually take serious issue with Goetz's explanation on the second page of the article about software:

Note that these terms are all consistent with a manufactured product: research, competitive analysis, functionality, specifications, operational environment, operating characteristics, interfaces, modules, engineering, implemented, debugged, tested, quality assurance, alpha and beta testing, documentation, installation, training, OEM, component, system, repackaged, maintenance, warrants, workmanship, guarantees, errors, defects, improved, enhanced, upgraded, and models.

Dude, you can make software development as complicated as goddamn rocket science. But at the end of the day some kid in a basement can also write software sans all that shit. I know where we work, we use Agile Methodologies, high communication, we work in very small teams and we depend on our developers not to be complete liabilities. Sure your control gates and extensive product assurance works too with just about any level of competence in your developers but I feel that's why software is so unreasonably expensive these days.

There are plenty of resources online that get you from nothing to your first "Hello, World!" program in a matter of minutes. The same is not true of hardware circuits -- especially if you want to manufacture them at all in a commercially viable way.

This analogy is rather flawed.

Re:Mask Work Law and Why the Heavy Process?

[mathew7]

What the fuck does that have to do with anything? A kid in a basement could develop a new hardware design as well. Or invent a new mousetrap, which would be patentable. But in the end, the cost and complexity of the equipment used to manufacture an invention have absolutely jack shit to do with the patentability of that invention.

Sure it does. Patents provide you with control over your idea and offer a monopoly to it's implementation. If you don't license your idea (which is legal), you can only obtain money if you sell end products. Those end products have high margins for SW, where you can have 90% profit easy (note: not the research and development, only manufacturing=copying), whereas a HW product will probably not have more than 50%. Also, for SW you can sell at least 10x the number of units compared to HW, since the replication of HW will take you a lot of time (let's face it, a basement kid will not have a production line). So this rough computation means you get at least 18x money in the same period. Also, the investment (R&D) is much lower in SW because testing is done much faster and cheaper than a HW product. Why do you think almost everything goes into SW?

As for the process comparison, many SW patents are actually too small modules. Companies are not interested in quality patents, just in numbers. They throw patents hoping that 10% are approved. For instance Apple's "slide to unlock" patent I would compare it to "low-pass filter". Surely nobody used it until iPhone, but they were not many touchscreens at that time (PDAs and tablets were limited to business users). So while Apple did show the idea, it was approved too late (because of the delay in millions of other SW patents pending) and now is common because of their marketshare in the category they created.

In summary, you have much bigger profit and faster time to delivery, but you still get the HW's exclusiveness period. This is just milking the system and NOT supporting innovation. Supporting innovation is allowing your competitor to improve your idea if you cannot or, just the fact that he can improve, to offer you an incentive to improve it first. If your idea does not catch immediately, then you are in the wrong time and deal with it: innovate more. If it does, then learn to profit in the shortest time but still innovate while in peak, don't turn into a potato-couch for the rest of your life.

Yeah, exactly.

[lorinc]

That's why all patents are just plain bullshit and should be nullified...

Times change

[Aladrin]

At one time, innovation and invention were hard to come by, and needed to be encouraged greatly. They weren't viewed as profitable and useful enough otherwise.

Today, innovation and invention is far cheaper, faster, and more profitable. The low end of the range can have massive changes in weeks. The high end still take years, money, and knowledge.

It's still worth protecting the high end. But 'protecting' the low end just stifles innovation. The system was supposed to correct for this by refusing 'obvious' patents. Now, either we need to redefine 'obvious', or there needs to be more unpatentable things. If you can bring something to market in a single month, there's no way that should be patentable. (1-Click, I'm looking at you!)

But if it takes a year, then it needs protection for the creator to recoup their money.

I don't know if it would work, but it would be an interesting twist to see patents expire when their cost has been recouped. So if you try to license patents out instead of creating the item, the best you could do would be to break even. (This would not include operating costs in the mean time, only profit, so you could probably make a decent living by stretching it out, but you couldn't possibly grow your company with it.)

Reality check?

[zmooc]

I totally agree with what he says. However, he misses the point that's called reality.

Problems that could easily be solved in hardware, would require a description of this hardware. A transistor here, a flywheel there and a plate of metal here in order to achieve X. It's that description that is then patented, not its functional result. Alternative implementations would then not be covered by the patent so anybody that finds a way to achieve X with a piece of plastic instead of metal should not be bothered by it.

That's not what we see in software patent land today; instead of the technical design, the functional result is patented. There are a million ways to implement slide-to-unlock and somehow Apple has a patent on all of them. That's odd, since most of these million ways have nothing to do with the original research Apple has done in order to obtain the patent.

If the same standards for granting a hardware patent would be applied to software patents, I could probably live with them (since in that case there would hardly be any software patents). Now I cannot, especially not when selling software in the USA.

Also, mathematical formulas cannot be patented. Therefore the comparison with mathematics is moot. What can be patented, is the practical use of that specific formula. Also note, the patent Marty Goetz was granted was not a software patent in the sense that we think of it today; his patent was valid ONLY when used on a machine using two tape reels. A pure software implementation would NOT have been patentable. Therefore using this patent as an example of a software patent is misleading; it is not a software patent, it is a regular patent that has part of the solution implemented in software.

Absolutely right

[TheRaven64]

The problem with software patents is not that they are software, it's that the vast majority of them are obvious and they tend to be overly broad. Anyone encountering vaguely the same problem would produce the same solution, or one sufficiently similar that it would be covered by the patent.

The other problem is that patents really fail at their primary purpose: encouraging disclosure. No one looks for algorithms by doing a patent search. People look for published papers, existing libraries, or invent something themselves. Any one of these can end up violating a patent, but without gaining any of the advantages of the patent system.

People campaigning against software patents often get bogged down by assuming that this is something that is limited to software patents. It isn't. Talk to people in almost any industry, and you'll hear the same thing. They either have big cross-licensing agreements that let them ignore all patents and just keep out new people, or they find themselves constantly having to pay royalties for things that they invented independently.

I would have no problem with software patents if they were limited to the scope that the patent office claims: novel, non-obvious, useful. If this happened, I doubt even 1% of current patents would stand up.

Re:New Idea

[foniksonik]

Should you be able to patent the combining of two old ideas? Many software patents are not new but are simply an old idea implemented in a new way but using existing techniques which are obvious to anyone skilled in that area. Typically it is just an extension into a different field of use.

Eg. A database to keep track of bird sightings. There is prior art of how to record such things in a log book (spreadsheet/table format) and anyone skilled in the creation of digital databases can duplicate this format and write CRUD operations.

This would probably receive a patent in the current atmosphere. I've seen many patents just like it but for other topics.

IMHO the above example and all "software" or "business method" patents like it should be nullified. They are not new or innovative in any way. They are just applications of existing technology.