4 ms·
"For example, almost every computer system can be represented by a physical equivalent." NO! Outside of the lower end of the embedded realm, computer systems
by hga 17y ago
"For example, almost every computer system can be represented by a physical equivalent."
NO!
Outside of the lower end of the embedded realm, computer systems are fantastically more complicated than anything physical we've every tried to construct, and they contain so many ideas that could be subject to patent in the current regime that you analogy fails.
- fnid2 17y agoSo this adds even more to the argument that software should retain its patentability. If it's even more complex than the old stuff we thought was so complex it deserved special protection. > NO! It's hard to believe, but it is true. Imagine simply replacing tiny transistors on a chip with big relays and physical switches that flop back and forth. Yes, it would take a long time and be a very big device, but it's a mathematically proven fact.
- hga 17y agoNo, the point here is that no one can build a software system of any complexity without infringing on many patents. Tracking that and even paying RAND fees without the need for negotiation makes this impossible, not just difficult or impractical. And it kills open source software, which doesn't have a model to pay license fees. Plus there are so many not formally documented ideas that the Patent Office regularly issues patents covering well known prior art; surely you've heard of this? E.g. the XOR patent? With any luck http://en.wikipedia.org/wiki/In_re_Bilski http://en.wikipedia.org/wiki/In_re_Bilski will drive a stake into the heart of this madness.
- fnid2 17y agoThe legal system, unfortunately, is where patents are disputed. If there is prior art, the patent will be thrown out.
- hga 17y agoUnless of course it's litigated in a jurisdiction like the Rocket Docket AKA the Eastern District of Texas, in which case the plaintiff generally wins. And it's very expensive to litigate this, to defend yourself you'd have to retain one or more very expensive expert witnesses. While I like the idea of putting money in the pockets of MIT professors, this is not the best way ^_^. Anyway, we're still talking about adding so much friction to the writing of software that it would utterly change the field and curtail most development, especially in small firms ... well, I suppose a healthy underground would develop. But there wouldn't be much for YC to do. No thanks.
- fnid2 17y agoOkay, I agree for sure that the system is messed up. But my real question is, "Is the solution deleting software patents?" Maybe the solution is holding software patents to the same rigors as mechanical patents. Why the distinction? That's what I don't understand. And still don't. Why are they treated differently? By everyone. Even outside the courts.
- hga 17y ago1. Originally, the biggest issue was "you can't patent math", i.e. where do you draw the line between discovering a rule of the universe vs. an "invention". Square roots are not an invention per se. A method for finding them...??? 2. Scale: just how many are used in any one system makes "deleting software patents" a good and probably the best outcome. Especially since there was no lack of health in the software industry before they were allowed, and there's no evidence I'm aware of that they now help it and plenty that they don't. It also seems to break down, mostly due to scale, when you insert patent trolls into the game. In the physical world there's a lot of cross licencing, where one patent, revealed to the world by the very act of patenting it, often results in an improvement, whereby the two companies cross license. Let's take a step back: the idea of patents is that for their absolute monopoly of limited duration you reveal your secret sauce instead of keeping it indefinitely hidden behind trade secret law. I fail to see how the publication of software patents helps the field in this way. Even if they weren't obscurely written, there are again just too many of them (too many ideas).
- ismarc 17y agoYou're mistaken. Every algorithm used in software can be created as a specific integrated circuit. Software is a means of controlling a general processor that performs sets of operations based on instructions it receives. No matter what level of programming you are doing (from interpreted web dev down to assembly) it gets translated and passed to the processor the same way. The problem isn't that it's too difficult to be specific enough in the patents due to the complexity. The problem is that there isn't a REQUIREMENT to be that specific in software patents. My personal opinion is that if the patents required a pseudocode implementation using a common-standard language to represent the components and operations, it would be much easier to determine patentability, would prevent generic, wide-sweeping software patents and require the investment of actual invention rather than just an investment of dreaming to be patented. Physical device patents are required to be specific enough that it can be used to create the device using only the patent and references in the patent. The same should be applied to software patents, with the translation from the pseudocode components to whatever language or platform be the same as physically constructing the components of a machine.
- hga 17y agoI think we're talking past each other in that I'm focusing on the macro level of an entire system with N (where N is large) algorithms whereas you're focusing on the concept of an individual one. My thesis is that even if I grant you a sane patent regime the macro level scales so fast and big that that changes the game and makes the concept impossible in practice.
- ismarc 17y agoWe have our common ground in that the current system is broken. My example of a single algorithm was based on the premise that for any algorithm, others can(are) incorporated into it (the algorithm can become synonymous with a cog, which would itself be patentable as well). Therefore, by my viewpoint, the macro level is identical to the micro level. The level of detail in patent applications for physical devices are mindnumbing. A key method of overcoming this problem is by using references to other patents. In this way, you could hold patents X, Y and Z and apply for a third patent A that combined the patents into a macro system without having to detail X, Y and Z all over again. My opinions on how to fix it are based on a blank slate system, all software patents should be of this level of detail, etc. rather than a just "going forward they should be". With that level of detail in a patent, resolving unintentional infringement is trivial while still providing the same output by simply referring to the patent and adjusting your implementation accordingly. This also means that prior art is easier to assess. The difficult part would be obviousness, but given the proper staff and training can be dealt with. EDIT: Reading your recent response to another commenter below, I don't want to give off the impression that I think any current software patents should be maintained. I think they should be all completely removed (or required to be resubmitted fitting new requirements that include specific details on implementation). My ideas are in the "dream world" where the rigor was applied since the first software patent was applied for.