Hacker News Reader

Well, actually, it does help a lot. In particular, they said ""claims, which merely require generic computer implementation, fail to transform that abstract idea into a patent-eligible invention""

Historically, federal circuit precedent said exactly the opposite - that when a generic computer was programmed for a specific purpose, it was now a special purpose machine and patentable.

"For computer implemented processes, the “machine” is often disclosed as a general purpose computer. In these cases, the general purpose computer may be sufficiently “particular” when programmed to perform the process steps. Such programming creates a new machine because a general purpose computer, in effect, becomes a special purpose computer once it is programmed to perform particular functions pursuant to instructions from program software."

This decision should, if the federal circuit actually follows it, knock out a large set of software patents.

>Well, actually, it does help a lot. In particular, they said ""claims, which merely require generic computer implementation, fail to transform that abstract idea into a patent-eligible invention""

Historically, federal circuit precedent said exactly the opposite - that when a generic computer was programmed for a specific purpose, it was now a special purpose machine and patentable.

1) Your argument does not refute the GP's statement. A "specific purpose" is not necessarily an "abstract idea", depending on how the claims are structured. A claim saying "presenting ads in such a way that it is unobtrusive on a computer" is an abstract idea[1]. A claim that specifies how a computer can, say, examine visual or DOM elements on a page to determine where to place ads such as not to be obtrusive, is much more concrete.

2) A generic computer running a program is a special purpose machine: it's now doing things it could not do before it was programmed to do so, and as such, it is performing potentially novel functionality. Otherwise, in my mind, you might as well argue "a kitchen knife could not be a murder weapon because it's a kitchen knife."

> This decision should, if the federal circuit actually follows it, knock out a large set of software patents.

I've looked only at a few particular areas, but all the "old idea, only on a computer / on the Internet" patents I've seen so far (like [1]) the one should have expired by now. However this is purely anecdotal, of course.

1. Such a claim (I only paraphrased it a bit) exists, and was referenced in one of Lemley's papers. On mobile so can't find it right away.

"2)a generic computer running a program is a special purpose machine: it's now doing things it could not do before it was programmed to do so"

This is of course false. It always could do them, it just wasn't told to. To view it otherwise is nuts.

Otherwise, I could argue that when you think about things, you are activating new pathways in your brain, forming neural connections that didn't exist, and otherwise changing and programming the state of your brain to know and do things it could not before.

> It always could do them, it just wasn't told to.

"A board, spring and wires could always trap a mouse, they just weren't configured to do so."

How is programming a computer different from assembling existing components in a new arrangement to make a new machine?

> otherwise changing and programming the state of your brain to know and do things it could not before.

Not a neuroscientist, and this risks derailing the thread, but isn't that exactly how brains and learning work? This has also been observed at the macro level, e.g. that study where London cabbies' brains were found to have enlarged portions that dealt with spatial processing.

A board, spring, and wires is a bit different from a programmable computer, because the computer was designed to be programmable. Putting a program in changes the computer's state, true, but it's changing it to a state that it was designed to be able to achieve.

That difference ruins your mousetrap analogy.

""A board, spring and wires could always trap a mouse, they just weren't configured to do so."

How is programming a computer different from assembling existing components in a new arrangement to make a new machine?"

You are not "assembling existing components in a new arrangement with a computer". Also, if I gave you a board, a spring, and wires, you could not trap a mouse with them.

You actually have to physically change them to accomplish anything. The spring won't stick to the board by itself, etc.

You are also taking things that were previously limited in some way (a board is not elastic, etc), and assembling them into a new thing that is not so limited.

When you program a computer, you are not changing the computer in any way, shape, or form.

It was never limited. By programming it, you are doing exactly what it was built to do: Be programmable.

You are simply picking a subset of the things it could already do, and saying "do this subset". The computer, no matter how hard you program it, will never trap a mouse. It is the building of the computer that was the hard part. You, on the other hand, are always operating entirely within the existing limits of the computer.

"Not a neuroscientist, and this risks derailing the thread, but isn't that exactly how brains and learning work? This has also been observed at the macro level, e.g. that study where London cabbies' brains were found to have enlarged portions that dealt with spatial processing. "

Yes, this is exactly how they work. That's my point. By your argument, you can replace "computer implemented" with brain implemented, and end up with the same patentable result. Given that, i'm arguing there is nothing special, at all, about computers, or implementing things using them, any more than thinking about things or learning to play guitar. Enlarging the center for spatial processing in cabbies is something the brain was meant to do. The cabbies did not create it out of whole cloth. They did nothing but exercise their normal skill in spatial reasoning, and their brain did all the actual work.

Programming a brain does not make them do new things they could not do. It simply tells them what task to perform.

IE There are a set of tasks they can both perform. That set of tasks is fixed at creation, both of brains, and computers. You cannot make lasers shoot of your eyes no matter how hard you try.

Much like a brain implemented invention, a computer one is just picking a certain task the computer already could do.

> You are also taking things that were previously limited in some way (a board is not elastic, etc), and assembling them into a new thing that is not so limited.

I think you have this part of it backwards. The fact that you can assemble a board and some metal into a mouse trap shows that the component parts weren't limited.

Computers are the opposite. They're extremely limited -- they can't physically do anything, they can only perform calculations. Software is capable of nothing but specifying what calculations to perform.

Here's a great example of the absurdity that occurs when you allow the narrow thing computers do to be the subject of a patent. Suppose that the text of a particular software patent consists of source code implementing the invention. If software can be patented then that patent would infringe itself. All you would need to constitute an embodiment of the "invention" would be a prior art general purpose computer and the patent's text. Imagine an infringement suit against the Patent Office (or Google Patents) for storing the patent's text on a computer capable of executing it.

This is so completely absurd I'm almost speechless. You have so completely missed how the world works I can't even fathom it. The more arguments I read, the more delusional people sound.

People are not getting patents for some sequence of calculations, they're getting patents for producing a useful, new effect on the world. Yes all computers are universal in what they can compute, but I see no reason why coming up with a new, greatly improved and economically useful way of using such a restricted machine should not be patentable.

Furthermore, people seem to think that just because a computer program can be described as a sequence of mathematical steps that it is therefore a mathematical algorithm and shouldn't be patentable. Algorithms have no effect on the world, they do not improve anything in a tangible way, but software absolutely does (and it's why you can read this now!) and it is the processes that lead to producing these useful effects that is patentable. I can't patent an algorithm without specifying why it is useful to the population at large.

Finally, people seem to have got the bizarre notion that there are patents claiming computers which are capable of performing a certain task, which is just absurd and untrue. The claims are usually directed to a system which actually does perform the steps in the process, not one which is capable of it (therefore allowing them to exclude competitors who make their own implementation of some software process that has been patented and attempt to make money from it).

It's one thing to patent a specific arrangement of a spring, board, and wire to build a mouse trap it's another to patent ALL arrangements of a springs, boards, and or wire to build any device that can possibly trap a mouse. The first is an invention the second is the abstract idea of a mouse trap.

The problem is software patents have long used the second approach. Rather than patenting a process they attempt to patent the result. And this is where the fundamental difference between software and physical reality shows up. If your building something then 'optimal' solutions are great, but if your building software anything that works is probably good enough.

" People are not getting patents for some sequence of calculations, they're getting patents for producing a useful, new effect on the world. Yes all computers are universal in what they can compute, but I see no reason why coming up with a new, greatly improved and economically useful way of using such a restricted machine should not be patentable."

I think you missed the point, actually.

They have argued, and precedent until yesterday was: It is patentable because it is implemented on a computer, because it "transforms the computer" into a "specific device". IE that the computer, programmed, makes something unpatentable, patentable.

"Finally, people seem to have got the bizarre notion that there are patents claiming computers which are capable of performing a certain task, which is just absurd and untrue. "

As a patent attorney, having fought software patents for many many years now, i'm just going to say this is completely and totally false.

" People are not getting patents for some sequence of calculations, they're getting patents for producing a useful, new effect on the world. Yes all computers are universal in what they can compute, but I see no reason why coming up with a new, greatly improved and economically useful way of using such a restricted machine should not be patentable."

I think you missed the point, actually.

They have argued, and precedent until yesterday was: It is patentable because it is implemented on a computer, because it "transforms the computer" into a "specific device". IE that the computer, programmed, makes something unpatentable, patentable.

"Finally, people seem to have got the bizarre notion that there are patents claiming computers which are capable of performing a certain task, which is just absurd and untrue. "

As a patent attorney, having fought software patents for many many years now, i'm just going to say this is completely and totally false. People patent both computers capable of things, and computer implemented inventions. To think otherwise just means you've not read enough patents :)

I agree with your first part, but the last part is incorrect.

>If software can be patented then that patent would infringe itself.

To infringe a patent, you must make, use or sell a product incorporating the invention. A patent by itself does none of those and hence cannot infringe anything.

> To infringe a patent, you must make, use or sell a product incorporating the invention. A patent by itself does none of those and hence cannot infringe anything.

How are you intending to distinguish program source code in an interpreted language incorporated into the body of a patent specification from a software product consisting entirely of that same code?

A patent is essentially a document with just two purposes: 1) instructing a practitioner how to reconstruct the claimed invention, and 2) defining the legal scope of ownership for the claimed invention. By itself, it does nothing else, and thus does not constitute making, using, or selling.

Now if somebody other than the patent-holder takes the source, compiles it and uses and / or distributes it without license, they are infringing. Note that by taking and compiling the source, they are essentially just fulfilling the first intended purpose of the patent: instruction. Where they run afoul is the second purpose.

It seems like the distinction you're trying to draw is between source code and a compiled program, but that doesn't apply to interpreted languages. There is no compiled program, the source code is the program. And you can effectively convert any compiled language into an interpreted language just by associating its file extension with a two line shell script that invokes the compiler and executes the output. Trying to differentiate one from the other on that basis is frivolous.

So I still don't see how you're trying to refute the conclusion that a software patent containing code covered by its claims would infringe itself. You couldn't reproduce or distribute the patent without reproducing or distributing the program.

Obviously a patent sitting on a desk doesn't self-execute, but neither does a program on a disk sitting on the same desk. I see no relevant difference between one and the other any more than there is between a USB stick and a CD.

Compiled versus interpreted has no bearing on this. The patent and the source code in it exists only to instruct, and by itself the invention claimed is neither made, sold or used and has no commercial effect. Unless an unlicensed third party does anything with it, it is not being made, used or sold, and so cannot infringe. Software that is lying unused and unexecuted on some computer readable media is still infringing if it was created without license because the act of copying code on to or distributing that media counts as "making" or "selling". I I cannot imagine under what interpretation a patent itself could count as an "unlicensed third party".

Are you proposing that if my patent containing source code is published by Google on Google Patents, I can sue them for infringing it because they are an "unlicensed third party"?

I am not able rightly to apprehend the kind of confusion of ideas that could provoke such a question.

Let me take a stab, though. Keep in mind that the text of the code does absolutely nothing until it is interpreted or compiled. Now, say your code implements a way to detect porn in an image, which is what your patent claims. By publishing the code in the patent, is Google Patents, or heck even the USPTO, making, using or selling "porn detection in images"?

A simple answer to this question is the answer to the question, "Can I submit an image to USPTO/Google Patents and get a determination of whether the image contains porn?"

> Computers are the opposite. They're extremely limited -- they can't physically do anything, they can only perform calculations. Software is capable of nothing but specifying what calculations to perform.

I think this is the heart of your argument here. (Correct me if I'm wrong.)

The problem is that in terms of calculation, computers are not limited. In fact, that was proven not once, but twice, in 1939. Any computation that can be performed can be done with any general-purpose computer. The Church-Turing Thesis proves that that computer need not even be specialized: it only requires the ability to manipulate symbols deterministically in some general manner.

What these patents cover is, in fact, only those computations with the aid of a general-purpose computer, and the only reason those patents have been allowed is by this fallacious argument that a general-purpose computer somehow becomes a special-purpose computer when it is programmed. This is not how computers work. A general-purpose computer does not become special-purpose merely by assigning some numbers to a pool of memory any more than an FPGA becomes an ASIC by uploading a ROM image. It is still a general-purpose computer, and its nature does not change.

If, perhaps, by programming a computer, the program magically became a new, immutable physical wiring, your argument might have merit. But we do not live in a world of magic.

> The problem is that in terms of calculation, computers are not limited.

If you exclude all the things computers can't do then they can do everything else. That's just begging the question. There is no way for a software program to allow a general purpose computer to trap a mouse, or generate electricity, or convert raw materials into manufactured goods etc.

Depending on how you look at it computers may perform physical operations based on instructions; which may amount to calculations. The physical computer vs. the conceptual computer.

> When you program a computer, you are not changing the computer in any way, shape, or form. It was never limited.

But that is untrue, on two levels! Firstly, at the physical level: by programming the computer you are physically changing it. You are changing the electronic charges on a multitude of molecules arranged at very specific locations in a very specific manner. This, in turn, when activated by other specifically applied charges, controls changes to yet other charges on other molecules at other very specific locations as these charges are evaluated by the laws of physics and the arrangements of various conducting materials connecting these molecules.

Secondly, at the functional level: these patterns of charges can be evaluated, by humans or other machines, to represent useful information, and hence the exact same circuit with different patterns of charges can be used as a word processor, a gaming machine, a media player, a database...

Consider the converse: without any such charges lighting up any of its circuits, the computer can do literally nothing. It is wholly limited.

I completely agree that the computer's potential utility is never limited -- except maybe by our imaginations. However, that could be said for a hunk of metal or a slab of wood too.

> By programming it, you are doing exactly what it was built to do: Be programmable.

By fashioning steel into various shapes, I'm doing exactly what steel was created to do: be malleable when very hot but very rigid at room temperatures. Yet every new shape given to steel, from rods to blades to ball bearings to I-beams to frying pans, has enabled a new use.

By combining various atoms of elements in different configurations, I'm doing exactly what they were meant to do: bond with each other. Yet every new combination could result in a brand new compound.

What something could do has no bearing on what is inventive. Inanimate things do precisely nothing until we invent a use for them. All technological development is the creative use of things doing what they are meant to do, arranged to achieve things that we want them to do.

> Enlarging the center for spatial processing in cabbies is something the brain was meant to do. ... They did nothing but exercise their normal skill in spatial reasoning, and their brain did all the actual work.

True, but their brains have become physically different from ours, optimized for spatial reasoning. You could in fact say that, with the right programming, they became special-purpose brains.

Hold on. You can't just say: Computers are implemented physically, so grounding a patent in a computer-based implementation makes it a physically-grounded patent.

If you want something physically-grounded you'd have to say: I want to implement this idea on that specific computer with these specific physical manifestations. And even if you got that patent, it would be effectively worthless. Changing what we view as trivial details of the implementation (e.g. swapping a Pentium for a Pentium Pro) would totally change the physical manifestations at the level of electrons and molecules. In other words, that would completely avoid any claims in the patent that are described at the physical level.

And the minute you retreat to considering claims at the functional level (i.e. what our collections of electrons and molecules mean) you're no longer talking about doing something physical to achieve a useful result: you're talking about combining abstract ideas with a wide range of possible physical implementations to achieve a useful result, which is something entirely different.

This is true for most physical inventions. Machines, for instance, may be claimed as a particular configuration of physical components, but the invention is not restricted to any precise composition materials used in those components unless the claims require it. Indeed, claims often don't even mention the materials being used.

For instance, you can use any comparable material or fluid to provide the necessary mechanical support or hydraulic force. Replacing, say, steel (or oil) with any other alloy (or incompressible fluid) with comparable properties will not allow you to sidestep the essence of the invention. This is analogous to switching one processor to another: it makes no difference because they are equivalent. As such, pretty much all patents (except maybe those on compounds) are abstract to some degree: they all claim a useful application with a wide range of possible physical implementations.

This is covered by the Doctrine of Equivalents, which specifically exists to prevent non-infringement through trivial workarounds. The counterbalance to this is that in examination, claims are given their broadest reasonable interpretation to find as much prior art as possible. For instance, examiners are free to use equivalent elements in the prior art to read on those in the claims.

You're trying to simultaneously claim that the machine is what makes it patentable as a result of its physical operation and that the machine is an abstraction completely divorced from its physical operation. Those two things don't go together.

No, let me do a diff on your statement to highlight my meaning (diff'd words in italics) I'm claiming that:

1) ... the machine is what makes it patentable as a result of its physical operation and

2) The patent on that the machine is an abstraction largely (to the extent it concerns irrelevant specifics) divorced from its physical operation.

You see the difference now? Those two arguments reconcile because they address different things. To see what I mean, do a search on google patents for, say, "hydraulic modulator", and count how many have claims that specify the precise materials they use. Without such details, these claims are about as abstract as software claims.

The difference is that in the software case the patent claims are completely divorced from the computer's physical operation. You could be using a gravity-based computer built out of planets and stars and it would infringe all the same. Because the computer isn't required to do something, it is only required to calculate something.

Let's try this another way. You understand that it is possible to build a computer out of anything. Let's build a computer out of nothing. We want to perform a calculation without building anything whatsoever, so all we are going do is observe nature. There are any number of ways to represent a calculation. If there are three pigeons and five more arrive then you have only to observe the resulting number of pigeons to learn the sum of three and five, etc. Nature is going to do lots of calculations we don't care about the answers to, so we ignore those and only pay attention to the ones necessary according to the algorithm to be executed.

By doing this you can execute any algorithm. Doing strictly nothing but observing natural phenomena. There is no machine. Because the result you are trying to obtain is not a thing, it is a representation of information. And information can be encoded into anything and the representation is defined by the context. By changing the context -- an entirely intellectual undertaking -- you determine what information is represented by the physical state. What the machine (or the birds or the stars) do is comprehensively irrelevant, as long as things are happening you can use them to represent the execution of computer software.

> The difference is that in the software case the patent claims are completely divorced from the computer's physical operation.

I'd like you to show me a granted software patent that cannot be implemented on any computer. If you can find one, I'll show you one that should be chucked in with perpetual motion machines.

> You understand that it is possible to build a computer out of anything. Let's build a computer out of nothing. We want to perform a calculation without building anything whatsoever, so all we are going do is observe nature.

It's amazing that each of these three consecutive statements are inter-contradictory. ("build a computer out of anything" != "build a computer out of nothing"; "nothing" != "observing nature"; "build a computer out of anything" != "without building anything whatsoever".) I have no idea how to proceed.

> It's amazing that each of these three consecutive statements are inter-contradictory. ("build a computer out of anything" != "build a computer out of nothing"; "nothing" != "observing nature"; "build a computer out of anything" != "without building anything whatsoever".) I have no idea how to proceed.

Oh sorry, let me clarify. I'm taking it as a given that you accept a computer can be built out of anything, i.e. that you can have electrical computers, mechanical computers, biological computers, etc. and they're all computationally equivalent and can execute all the same algorithms. The whole Church-Turing thing. My point was that you don't even have to build anything. So you can completely ignore the "build a computer out of anything" statement if you think it's contradicting anything.

And observing is not building. It seems exceedingly obvious that you can't patent squirrels gathering nuts, or leaves growing on trees, or raindrops falling from the sky, etc. But you can map desired calculations onto the occurrence of things in nature in much the same way as you map them onto electrical signals in a PC, wait until they enter the state required by a given algorithm, observe the result, and thereby execute any algorithm you can execute on a computer. The machine is a completely abstract concept. Trying to patent an algorithm because you can execute it on a computer is exactly the same thing as trying to patent a formula because you can compute it with a calculator. The calculator doesn't become a new type of machine based on which buttons you press.

> > I'd like you to show me a granted software patent that cannot be implemented on any computer. If you can find one, I'll show you one that should be chucked in with perpetual motion machines.

Obviously a patent on a solution to the halting problem is fraudulent, but you have the issue reversed. It isn't that you can't use a machine to execute an algorithm, it's that you can execute an algorithm without a machine. The machine isn't a necessary component, it's just a convenient and efficient way to do it.

Without arguing this back and forth much more (I simply don't have the time to sit on hacker news all day, sadly :P):

The problem with this type of argument is that it essentially says everything is an unlimited physical process that could be patented through competent drafting, because everything always affects change at some level. Heck, even the process of observation affects things at a physical level. Allowing things to be patented, simply because they can be described as a physical process, without anything more, is ridiculous.

This was not was anyone had in mind when they created the patent system, nor was it even allowed until recently.

That's ad absurdum. The current statute is "anything under the sun that is made by man". That is pretty broad, but even that drastically reduces what can be patented, because you'll notice the number of things that could be described by a physical process is infinite, but an infinitesimal fraction of those is made by man. An even smaller fraction of that is useful and novel. I guess it's just my opinion, but the current theory captures the intent of the patent system just fine, though the implementation can use a little tuning.

> You are changing the electronic charges on a multitude of molecules arranged at very specific locations in a very specific manner.

The problem is that the physical aspects of the computer are not what infringes the patent. They exist in the prior art. A mechanical or biological computer would infringe the same as an electrical computer because the physical aspects are irrelevant to the software. You're essentially try to claim that authors should be able to patent literary works because a pen or a printing press is a physical machine that moves all manner of ink molecules around into a particular configuration.

> A generic computer running a program is a special purpose machine: it's now doing things it could not do before it was programmed to do so, and as such, it is performing potentially novel functionality. Otherwise, in my mind, you might as well argue "a kitchen knife could not be a murder weapon because it's a kitchen knife."

I don't think that analogy goes where you want it to go. Suppose that a kitchen knife exists in the prior art and you want to get a patent on "homicide weapon with hand grip and pointed edge." That which infringes if later anticipates if earlier. So if the kitchen knife would infringe that patent then it's also novelty destroying.

I meant the kitchen knife analogy as more of a general illustration that actual use of something is more important than straightforward interpretation of that object's purported purpose, rather than as a case of patentability :-) I fully agree that the use of kitchen knife as a murder weapon would be non-novel, given that sharp objects have been used to cause death since prehistoric times, as well as obvious, given that sharp objects are known to cause bodily harm, which is known to cause death.

> One of the horrid things about law is how it can be dissembled by those whose job is to practice it. This drives people crazy. They know something is real. Yet, especially if they are caught in its hooks, they watch how law can be applied such that results are absurdly divorced from the reality they know.

I think this is maybe overly harsh. The plaintiffs in CLS v. Alice weren't trying to obtain a result "divorced" from reality. Their argument was simply that 101 is written broadly and should be interpreted broadly (i.e. judicially created exceptions should be interpreted narrowly). There's no truth--just a statute of Congress that is amenable to differing interpretations. Some of those interpretations are favorable to some people, and others to other people. Who knows what Congress really intended. Indeed, there's 535 differing sets of intentions...

>> the Court not only did not address this sort of argument but it also carefully emphasized the idea that it was to "tread carefully"

I'm very much against the software patent industry, but in general, judicial restraint is usually more good than bad. Sweeping redefinitions of law by the Supreme Court that go beyond the primary issue at hand in the case more often than not have very problematic unintended consequences. While it would be great to kill the software patent industry overnight, that kind of change is better left to the legislature, while courts deal with challenges to specific claims of patent trolls. It's a slow process, but it's much tougher to roll back a Supreme Court decision with unintended consequences than legislation.

>"...cut through such dissimulation in the area of patents and to look at the reality by asking, in effect, "is this really inventive or is it simply a product of the draftsman's craft parading as being somehow inventive."

If only we could create an office that could somehow, you know, examine the patent before granting it and do just this. These so called "examiners" could then prevent it from going all the way through a court to a judge before we decide if it was a good idea to grant the patent in the first place.