![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
Weird Corollaries of the Way Things Work Now
Number 3,293: finding out from LinkedIn that you have your name on six software patents.
Seriously -- the most recent of those I didn't even know was still in the works until I went to update my profile just now. It's a bit jaw-dropping, because (a) the company (Convoq) has been out of business for a dozen years now, and (b) it's a somewhat dangerous patent. Unlike many software patents, this one isn't pure BS -- we were well ahead of the game in 2004, and pioneered a bunch of techniques that were new at the time.
Now I'm wondering whether it's been used against any of the players in the web-meeting industry. One of our headline features (Meet ASAP) still doesn't exist anywhere that I've seen, and I'm wondering if that's because this patent has prevented anybody from implementing it.
Sigh -- I hate, hate, hate the software-patent game. I wish Congress would get up the nerve to just obliterate the entire system: it's nothing but sand in the wheels of the software industry, making the rich richer and screwing the little guys. Unfortunately, as a senior engineer, you rarely have much choice but to play the game...
no subject
Another point of possible disagreement: I firmly believe that software patents are *not* a fundamental human right, the way many companies treat them. The system exists to promote societal good, specifically to promote innovation. If it isn't being a net positive for society, it should be trashed, IMO, and I'm dubious that there is a way to shift the needle that far.
All that said: I've seen an awful lot of the sausage-making here (I've spent several full-time months of my career writing and editing software patents), and have wound up especially cynical about the whole thing as a result. YMMV...
So...
I don't think anyone ever asserted that any patent is a fundamental right. A patent is a form of government-granted limited monopoly. It's unclear to me why inventors in this domain are any more or less deserving of these grants than inventors in any other domain. One might assert (and I might agree) that the current US patent system has strayed so far from the original intent in the Constitution as to be unconstitutional. In which case, that applies to all patents; I see no cause to apply that reasoning solely to software patents.
Re: So...
No, but the financial design of the industry is wildly different. "Small" companies in the software-industry sense are difficult-to-impossible to create in pharma, for reasons largely unrelated to their patent situations -- you can't really have a true garage startup for pharma, because even getting to the starting line requires a good deal of money. So the money to research, submit and fight patents exists in a way that it doesn't for bootstrapping software startups.
Moreover, without patents much of the pharma industry *couldn't exist* -- the cost of entry is so enormously high that, without the promise of patents, nobody would enter it. That's exactly what patents are for, in a way that has never been especially true for software and is even less true today.
There's also a difference in defining the scope of a patent. The problem with software is that it is so very *squishy* -- it's often hard to state precisely what problem a given innovation solves (which is part of why software patents weren't even legal until relatively recently; they were considered a non-sequiteur to the whole patent concept), and the "process" being described is necessarily relatively vague if you're going to produce a patent that is more than a glorified copyright (which is why software patents are dominated by the sorts of useless flow charts that the rest of the industry stopped using decades ago). I'm skeptical that there is a straightforward technique for properly defining software patent scope, at least in the general case. It's telling that such a large fraction of software patents are obvious, over-reaching bullshit.
And that is *precisely* what I'm arguing against. "Deserving" has *nothing to do with it* -- the question is, is it in the best interest of society that this specific field be patentable? I say no: because software is so cheap to create and to innovate in, it requires far less *incentive* to innovate -- and providing that incentive is the sole purpose of the patent system. You're making exactly the category error I'm talking about when I talk about people speaking in terms of "rights". It's not about rights or deserving: it's about an overall societal good being fostered by providing this monopoly within a given field.
In the software case, it's unnecessary *and* it retards innovation, by providing a mechanism whereby giant companies can obliterate small ones -- not by out-inventing or out-competing them, but by out-suing them. Buying up gigantic warchests of patents is a central part of the business plans of many of the big companies -- The Economist, for one, has devoted non-trivial space over the past few years to describing this effect, and the way that it is gradually squashing the small and innovative part of the software economy.
Software patents (indeed, business process patents in general) are both harmful and unnecessary. Is it possible to devise a system that addresses both of those? Maybe, but like I said, I'm skeptical...
Re: So...
Biopharma has not, yet, seen the kind of garage start-ups that software has, but that's in part because the means of production have been expensive. They've fallen by an order of magnitude in each of the past two decades and I expect they will continue on that curve. It's not Moore's-law cheap but it will be "four guys in an attic" cheap this decade where it's not already.
That said, the point was not the start-up cheapness, but rather the idea that patents dealing with complex topics (software) are somehow qualitatively different from patents in other fields. I picked biopharma because although data are harder to come by quickly now, it was a field with roughly comparable numbers of patents filed/awarded to software last time I looked. (I'm too lazy/busy to dig into the USPTO charts, which are differently structured now.)
If your assertion is no longer that the amount of information in software is not more than biopharma then I am confused.
without patents much of the pharma industry *couldn't exist* -- the cost of entry is so enormously high that, without the promise of patents, nobody would enter it.
This is almost exactly backward. Patent protection is most useful when the machine is easy/cheap to copy. If the thing is hard to copy that itself provides a barrier and gives you a de facto monopoly. The patent gives you a de jure monopoly, which is useful when you expect copy competition.
Patents are useful in biopharma because of the cost recovery/profit curve. Producing the product involves a very large up-front outlay, and large uncertainty due to externalities like regulatory approval. In effect, you're making money on product A to cover your expenses and often losses on sunk costs of product B. Monopolies extend the lifetime of your exclusivity so that you can reach the point where expenses are paid and all the revenue is profit. Patents are also useful for the smaller companies that are, effectively, betting the house.
Pharma exists and flourishes in places with weak or no IP regimes; they just don't make the obscene profits that first world biopharma companies make. But this is way off topic.
The problem with software is that it is so very *squishy* -- it's often hard to state precisely what problem a given innovation solves (which is part of why software patents weren't even legal until relatively recently; they were considered a non-sequiteur to the whole patent concept)
So now we're back to the "software is special" concept. In this model, you can't state precisely enough what problem a program solves to patent it, but you can state it precisely enough to convince people to give you money for it. I am VERY VERY dubious of that paired claim.
Benson was decided in 1972, so "relatively recently" means 47 years (if I've done my math right). That's something like 75% of the lifetime of electronic programmable computers, so I disagree with this claim as well.
they were considered a non-sequiteur to the whole patent concept
What I think you're pointing to is the machine/method distinction that Benson tried to make, and that brings me back to asking you whether you believe Church-Turing. Do you, in fact, believe that any program is equivalent to a machine of a certain description?
If you do, then you need to explain why some machines are special and can't be patented while others can. You also need to tell me why I can't get a patent on my program if I write it in a compiled language but I can if I create a custom chip with that program burned into it.
It's telling that such a large fraction of software patents are obvious, over-reaching bullshit.
This is the equivalent of claiming that because people can use cars to do terrible things and some classes of drivers are terrible that therefore cars shouldn't be allowed. I'll cheerfully agree that software patents are a mess, and I've provided three concrete steps that should be taken to untangle the mess (which are similar to steps taken in other patent disciplines). I'm not sure if you're just angry at the mess and so ignoring that, or you have reasons why you think these changes wouldn't be helpful.
the question is, is it in the best interest of society that this specific field be patentable? I say no: because software is so cheap to create and to innovate in, it requires far less *incentive* to innovate -- and providing that incentive is the sole purpose of the patent system. You're making exactly the category error I'm talking about when I talk about people speaking in terms of "rights". It's not about rights or deserving: it's about an overall societal good being fostered by providing this monopoly within a given field.
So in your world, we'd restrict the categories of machines that are patentable to those that have social benefits? Who gets to decide those benefits? Is there a threshold of benefit that allows patenting? Can I patent a machine that benefits only a small percentage of society (like people with very rare diseases, or people trying to live on Mars)? Can I patent a machine where the benefit is speculative - I can't demonstrate current specific benefits? Can I patent machines that are targeted to hobbies and frankly socially wasteful activities? I mean, why should I get a patent on a new kind of confection manufacture?
My point is not that anyone is deserving; it's that you're singling out software with nothing to sustain why software should be singled out. I don't think you've thought this through, but perhaps I'm missing something.
In the software case, it's unnecessary *and* it retards innovation, by providing a mechanism whereby giant companies can obliterate small ones -- not by out-inventing or out-competing them, but by out-suing them. Buying up gigantic warchests of patents is a central part of the business plans of many of the big companies -- The Economist, for one, has devoted non-trivial space over the past few years to describing this effect, and the way that it is gradually squashing the small and innovative part of the software economy.
And yet, the number of software start-ups continues to climb. I don't dispute that people behave badly with software patents. They do horrible shit with drug patents, too, such as keeping generics from coming to market by chaining patents. I believe that my stated solutions would help with some of these problems (particularly reducing junk patents and compulsory licensing). Again. bad behavior exists in all kinds of businesses. Railing about the particular badness of software companies is important and I enthusiastically join in. But as I said above, bad drivers is not a reason to ban cars.
Software patents (indeed, business process patents in general) are both harmful and unnecessary.
That's both conflating two things (software machines and business processes) and argument by emphatic repetition. I don't know how to respond to that.
Re: So...
That's probably true.
Which is what I just said -- the cost of entry is enormously high. You can't just build a better mousetrap: you have to spend a good fraction of a billion dollars to get that mousetrap to market because of the FDA. That being the case, patents are necessary.
You're entirely free to disagree, but anything in my lifetime is "recently" compared to the overall existence of the patent system. And it wasn't just about software, it was a change in the fundamental understanding of what is patentable.
I'm not saying that the latter should be. And yes, I'm well aware of the chains of reasoning involved here, but the ability to patent algorithms in the *general* case is, IMO, where the system fell off a cliff.
I'm dubious about that assertion. Most folks I know who are paying close attention to the industry consider the startup world to be in significant trouble, because of the growing concentration of power into the giants. And much of that power comes from their gigantic patent arsenals. The assessments I've been seeing for the past few years are that startups are getting less ambitious as a result.
Dude -- I've spent 20 years thinking about this. 20 years of observing it from the inside, being forced to write these stupid patents. 20 years of worrying about companies being put out of business by this idiocy. I've thought about it a lot.
We're done here -- I think we're unlikely to come to an agreement. Suffice it to say, I think that implementing your ideas in a way that isn't easily abuseable, and results in benefits that are actually worthwhile, is unlikely. I'd love to believe that you're correct, but as I said -- I'm skeptical. You're welcome to your optimism, but there are a lot of devils in those details. And for the time being, the system as it stands is destructive...