Broken Patent System

Politics and religion
41

In that case, your experience isn’t related to software or it’s very old. Because what you describe is not how it has worked for the past 6-7 years.

Eh, I’m speaking from present experience. So, the most recent software patent I’ve got my name on was awarded last month. The example that I used of the acquisition is one which happened about 3 years ago, if I recall correctly.

Writing software today without infringing on someone’s patent is becoming increasingly difficult in the US. There is a reason why Microsoft et al. are involved in an arms race to file as many patents as possible - it gives them ammunition with which to smack anyone who tries to threaten their turf - in addition to which they can use the patents to kill any small companies who might think of actually competing with them.

Eh, there’s actually a dual purpose there. Patents are not only offensive weapons, but defensive ones as well. They can be used to invalidate other patents. Similarly, you can look at a company like IBM who not only makes more patents than any other company in the world, but also publishes their “run-off” in their own journal, which anyone can read… basically, because disclosing ideas in a journal is as good as patenting them from a defensive perspective.

Also, I think the idea that it’s actually difficult to write software without infringing on patents is pretty overblown.

How many people here have been involved in patent litigation, because they “accidentally violated a patent”? Even in the rare case where such a thing happens, I would suspect that you can almost always design around it… Unless a patent is truly bad, in that the sense that it does not conform to the normal requirements of a patent (ie it’s too vague or general). But in that case, the problem is one of the patent office failing to do its job, not an inherent problem with patents themselves.

In which case you are protected by copyright and can sue for copyright infringement.

Not really, because it’s trivial to modify code to circumvent a copyright.

That’s the thing… folks say, “Oh, you don’t need patents, because you can just use copyrights for protection!” But copyrights don’t do the job here. And therein lies the problem that we’re ultimately running into.

Software exists in a strange limbo between physical objects, traditionally protected by patents, and written material protected by copyright. Source code is written material, but it differs from most other written material in that it is not valued for its aesthetic appeal.

That is, it I were to take a great book like “Of Mice and Men”, and change the language around in it, it would no longer violate the copyright, even if the overall story remained the same. But the book would be different (and likely worse). I would have lost some element of what made the book great.

With software, I could change the underlying code around all I want, and not necessarily lose any value at all, because the value of the product has nothing to do with the source code itself. It has to do with the functional behavior of the software after its compiled and run on a computer. If that functional behavior stays the same, then the particular sequence of characters in the source code is irrelevant. So copyright doesn’t really provide the same type of protection here.

Again - this is not the form of software patents today. Many - if not most - are written to be as broad and non-specific as possible, so as to be usable in litigation. That’s the purpose after all - a patent is only as useful as the claims you can make with it.

Of course, the more broadly applicable a patent is, the more valuable it is. But there are rules for creation of a patent.

Now just because you invent something which is so useful that everyone really wants to use it, that doesn’t mean that it sudden’t becomes “obvious” or “too broad”.

Again, the argument that something is obvious has specific meanings in the context of patents… Looking back in retrospect, and saying, “Oh, well duh… everyone knows how to implement a cursor!” doesn’t really cut it. Someone figured that out, for the first time. At the time, no one else had done it. It’s a simple, elegant solution… but it’s not “obvious” when you take it within the context of the time. Hell, I think most folks here who are arguing about it don’t actually understand what the patent actually means anyway, even today, since folks are suggesting that it’s a patent on the abstract idea of a cursor.

42

This is just a funny statement to me, as I hear similar stuff in my work in AI. “Oh, this is a finite state machine!” In reality, it’s not a finite state machine, any more than any piece of software can be represented as a FSM. To describe it as such is to fail in understanding what is going on.

When executed software is certainly math… in that all a computer can do is essentially mathematical operations, and moving symbols around in memory.

But to describe software as simply “math” is like describing engineering as simply “math” or “physics”.

Have you ever created a UML design for a large software system? Because it becomes obvious at that point that it’s more than math. When designing software on that level, it takes the same form as any other systems engineering design. You’re specifying component parts, and how they work together. You’re not writing down expressions.

And in the realm of engineering, that exact same type of thing is accepted as patentable.

Well, honestly, so have cursors at this point.

The broader statement though is that software has those qualities as well.

43

Yes. I also believe that a “one size fits all” approach is wrong. Patents should exist only to the extent that they a) make it economically feasible to sink money into R&D and b) promote disclosure of innovation. I’m sure there are plenty of products where a 17-year term of monopoly is necessary to achieve the former, and the disclosure in the patent is adequate to achieve the latter. With software patents, 17 years is massive overkill and the patents themselves are so vague that they do not constitute adequate disclosure. A sliding scale of patent term and more detailed disclosure would better serve the goals that justify the granting of patents.

So are extremely high taxes.

You’ve put the cart before the horse. Ideas are only deemed property because of government intervention. Without this anti-competitive intervention, there is nothing to “steal.” It’s also not an accepted point of law that all scientific discoveries and innovations are property–the scope of patentable subject matter does not encompass phenomena of nature, even though recently discovered, for instance.

Not at all.

44

Yes. I also believe that a “one size fits all” approach is wrong. Patents should exist only to the extent that they a) make it economically feasible to sink money into R&D and b) promote disclosure of innovation. I’m sure there are plenty of products where a 17-year term of monopoly is necessary to achieve the former, and the disclosure in the patent is adequate to achieve the latter. With software patents, 17 years is massive overkill and the patents themselves are so vague that they do not constitute adequate disclosure. A sliding scale of patent term and more detailed disclosure would better serve the goals that justify the granting of patents.

I think I could agree with that. Frankly, 17 years seems excessive to me as well.

Although it could perhaps be considered just as excessive for other types of inventions.

Again, I think part of my unease regarding this seems to be a perception that software inventions aren’t “real” inventions.

You’ve put the cart before the horse. Ideas are only deemed property because of government intervention. Without this anti-competitive intervention, there is nothing to “steal.” It’s also not an accepted point of law that all scientific discoveries and innovations are property–the scope of patentable subject matter does not encompass phenomena of nature, even though recently discovered, for instance.

Certainly, concepts are not property. However, inventions are, and have been regarded as such since the creation of our country. Generally, it seems to have worked fairly well.

But this doesn’t really have much bearing on the notion of a software patent. A systems design for software is essentially identical to a systems design for a physical device. Indeed, you could actually choose to implement software as a physical device, if you were so inclined, and had a ridiculous amount of time on your hands.

Perhaps that’s a notion worth considering… If you implemented a software system as a physical device, describing some complex set of pulleys and levers and junk, would you consider THAT device to be patentable? I can’t imagine why it wouldn’t be. Thus, it seems as though the design of the device is patentable, regardless of whether or not it describes a physical device or a virtualized one.

45

I have 5 patents. They are cool because they are mine. All other patents are stupid.

Seriously, the applications would never have been filed if someone wasn’t paying me $500 a piece for the invention disclosures. IMO, patents are totally ineffective (at their supposed purpose of encouraging innovation), and do society a lot more harm than good.

People and companies will still try to come up with new and cool things without them, and can protect themselves by being first to market. If it’s that easy to rip off the invention that there is no benefit to being first, then it obviously wasn’t much of an invention… You don’t agree? Fine, make the patent period 2 years, then. Then the inventor has plenty of time to profit without hosing the rest of the world.

I guarantee that pharmaceutical companies will not stop research just because they lose patent protection; the typical style of research may change, and some of it may shift more to academia and institutes, but I’d rather wait a few years for a new drug than have Astra Zeneca charge 100x what it should cost anyway.

If the primary function of patents was indeed to protect the inventor, I guess that would be OK, but of course, that’s not the case, and has not been the case for many years.

46

Except that this doesn’t happen. Isn’t happening. And lawsuits are filed based on these dupe patents.

Thus the suggestion that perhaps the system is broken.

47

Has anyone in this thread adjusted their arguments to encompass the just-signed America Invents Act?

Biggest deal is “first to file” determines the winner now, not “first to invent”. Also a fair bit of somewhat legally technical bits that fuck up patent review and whatnot, essentially a power boost for patent trolls. Or Apple and Microsoft.

48

Ya, I mentioned this a few times. The change is pretty detrimental to small businesses, since we cannot afford to file patents for everything.

Our approach has been in the past to record inventions in bound logs, which could be used to demonstrate the invention as prior art.

Large companies like IBM and Microsoft can afford to file far more patents.

Of course, we can still achieve some level of protection by disclosing ideas publicly, via journals and stuff.

49

The shipped product is running code, not UML diagrams. UML diagrams and the design process are not math, they’re business methods. Which are not patentable.

Timex, I guess what I’m disagreeing with is that I think it’s impossible to make useful software patents if you approach them from first principles of property rights or “how unique” they are; you have to use the clearly-intended “does it do what it’s intended” question. Which in the case of a blinking cursor, no, it doesn’t remotely give society anything to patent that.

As an analogy to the blinking cursor case to show what concerns me about a first-principles approach: there’s a fixed number of ways to write text output to the console in a C# application. Imagine Microsoft for some reason released the libraries for C# before any sample code, and in the intervening time period someone writes all possible methods for console display into an application that they then start selling for a dollar. Should they be given royalties anytime someone writes to the console?

50

Well clearly no, because something like that would likely fall into the realm of something that would be done by anybody reasonably skilled as programmer. In other words, not creative enough to be patentable. However, the patent system does seem to allow these type of patents through, and we end up in a mess of lawsuits.

Timex’s belief that this will be sorted out properly by the courts may be true in some cases, but not all cases. Patent trolls try to license out such patents by pricing the licenses so that the cost to license and make the problem go away is lower than the cost lawyer up and hit the courts. So in such a situation, the court never decides.

51

In both cases anybody reasonably skilled as programmer could do it; in both cases the solution is not immediately obvious (there’s some obscure ways to do lots of things in C#; pick a different simple operation if you’d like), and the creativity involved is minimal. What’s the difference?

52

The shipped product is running code, not UML diagrams. UML diagrams and the design process are not math, they’re business methods. Which are not patentable.

No, they’re really not.

When you make a system design, it’s more than just a bunch of processes. It’s a design of components and their interoperability.

How is this different from the design of a physical device? You’re not patenting a specific instantiation of that device. You’re patenting a design of notional components and how they fit together to perform a specific function.

Timex, I guess what I’m disagreeing with is that I think it’s impossible to make useful software patents if you approach them from first principles of property rights or “how unique” they are; you have to use the clearly-intended “does it do what it’s intended” question. Which in the case of a blinking cursor, no, it doesn’t remotely give society anything to patent that.

Could you expand on this? I’m not sure what you’re trying to say here.

As an analogy to the blinking cursor case to show what concerns me about a first-principles approach: there’s a fixed number of ways to write text output to the console in a C# application. Imagine Microsoft for some reason released the libraries for C# before any sample code, and in the intervening time period someone writes all possible methods for console display into an application that they then start selling for a dollar. Should they be given royalties anytime someone writes to the console?

Eh, not really.
There’s a function which writes text to the console. Simply using such a thing would be considered obvious, and thus non-patentable.

This isn’t really the case in the old XOR patent. Regardless though, let’s not get hung up on that particular one(although, if you like, you can read that ACM article I posted, as it specifically talks about that case).

53

So it takes what, a half an hour or less time, no court dates, and at most a hundred dollars to demonstrate this? That would be a trivial amount of effort. I’m suspecting that the actual expenses would be months or years, hundreds of thousands to millions of dollars, and massive amounts of time utterly wasted in courtrooms.

54

Generally, it’s a pain in the ass, although if you are able to simply demonstrate prior art, it can end the case pretty quickly.

Your point is valid though. Patents should not be allowed to be used as a legal cudgel by those with the means to fund extended legal fights.

55

Using a rights-based approach to patents, where “I made it, it’s creative, royalties are mine” it’s very difficult to say why you shouldn’t be able to patent “all methods of doing X in Y language”, or why you can’t patent math. The theory here is entirely goal-based - patents exist only to get people to produce something valuable for society by creating a profit motive.

56

Since when was this not really the case anyway? The standard way to demonstrate precedence for an invention is to file it. This is why you remember names like Morse and Graham Bell. Of course if inventor A chooses only to publish and inventor B wants to patent later, that’s something else.

57

Up until the recent change, it wasn’t the case. There are numerous ways to demonstrate invention that do not require filing a patent.

A common method is to record a timestamped description of the invention. Making such a record does not prevent others from developing and using that same idea, but it would prevent someone else from patenting the idea and then suing you for use. In the hearing you show the evidence that you had invented it prior to the patent being filed, and it renders the patent invalid.

58

What Timex said. I suppose it also goes hand-in-hand with the elimination of the one year grace period. Ie, you produce your invention and begin using it. You have one year to file your patent. During that time someone else can independently also create the same thing and file. Under first to invent, if you could prove you thought of it first your patent would (should) prevail. Under first to file, it doesn’t matter that you invented it first. You waited too long to file and you lose.

And the grace period has also been been eliminated. I’m probably not explaining it very well. Law and the Multiverse, of all places, covers this particular bit quite well.

59

I think that you still get protection though, even without filing it.

That is, I think being able to demonstrate prior art would still invalidate the new patent. I’m not totally sure though.

The thing that sucks for small companies is that filing for a patent is an expensive process. We simply cannot afford to just file patents for every idea we come up with. The big companies can though.

60

I’m talking practicality, obviously. When the vast majority of people want to establish invention precedence, they file an application; they do this because they want the patent for themselves, of course. But yes, you could have just published the information in a journal or timestamped it with a nonrepudiable signature if you just wanted to deny future patents.

As it happens, I prototyped the very first digital timestamping system at Bellcore, so I’m familiar with how that works :)