That's the thing: It doesn't really matter. Take any given software patent and throw a research team at it, chances are you'll find invalidating prior art. Take any given software patent and give it to a CS major, chances are you'll have a work-around inside of a few days.
But that's not the problem. The problem is that you never know which patents will be asserted against you, and by that point the horse is out of the barn. You've now got to spend millions of dollars on lawyers for years to invalidate their improvidently granted patents. And the work-around doesn't do much good when they're asking for damages for past infringement on the millions of devices you've already shipped or suddenly having your trucks stopped at the border (and interrupting your business and straining your relationships with retailers) during the period before you can retool your supply chain to incorporate the work-around. Especially when the second you release a new product with a new set of work-arounds, they come at you with a new set of patents.
I think there is a middle ground - patents, if done "properly" is useful.
Perhaps, firstly, you have to have a more vigorous application process that is peer reviewed (where peers are basically your competitors!).
Secondly, you have to declare the valuation of the patent. This is then taxed by the government, similarly to property tax. If you decide to value your patent at $0, then you pay $0 tax.
The valuation of the patent determines the damage payment.
This combination means that you can't just keep a patent indefinitely, waiting for someone to infringe and then extract money from them. Peer review means that you can't patent something that will stop your competitors from continusing business. Real innovation will get thru because your peers will have no case of prior art, but "bad" patents won't get thru as your competitors will search hard for prior art to stop you (aka, incentives are aligned amongst the parties).
Secondly, you have to declare the valuation of the patent. This is then taxed by the government, similarly to property tax. If you decide to value your patent at $0, then you pay $0 tax.
I'd rather just see software patents thrown out entirely but I think this is the most reasonable suggestion I've seen for fixing the current system. If that tax were rolled back into funding real scrutiny of incoming patents the whole system might not be the cancer it's become now.
I love this idea (there would have to be careful thinking around patent resell/value restating to avoid sneaky shell company tricks).
It means you don't have the problems of special-casing software (no software patents) or vice versa special-casing like big pharma stuff (only patents on drugs - which most people seem to think is a needed case for patents) - big pharma would be much happier to claim up front that drug patent is worth millions on pay tax on that.
It also means that you'd get lots of press around big stated-value (and so big potential claim) patents, so people in the same industry would know which patents to avoid infringing, hopefully much reducing risk of accidental infringement.
>Secondly, you have to declare the valuation of the patent. This is then taxed by the government, similarly to property tax. If you decide to value your patent at $0, then you pay $0 tax.
>The valuation of the patent determines the damage payment.
This sounds more clever than it actually is. Think about it:
First of all, this makes the patent system totally worthless to the little guy. If Joe Inventor has the Billion Dollar Idea, he can't value it at a billion dollars even if it is worth that much because the first year's taxes would bankrupt him before he can recoup the money. But if he values it at substantially less than that, the big guys come along and infringe and just pay damages that are way less than the value of infringing.
But you can't allow the patent holder to periodically revalue the patent and change the amount of tax being paid on it because that breaks the system, since the big guys could just value all their patents at $0 until they see someone infringing, then revalue them at a trillion dollars the day before they file the lawsuit.
And the alternative of not allowing revaluation ignores what a crap shoot patents are. Sometimes you don't know that something is going to take off until it does. If you have five or ten genuinely good ideas but can only expect one or two to gain market acceptance, and you don't know which ones at the time of filing, you can't value them all at a billion dollars or you'll be crushed by taxes, but if you value the wrong ones highly then the patent system isn't providing you with the desired incentive because when the infringers come around you won't be able to collect sufficient damages to deter them.
More than that, think about what you would be doing to patent litigation: Valuing an individual software patent at a billion dollars is insane for anyone because of the taxes that would be due on it and because of the potential of someone finding a work-around that makes that valuation worthless, so instead the strategy (much like today) will be to collect many thousands of patents, valued in the low millions individually, but instead of asserting five or ten in litigation, plaintiffs will have to assert a much larger number of them in order to arrive at the desired amount of damages. These trials are a giant mass of spaghetti even with five or ten patents, can you imagine a hundred? So it does nothing to solve the problem of upstanding defendants being coerced into unjust settlements for fear of being crushed by litigation costs.
So the taxes thing is not really going to do it I don't think. (And that's before the Tea Party gets a billion dollar donation from Intellectual Ventures to go on TV and scream about taxes on innovators, and before the tax provision gets eaten by all the same loopholes that allow Microsoft to not pay income tax.)
Which leaves the suggestion that we need a more rigorous vetting of patents -- and I'm all for that, but easier said than done. Peer review probably wouldn't hurt, but you've got a serious free rider problem. If IBM files for a spurious patent then in theory Google and Microsoft each have an incentive to hire lawyers and try to oppose their application, but they also have an incentive to save money by letting their other competitors do it. And because patents are bought and sold like commodities, you can't just concentrate on fighting your own competitors' attempts to get bad patents, because anyone who is awarded a low quality patent is a potential supplier of them to your competitors, so anyone who decides they want to subsidize their competitors by being the first to start policing bad patents in the industry will have to do so against the world rather than only their own actual competitors. You end up with a situation where nobody wants to be the first mover to start opposing competitors' patents because there will inevitably be retaliation, so nobody does, and all the big players are happier to continue the existing system of everyone big having an arsenal of low quality patents that in most cases just get cross-licensed to one another.
It seems to me it would be better to just throw out software patents and be done with it. No one can even agree whether the existing or proposed systems would encourage more innovation than they impede, but no one can dispute that any system of software patents involves the employment of a large army of soul-crushing patent lawyers with engineering degrees who the world would be better off if they were set to work doing actual engineering instead.
A well thought out reply, and i agree with the problems you have stated. Good point about the first corp that moves to peer review their competitor's patent application in order to prevent them from getting it. I hadn't thought about how they might collude.
So may be there isn't a proper solution, because the premise is that the idea of patents is flawed?
>So may be there isn't a proper solution, because the premise is that the idea of patents is flawed?
I don't think the idea of patents is flawed. They work well in certain industries (namely pharmaceuticals). The problem is that the patent system was never designed to cover intangible goods. The idea that one single device can simultaneously infringe tens of thousands of separate patents is an abomination.
That leaves the problem of how to exclude software from patentability without causing problems for Pfizer and General Electric, but I think that's coming at it from the wrong end. Never mind what you can get a patent to cover or not, just make intangible goods incapable of patent infringement. Pass a law that says that if all you're doing is distributing bits or executing a program on a general purpose computer, you can't be liable for patent infringement whatsoever. That should have no effect on the makers of cholesterol medication and jet turbines while solving the problem for software developers.
But that's not the problem. The problem is that you never know which patents will be asserted against you, and by that point the horse is out of the barn. You've now got to spend millions of dollars on lawyers for years to invalidate their improvidently granted patents. And the work-around doesn't do much good when they're asking for damages for past infringement on the millions of devices you've already shipped or suddenly having your trucks stopped at the border (and interrupting your business and straining your relationships with retailers) during the period before you can retool your supply chain to incorporate the work-around. Especially when the second you release a new product with a new set of work-arounds, they come at you with a new set of patents.
Software patents have got to go.