Recipient of First Software Patent Defends Them

Arguendo writes "Martin Goetz, who obtained the first software patent in 1968, has penned a thoughtful defense of software patents for Patently-O. Goetz argues that there is no principled difference between software and hardware patents and that truly patentable software innovations require just as much ingenuity and advancement as any other kind of patentable subject matter. The Supreme Court is of course currently considering whether to change the scope of patentable subject matter in the Bilski case, which we've discussed before." Does it weaken Goetz's argument that his description of the software lifecycle harks back to the waterfall days and bears little resemblance to current development practice in open source and/or Internet contexts?

Comments

Patents aren't the problem

[dintech]

Patents by themselves weren't a problem back in 1968 and shouldn't be problems now either. The issue is how they are used as legal clubs to beat down competition with or simply as a way to make money through litigation alone.

Re:Patents aren't the problem

[Tim C]

Software is already protected by copyright, and should not be protected by patents. If you come up with an amazing new algorithm your implementation is protected, and cannot be copied by other people.

Patents are required on physical objects because they are not covered by copyright, and so absent patent protection an engineer could simply disassemble your new vacuum cleaner (for example) and produce a clone, cheaper than yours as they don't have to cover the R&D costs.

Re:Patents aren't the problem

[wrook]

I'm going to have to disagree with this.

Software patents were a problem then and are still a problem now. Patents have always been a trade off. At one time everyone making inventions hid their work through secrecy and obfuscation. This is clearly their right and in order to protect the upfront investment in research and development it was necessary. Otherwise someone could use the information that other people had developed in order to create a product without the upfront investment. They could then undercut the initial inventor.

The result of secrecy and obfuscation was that society couldn't build on top of new ideas. Progress could stall because only a few people knew how something worked. So a limited monopoly was granted to inventors in exchange for full explanations of their inventions. It was hoped that this monopoly would allow others to build on previous inventions and accelerate progress.

This is fine when we are talking about real physical machines. We aren't giving monopolies on ideas, rather their expression in physical form. In the world of physics, only certain things work for a given problem. For example with velcro the hooks have to be at a certain angle and the loops have to be at a certain density. Nobody can patent the idea of a hook or a loop. But a specific physical arrangement of hooks and loops with a specific application is patentable.

The problem with software is that it isn't constrained by physics. For any given problem there are many expressions that can work. Software *isn't real*. If my hooks and loops don't match up I can change the laws of physics to make them match. The work in developing software is *not* trying to discover the angle with which a hook must be made, but rather the sheer volume of describing the hook.

Because people misunderstand the purpose of a patent, they believe that patents exist to protect the upfront investment of development. In this case, if it takes a year worth of work to type in the description of my hook, that hook must be worthy of protection. It is, after all, a considerable upfront cost in my act of invention. But the *specific description* of the hook (i.e., the source code) is not what they want to patent. After all, they already obfuscate it and have a copyright for it. It is amply protected. What they want to patent is the idea of a hook for a given application. *All* descriptions of hooks for that application are now forbidden.

Now we could argue that some techniques are difficult to develop. This is true. But on a computer, all techniques are mathematical algorithms. These are not, and never were patentable. If we accept the argument that a computer program is a "software machine", then the patentable part must be the description of the algorithm in the computer. But this is already obfuscated, secret and protected by copyright. They wish a higher level protection on the concept itself. This is a problem because it has never been patentable (if you wish a reason, I direct you to read the original arguments given when instating the patent system).

Not only are software patents a problem, their very nature is what is causing "bad patents" and "money through litigation" schemes. Any software patent must, by necessity, by overly broad and provide an opportunity for abuse. Not only should software patents be disallowed, but also "hardware" patents that can be fully implemented in software. To do otherwise is to fly fully in the face of the original intent of patents.

Re:Patents aren't the problem

[Mr2001]

Patents are required on physical objects because they are not covered by copyright, and so absent patent protection an engineer could simply disassemble your new vacuum cleaner (for example) and produce a clone, cheaper than yours as they don't have to cover the R&D costs.

An invention that can easily be examined and cloned doesn't need patent protection.

The term "patent" comes from the use of the word that means "apparent to everyone", as in "patently obvious". Patents further the useful arts not by enticing people to invent new things (when there's a problem that needs solving, people will solve it anyway), but by enticing inventors to share their knowledge: before patents, inventors were reluctant to share details of their inventions because they feared competition, so patents grant a temporary monopoly in exchange for disclosing those details.

This rationale for patents rarely applies to software. A program has to be executable, and if a CPU can figure out how to make it work, so can a person. Even intentionally obfuscated software can be understood with moderate effort (see any warez site).

Re:Patents aren't the problem

[vtcodger]

***Patents by themselves weren't a problem back in 1968***

A common belief. But probably wrong. Patents are, and always have been, a dubious idea. I don't have time to write a real essay. But just one example. In the early 20th Century, many of the basic patents on aircraft technology were held, naturally enough, by the Wright brothers. However, a lot of the early aircraft were built by Glenn Curtiss who was unable to get a licensing agreement from the Wrights. Curtiss built his planes anyway using (and patenting) alternate technology where he could. The ensuing lawsuits dragged on and on, draining the resources of both parties and crippling the development of aircraft in the US because no one was interested in building aircraft until they knew who they had to pay for the privilege.

http://en.wikipedia.org/wiki/The_Wright_brothers_patent_war

By the time World War I started, the US was years behind the Europeans in aircraft design. Congress eventually solved the problem by enacting compulsory licensing at a fixed royalty.

How, exactly, did this mess -- which was far from unique -- benefit anyone?

Of course, things are far worse today. The average patent is (deliberately as far as I can see) unintelligible, the claims absurd, prior art is ignored, stuff that is obvious to practicioners is patented, natural laws are patented, etc, etc, etc.

IMHO, the patent system is broken. Badly broken. We would be well advised to carefully -- very carefully -- scrap the thing. Software patents would be a terrific place to start.

Re:Patents aren't the problem

[Lord Bitman]

You can patent a new type of hammer.
If you draw a diagram of the hammer, that diagram is covered by copyright.
You can use that diagram in a patent application, in which case the diagram cannot be covered under copyright.
If you draw another diagram of the same type of hammer, that new diagram is covered by copyright.
The hammer itself, that is, the physical thing that is a hammer, can neither be patented or copyrighted.

The same logic should apply to software:
  - You can patent a new type of software algorithm.
  - If you write source code implementing that algorithm, that source code is covered by copyright
  - You can use that source code in a patent application, in which case the source code cannot be covered under copyright.
  - If you write a new implementation of the same algorithm, that new source code is covered by copyright.
  - The executable itself, that is, the black-box which tells you at most as little of the implementation details as a hammer would compared to a diagram of the hammer, can neither be patented or copyrighted.

Re:Patents aren't the problem

[swillden]

Without patents, it would be even easier for medium/large/mega cooperation to prey on small companies inventions, I write even because just having a patent today do not mean that your immune against this behavior.

That's the oft-quoted theory. In practice, little guys almost never win in patent disputes. Instead, they're bankrupted by legal bills and never see a dime of the royalties they should get. In theory, patents should help the little guy. In practice, they favor large corporations.

Re:Patents aren't the problem

[vivian]

The way I see it, patents on software are wrong for two reasons.

Firstly: The software is already protected by copyright, so there should not be "double" protection by also allowing patents.

Secondly: Patents are supposed to be a "social contract", where the inventor publishes the details of how his invention works and thus improves the "state of the art", in exchange for a reasonable period of protection that allows the inventor sufficient time to get the product to market, recoup the cost of inventing it and making a profit. Society as a whole benefits because other inventors can then use the details to incorporate the ideas and mechanisms described in other inventions (possibly subject to obtaining a license).

This works great for many inventions - but fails completely for software.
The reason it fails for software is that most software patents are so obfuscated as to be practically useless to a programmer to build upon.
How often do you actually see the source code for a software patent? I never have in ant of the software patents I have seen.

Imagine if patents were allowed on artworks - should the first guy who paints a picture of a bridge then be able to prevent anyone else from painting a picture of a bridge? Of course there are many "prior art" examples of paintings of bridges, so it would not be allowed. What if it were something a little different? Eg. someone drawing a stick figure picture of a guy in a red shirt doing a handstand on a bridge? or the more generalised case of just a humanoid figure doing a handstand on a bridge? This is the sort of wording that many software patents have - even if following artists are able to paint much better pictures of people doing handstands on bridges.

Worse yet, if you painted a picture of something entirely different, which just happened to have a picture of a guy doing a handstand on a bridge being one small element of the entire picture ( say, it's in the background or something) you would still be in violation of the patent - even though the handstand guy is just a tiny part of the whole.

now programming is to a certain extent like art - there are many ways of implementing the same "idea", in many different languages, and in ways that are better than the originally scrappy code that might have been written. Whats more, there are so many elements that go into a program, it is all but impossible to search through all the possible patents it might potentially infringe. Software patents are not making it easier for programmers to write better programs, so therefore are entirely useless for "improving the art", and thus the social contract of exchange of information in return for a limited period of protection is broken.

Re:Patents aren't the problem

[mdwh2]

Yes it's a good thing that mathematics was routinely patented over the last thousand years, otherwise we'd never had made any progress, and mathematical discoveries would never have been shared with the public.

(And even if you say it's only 20 years, how would quantum mechanics have advanced last century, if every discovery by one physicist resulted in a 20 year delay before others could make use of it? Or imagine Newton and Leibniz getting bogged down in a patent court battle for the rest of their lives over who invented Calculus first...)

Re:Patents aren't the problem

[rxan]

Many algorithms can be explained and profiled using mathematics, but they aren't a subset of mathematics. Take quicksort for example. Sorting data has nothing to do with mathematics. It just so happens that a mathematical model can predict the performance of the algorithm.

Mathematicians work on axioms and the like. There is no axioms for a lot of things in software. Saying software patents should be outlawed because it's based on logic is like saying that mechanical patents should be outlawed because it's all based on the atoms that produce the end product.

Re:Patents aren't the problem

[HungryHobo]

The problem isn't software patents, hell every now and then I see a really fantastic piece of code and think "I'd have never thought of that in a million years". Sometimes we see the software equivalent of the ball-bearing, simple, elegant but not obvious until someone comes up with it.

The problem is the hulk of intentionally obfuscated legalese in software patents which is utterly useless to an engineer trying to duplicate what's being patented.
The problem is not including the actual source code in the patent.
The problem is patenting a general idea rather than an exact way of doing it.
The problem is flow charts with descriptions of what some section of what you're patenting does because while you can engineer around an innovative break design in a car you can never engineer your way around a box in a flow chart reading "slows car down".

The problem is the bastardised combination of patents and copyright software enjoys.
Pick one or the other damnit.

truly patentable software innovations...

[fatp]

The major problem is that most software patents were not awarded to truly patentable software innovations