News for hackers. Stuff that matters.
Re:Obama Calls for Patent Reform to Topple Trolls by rayiner on Saturday February 16, 2013 @02:28AM (#5230767)
You make excellent points, but I have a bone to pick with your last one: "Copyright works really well for software. The implementation is the hard part, and that is the part that copyright covers."

Whether design or implementation is the hard part depends entirely what kind of software you're building. Think of something like Google's self-driving car. I can't imagine that the hard part of that is just sitting down and writing the code.

Re:Obama Calls for Patent Reform to Topple Trolls by AnthonyMouse on Saturday February 16, 2013 @10:59AM (#5231820)
>Whether design or implementation is the hard part depends entirely what kind of software you're building. Think of something like Google's self-driving car. I can't imagine that the hard part of that is just sitting down and writing the code.

That's kind of what I'm talking about. You can't get a patent claim on "a self-driving car" -- and if you do the patent office has failed miserably. Software patents don't cover the overall product, they cover the building blocks you use to make one. And each of the individual building blocks is not the hard part, the hard part is the overall implementation -- assembling them all into a cohesive whole that accomplishes the major goal you set out to achieve.

You want someone who starts with the Google self-driving car and sets out to copy it to have a hard time of it legally, so that they decide instead to do their own independent implementation and have their own costs (and maybe advance the field themselves) so that they can't just undercut Google having done no R&D, but at the same time still make it so that they can do an independent implementation without infringing. Those are the two important criteria.

Copyright pretty well does the first one. If you can't just copy Google's code out of their car and put it in yours, you have to write some tens of millions or more lines of code yourself. And then you have to test it and stamp out the bugs, which is by far the most expensive part of producing software in practice. (I expect this goes double when the thing you're making is a car.) Patents don't really add very much on top of that -- if someone is already doing an independent implementation, having to work around a patent is a small incremental cost, especially if the issued patent is not so broad that it covers the problem itself rather than a specific solution.

Which brings us back to the trouble: In order for the patent system to do anything for software R&D, you can't just have one patent, you need thousands. Which thoroughly violates the second criteria -- now no one can do an independent implementation without accidentally stepping on some fraction of the outlandish number of issued software patents. To clear yourself you have to clear tens or hundreds of thousands of issued patents, but to sue you the plaintiffs need only find a single patent for which you failed to do so. And to shake you down they don't even need to do that, as long as the cost of proving they're full of crap in court is more than the amount they're demanding for a license. These costs vastly exceed any possible incremental benefit over copyright of having software patents at all.