Comment by sillysaurus | Hacker News Reader

Comment by sillysaurus | Hacker News Reader

Parent Full thread

We really need to start naming and shaming the people willing to have their names listed on frivolous patents like this, and not just their employers. If you're listed as an inventor on an unethical patent, you have violated the implicit moral code of technologists and positioned yourself in opposition to progress.

http://paulgraham.com/6631327.html

Paul's patent would seem frivolous now, but it wasn't then. By applying for a patent, did he position himself in opposition to progress? Seems doubtful.

Here's another example: http://www.google.com/patents/US5204966 This patent was filed in 1990. I was two years old. The internet hardly existed then. Was this patent frivolous? Should these inventors be punished according to your ethics?

So what's our criteria for a frivolous patent? Perhaps "(a) it was filed since 2010, and (b) it covers some well-known technique." Yet that's completely arbitrary. Punishing someone for violating an arbitrary moral code is a recipe for evil.

It's a software patent, therefore it's frivolous.

"The basic algorithmic ideas that people are now rushing to patent are so fundamental, the result threatens to be like what would happen if we allowed authors to have patents on individual words and concepts."

- God himself, http://progfree.org/Patents/knuth-to-pto.txt

That's not very good reasoning.

If all software patents are bunk, so are many electrical engineering patents.

If software is built on hardware, anything I represent in software can be 1:1 related to the digital circuit equivalent. It may be silly or infeasible to do so - but I could make an ASIC or FPGA that runs my latest piece of code. Maybe those wouldn't count because they're too close to the way we think of software. Okay - I lay out my PCB with just 4011s NAND gates. Is this sufficiently far from software to get me a patent for my ideas?

I think software should move towards a trade secret model and abolish software specific patents entirely. Unless you can demand to see the source code, you can never prove whether or not I used your algorithm.

That - or, software developers need to make one gigantic document of every single idea they have implemented and used. If everyone has 5 ideas a day that they implemented and wrote down, we'd hit some sticky situations where a new problem is found. And, much like the patent system, if we put it online where anyone could potentially see it, we can say that it was prior art for any future patent.

Then - as they don't have to be good ideas necessarily, we run through and make all permutations of every single word on the page, leading to every possible idea that can be thought of in that context. Prior art for everything. Or just get Pi out and start translating it to English.

If all software patents are bunk, so are many electrical engineering patents.

And yet few people realize that. The USPTO continues to issue software-equivalent patents as well as software patents, and the courts uphold their validity, all while claiming that "real" math is not patentable, but offering no clear theoretical framework for distinguishing the two.

ASICs could be argued to be physical implementations and therefore patentable. On the other hand, when converting a software algorithm to an ASIC design involves no additional creative effort, how can the ASIC be patented?

We used to argue that inventions should be patentable, but abstract ideas around inventions are not deserving of patents. If I was interested in your supposed ASIC design, and in a position to be awarded a patent, I would argue that by taking the algorithm and realizing it into a chip fab (or by doing the requisite homework to bring it to a chip fab, if I didn't have the requisite billion dollars to actually market the chips) I had done sufficient work to be worthy of a patent.

As a good person (who doesn't want his patent thrown out when it goes to court), I would not try to submit an incomplete design and I would make sure that my filing included all of the necessary information to put the design into future generations' hands. (But not the billion dollars, of course!) Who's to say it involves no additional creative effort? I remember Computer Organization and Architecture classes as a Computer Scientist that were meant to teach me the left hand of the equation while making me aware of the tetris game that exists on the right hand side, but not bogging us down (since we were computer scientists and not electrical engineers.)

So out of self interest, maybe I should wish to be able to patent the algorithms. I know and I was raised better than that. Our pyramid schemes start and end with our brothers (and sisters) who majored in electrical engineering! Software is meant to be free, and hardware has innate marginal cost as a necessity of physics.

Remember it can be up to the patent examiner in an abstract sense with no rhyme or reason on whether or not to grant a patent, and many of the folks I hear would rather see "boots on the ground." In other words, without the billion dollars to put the fab into works: no patent for us, so no victory.

That's not very good reasoning.

If all software patents are bunk, so are many electrical engineering patents.

I hope that the second sentence is not supposed to be a substantiation for the first one. If anything, it's an example of why the system must be fixed.

Is this sufficiently far from software to get me a patent for my ideas?

It shouldn't give you that right in the first place. Patents were meant for engineering inventions, not for mathematical discoveries. Encoding a mathematical discovery in an engineered form does not change its substantial nature.

Thing is that using common sense will reveal that this patent is restrictive in a bad way since it helps potential attackers to get into system. No matter if it's year 1980 or 2000. When Volvo invited three-point seat belt it was shared with other manufacturers because they were well aware the obvious benefit to all. Same goes eg. with blocking simple passwords.

This brings up the problem of patent terms. At least in software, the ordinary term of a patent is obviously much too long. Even granting a ten-year monopoly would be onerous in this business; twenty is absurd - the technological landscape can change completely in that time.

Devil's advocate: by allowing software patents, that might tend to force technological change. You need to invent a new landscape if the current landscape is illegal or onerous to operate in.

For the record, software patents annoy me and I don't think they should be allowed.

But the patents are so general in scope that they tend to preempt any technological change.

What benefit is there in inventing a better mousetrap if the very idea of causing inconvenience to vermin has been patented, and the only people beating a path to your door are verm^H^H^H^H patent lawyers.

20 years was long enough for Microsoft to rise and fall.

Reply on news.ycombinator.com