Some Thoughts On Fixing Problems In The Patent System
from the lots-of-ideas,-but-the-will-to-implement? dept
I've been procrastinating for the last week and a half or so (admittedly with a Thanksgiving holiday thrown in the middle) writing up my impressions of the recent conference at Santa Clara University called Solutions to the Software Patent Problem. If you want something of a play-by-play on what happened, Groklaw has a pretty good tick-tock on the specifics. Unfortunately, I'm not sure that really covered some of the larger important themes that the conference brought out.First off, I know that some patent system defenders took great offense to the idea that the event wasn't "balanced" with system supporters. Nearly every speaker presented an aspect of how the system was broken with suggestions to fix it. But I see this complaint as being misleading. There's this random belief out there that conference panels need to be "balanced" with "pro & con" and then let the panelists argue things out. I've argued against this in the past when setting up panels for other events, and with the events that we run, we've tried to avoid that concept as well. While just setting up pro vs. con can make for an entertaining session, it rarely leads to productive discussions that move the conversation forward. They just lead to people arguing past each other. A productive event is one in which people agree on a basic premise or problem and are then working towards possible responses. That's what this was. It was a conference for those who believe the system is broken. Given that assumption, the point of the event was to suggest possible solutions. There are plenty of events where patent system defenders and skeptics can argue against each other, but I can't think of another where people were able to dig in deep on possible fixes.
Given all that, there were a lot of different ideas proposed -- some broad, some specific. Many of them required people to have pretty deep knowledge of how the system works. For patent system geeks, it was really fascinating -- but for those less aware of the inner workings of the system I could see how the basic play-by-play seems confusing or uninteresting. However, in reality, there were a number of interesting ideas presented on ways to fix various problems, from getting Congress involved to having the courts better adhere to the law to getting the Patent Office to better adhere to rules to increasing maintenance fees to carving out software entirely... and many more. The conference schedule has links to more detailed versions of various proposals, sometimes both the "academic" versions and more "mainstream" versions at Wired Opinion. I'll point to a few of the suggestions I found to be more compelling in a bit.
However, what I found really useful about the event was twofold:
- There are a lot of smart people thinking about the very obvious problems of the system, and they are trying to come up with solutions. That's a good thing, even if there are many (powerful) interests who don't want the system to change at all. You can only deny reality for so long.
- There are also many ideas for solutions, some of which are quite clever, others of which seem like blunt force attempts to deal with symptoms rather than actual problems. Also, as was raised by some people in the comments, many of the suggestions could have significant unintended consequences that weren't readily considered. This is important -- for as much as I believe that the system needs fixing (and I have my own suggestions for solutions), thinking through the consequences of any change is an important exercise (though always keeping in mind that defenders of the status quo seem to argue any change will destroy the American economy and probably take apple pie and motherhood with it).
- There was some interest in using patent fees as a way that the patent office might be able to fix some of the worst problems without requiring a change in the law. James Bessen pointed out that bad patents are a form of pollution and, as such should be taxed as polluters. Christina Mulligan pushed back against attempts at doing the opposite, as urged by some patent system defenders, noting that "Cheaper patent applications mean more – and lower-quality! – patents. Lowering filing fees will make more people more likely to file patents. But if the marginal difference in patent fees is holding inventors back from filing applications, the patent is probably covering a low-value invention."
- Related to this, Brian Love presented some research and a proposal for using higher fees to decimate trolls. Part of his argument is that the data shows that trolls quite frequently become much more active as a patent is close to expiring: "Of all patent suits litigated within the final three years of the asserted patents' term of protection, trolls file more than 70%. Of all companies accused of infringing a patent within three years of its expiration, trolls accuse more than 83%." His suggestion is that shortening the term length of patents would deal with this problem, and would likely hurt only trolls or failing companies that are lashing out at more innovative competitors (he highlights Kodak and Encyclopaedia Britannica). But rather than actually go through the process of shortening patents, he argues just increasing the maintenance fees to keep a patent for the full term. He notes that many other countries already do something like this. The data here was compelling, though the risk is that it would just mean that patent trolls would become more active in earlier years...
- Peter Menell picked up on the point that many have raised that patents aren't at all clear in what they cover. So he thinks that we should force patents to be much clearer in what the claims cover. Specifically, he argues for a form-based system where you have a series of check boxes and pull downs in defining a claim to make it clear what you're actually trying to cover. I like the idea that patents should be much more precise over what they claim, but I'm not sure it's realistic to think that a form would (a) do this well or (b) be properly used.
- Arti Rai pointed out that the software crowd probably screwed up years ago, by not doing what the bioinformatics world did in claiming that you can't get a patent unless you provide the actual working algorithm. This seems like something of a no-brainer. The USPTO has regularly rejected bioinformatics claims on the basis of a lack of "definiteness" for not supplying the algorithm. There's no reason that the same couldn't be done in software. Of course, that would require the USPTO to actually notice this point...
- I thought John Duffy's proposal was one of the most interesting (and, in the past, I've found myself disagreeing with Duffy more often than agreeing with him) in which he argued the most "elegant" solution to the software patent problem that isn't all that far off from my own suggestion: fix the "non-obviousness" standard by noting that multiple parties simultaneously, independently inventing the same thing is a sign of obviousness and should lead to no patents being allowed. Duffy's presentation was the one that set Richard Stallman off for arguing that Stallman's solution was "kludgey" rather than a full solution. On this one, I side with Duffy. Duffy's solution here would likely be a lot more effective that Stallman's because patent lawyers could easily write around Stallman's solution of not allowing patents on "general purpose" computing.
- Samson Vermont highlighted that we should take into account social harm (Word doc). That is, he suggests courts should base remedies on the amount of harm on the public rather than on the harm on the patentee. "More specifically, in cases in which the defendant's infringement clearly makes the world better off, the court should deny the patentee both money damages and injunctive relief." More specifically, he argues that if the patent isn't actually being used to put a product on the market (either by the inventor or a licensee), the accused infringer is an independent inventor and is actually putting a product out there and the "costs to find the patentee's version of the invention beforehand were greater than the defendant's costs to invent it on his own" then courts should reject any awards. As he notes, when all three conditions are met, "it is very hard to come up with plausible reasons to believe that the defendant's infringement is a bad thing."
Overall, it was a really great and thought-provoking event. I'm happy that there's a lot of detailed thought going into ways to solve the patent trolling problem, though I'd argue we need to be careful not to just think that this is a "software patent" issue. I think the problems of the system apply equally elsewhere -- which is why I like system-wide fixes, such as those suggested by Duffy and Vermont.
And, just to address this one point before one of our regular patent lawyer commenters tries to make it in the comments: there is an argument among patent system supporters that there is no such thing as a "software patent" and thus any argument that uses that term is meaningless. This is both slightly true and (more importantly) a distortion of the larger issue. As was discussed at the conference, there is a difference between software and hardware that can't be denied. One involves moving around bits. One doesn't. So it's not difficult to define software differently from hardware. The real problem is that if we did carve out software from patentability, it's likely that crafty patent lawyers would quickly figure out how to rewrite patent claims to make them broadly cover the same concepts in a way that could be seen as not being "software." Given all that, I think it's quite legitimate to discuss "patents that cover software" as "software patents," even while I agree that merely targeting "software patents" misses the larger problem.
Thank you for reading this Techdirt post. With so many things competing for everyone’s attention these days, we really appreciate you giving us your time. We work hard every day to put quality content out there for our community.
Techdirt is one of the few remaining truly independent media outlets. We do not have a giant corporation behind us, and we rely heavily on our community to support us, in an age when advertisers are increasingly uninterested in sponsoring small, independent sites — especially a site like ours that is unwilling to pull punches in its reporting and analysis.
While other websites have resorted to paywalls, registration requirements, and increasingly annoying/intrusive advertising, we have always kept Techdirt open and available to anyone. But in order to continue doing so, we need your support. We offer a variety of ways for our readers to support us, from direct donations to special subscriptions and cool merchandise — and every little bit helps. Thank you.
–The Techdirt Team
Filed Under: artia rai, brian love, christina mulligan, dan ravicher, fixes, james bessen, john duffy, mark lemley, patents, peter menell, problems, richard stallman, samson vermont, software patents
Reader Comments
Subscribe: RSS
View by: Time | Thread
Which is it?
You'll never solve the problems with patents until you get the motive of money out of it. Taxing the hell out of unearned profits would at same time remove much of the hell from other malapportioning of societal rewards. People who make gadgets are generally much preferable to those who extract money from various mercantilism.
Support Mike "Streisand Effect" Masnick's proprietary interest!
http://en.wikipedia.org/wiki/Streisand_effect
He innovated the term all by himself! He alone! It's HIS!
[ link to this | view in thread ]
Cost of patents
[ link to this | view in thread ]
I feel this is going to get some flak because I just don't think this is accurate at all unless this is just poorly worded. They both involve moving around representations of bits using their own various mechanisms (exception being analog electronic components). I think you need another definition to differentiate software and hardware because trying to separate them based on how many bits they move around is not the way to go.
[ link to this | view in thread ]
Re: Which is it?
Is it anger that you've never come up with a term or quote that anyone cares about enough to repeat, leading to you lashing out at someone who has?
Is it due to some hilariously incorrect misconception that it makes you look 'smart', or 'witty'?
Or is it simply that you just really like the color pink, and so are trying your best to fill every comment section with as much of it as possible?
[ link to this | view in thread ]
Re:
In the first place, what is a "bit"? I can build a general purpose computer using nothing but valves and water, or lincoln logs, or marbles & tracks. On one level, each device operates on "bits". At this conceptual level, however, I could plausibly argue that all kinds of machines that that aren't computers just move "bits" around.
[ link to this | view in thread ]
[ link to this | view in thread ]
Re: Re:
[ link to this | view in thread ]
Re: Cost of patents
Not necessarily. The idea is to increase fees *AT THE END* of the patent term. That then encourages only those actually profiting from/using the patent to continue maintaining it. Those that aren't making money from it would abandon it, and let someone else take a shot...
[ link to this | view in thread ]
I think we're looking at the problem wrong
1) Obtain the rights to a ridiculously overbroad patent
2) Find people who could be interpreted as infrining that patent
3) Send out legal threats announcing an intent to sue if they don't settle
4) Set the settlement costs at somewhat below the ridiculously high amount it would cost them to fight the suit, even if they're in the right
5) Profit
Everyone keeps saying "so fix step 1", but no one seems to be looking at the actual pain point here, which is step 4. The real problem is bigger than patents; it's the fact that it is possible to use the threat of legal action--not actual legal action, but simply the threat of such--as a coercive tool, because it would cost too much to fight.
That right there is the broken system that needs to be reformed. Not just for patents, but in general. No one should ever be in the position where they honestly, legitimately believe that they're in the right, but that they simply have to roll over because they simply can't afford the legal costs. Fix that, and you not only do away with patent trolling, but a whole host of other societal ills along with it.
[ link to this | view in thread ]
I just read the Groklaw coverage of the conference
Politically I probably fall here:
Groklaw - A Report on the Santa Clara Conference on Software Patents by Tkilgore ~pj: "Next up was Crystal Sheppard, University of Nebraska School of Law. In a nutshell, she said that the problem was Congressional action or, better, inaction, toward resolving a court-caused problem. ... In essence, the problem is to change the law without passing a law."
[ link to this | view in thread ]
But what if I mean it about respect?
I know that a lot of people are guilty for crying "wolf" and thus ruining things. That's what's going on here. But actual push for respect and civility is a good thing, and we should encourage it.
Otherwise, nice article. What do you think about the Defensive Patent License idea?
[ link to this | view in thread ]
[ link to this | view in thread ]
My issues
1. The obvious standard not being enforced and/or being too lenient. Many patents would not be approved if the USPTO was much stricter about what is obvious. There is way too much old textbook stuff done on new equipment or just simple extensions of known concepts that get patented currently.
1a. As a correlary to this, prior art were actually considered. "Prior art" to the USPTO tends to consist of only what was previously patented, and neglects what was previous done in the "real" world. Obvious combinations or taking clear "next steps" do not seem to be considered.
2. Poorly defined or overbroad claims. The patent should clearly say what it covers with language that is commonly used in the field and constructed such that a person in the field can understand. The patent should cover a specific embodiment of that invention, not every possible variation on the general idea. This is the major issue with software patents, but the issue is nearly universal. It's almost impossible to read most patents can come away thinking, "oh, this patent is for x" with no doubts as to what it could be extended to cover.
[ link to this | view in thread ]
Re: Which is it?
[ link to this | view in thread ]
8 suggestions to improve the American patent system
1. Grant reimbursement of all legal and other costs made by a defendant if they (mostly) win the case, to be paid by plaintiff. This should work just like anti-SLAPP to remove the worst extortion effect: defendants won't lose millions of dollars any more even if they win, as it is now. Perhaps even add a fine for an unreasonable losing plaintiff.
2. Make *all* licensing compulsory (and fairly cheap), like FRAND. If you sue someone for patent infringement, you are required to get a quote for the licence first, from some committee. That way, if you are sued, you know what your options are, and licensing fees will be reasonable.
Fees should be scaled to both the patent holder and the defendant: a small plaintiff should never get more than, say, 50 % of its profits in fees, because otherwise they would profit too much from the other company's other activities; similarly, a small defending company should not be made to pay more than, say, 10 % of its profits on each infringing product. Fees should never be based on perceived losses of the patent holder.
3. Remove patent and trademark appeals from the Court of Appeals for the Federal Circuit and revert jurisdiction to the regular appellate courts. This mitigates regulatory capture; a court whose existence depends for a large part on a single kind of litigation, like patent lawsuits, tends to expand the scope of such litigation, thereby creating more jobs and a sense of importance for itself.
4. Forbid non-practising entities from starting patent lawsuits. This should hinder trolls a lot, and the reselling aspect of patents is that for which there is the least evidence that it benefits society.
5. Forbid the reselling of patents, see 4. Only the original "inventor" can sue. Like 4., there are ways to circumvent this, but it will be much harder.
6. Never allow juries to decide on patent cases. Laymen tend to be unpredictable and less balanced.
7. Have each new patent judged by a panel of experts in the field. That should mitigate the problem of the patent office's rubber-stamping. Of course they should be thoroughly screened for conflicts of interest.
8. Institute as an official criterion that patents are only granted for inventions that it takes a considerable amount of time, effort, or money to come up with. The idea is that the incentive provided by a patent is only needed for inventions that would otherwise not be worth the trouble. If all it takes is a moment of genius, no incentive is needed, because there is no cost: the inventor will come up with and exploit his stroke of genius anyway, even without a patent. If you know you need to carry out field tests for years on thousands of patients, then you may *need* the incentive provided by a (potential) patent to make it worth your while to proceed.
The panel of experts from 7. should be the judge of this. Rounded corners and bounce-back are obviously out of the game. License fees should also be based on the amount of effort that was involved in the invention.
[ link to this | view in thread ]
Solutions
Another complaint I've made for years has to do with the way patents are written - patents are SUPPOSED to be written for people in the field of the patent... if it takes a lawyer to understand a patent, it's not written for the practitioners and should be rejected.
[ link to this | view in thread ]
Re:
[ link to this | view in thread ]
1 SUGGESTION
that PATENTS are only for individuals, NOT CORPS..
And you get to HIRE the idiot that thought it up..until you are finished with it..
[ link to this | view in thread ]
Re: 8 suggestions to improve the American patent system
But I think you missed one thing. These arguments make all to much sense for the legislature to take them up!
[ link to this | view in thread ]
Re:
[ link to this | view in thread ]
of course you would want and NEED to stack panels, after all you do not want anyone asking difficult questions, or making valid points that you have no answer for.
But at least you have admitted that you are BIASED, and do what you can to ensure that bias..
It also shows that you are either unwilling or unable to meet valid arguments, and that you have no intention of considering any alternative argument, and you prefer it that way..
No wonder you are seem as totally ineffective in any field, now you have at least admitted as to why that is.
[ link to this | view in thread ]
Re: I just read the Groklaw coverage of the conference
Groklaw - A Report on the Santa Clara Conference on Software Patents by Tkilgore ~pj: "Next up was Crystal Sheppard, University of Nebraska School of Law. In a nutshell, she said that the problem was Congressional action or, better, inaction, toward resolving a court-caused problem. ... In essence, the problem is to change the law without passing a law."
That's not a fully accurate summary of Sheppard's presentation. Her suggestion was to use a very specific, and totally useless, form of Congressional action to deal with the problem: "sense of Congress." It was widely panned by the experts in the audience who noted that the process she suggested would not work, because a "sense of Congress" is a totally toothless and mostly useless process:
http://usgovinfo.about.com/od/uscongress/a/senseof.htm
[ link to this | view in thread ]
Re: I think we're looking at the problem wrong
There actually was a fair bit of discussion about "damages" which is addressing point 4 as well. But, again, I see that as something of a kludge to deal with the symptom, rather than the root.
[ link to this | view in thread ]
Re:
You ought to try reading it again then.
[ link to this | view in thread ]
Defining General Purpose Computers
[ link to this | view in thread ]
Re: Defining General Purpose Computers
So if someone tries to patent for instance a widget which is code running on a FPGA, they should be able to say "which parts could be replaced by a general purpose computer?", and you would be left with a FPGA (already exists, so not patentable) and VHDL code (can run on a general purpose computer, so not patentable), and there is nothing left.
On the other hand, if they added some innovative mechanical thingamajig to this example, they would have a FPGA (already exists, so not patentable), VHDL code (can run on a general purpose computer, so not patentable), and the mechanical thingamajig you invented (patentable). So the patent would be valid only for the mechanical part.
[ link to this | view in thread ]
Re: Re: Defining General Purpose Computers
We attack a programming problem by subdividing it into its component parts, subdividing these component parts into their component parts, and so on. At the bottom, we have basic elements, like basic arithmetic operations, bit manipulation, assignment of values, and so on (the exact set depends on the programming language). We do this almost by instinct, since it is the only way to express our program to the computer, and thus something we do thousands of times every working day.
So, when we hear about a patent on software, we mentally break it into its component parts (like we would do while programming), and see only basic uses of math (all basic operations of any programming language are equivalent to mathematical statements). We see it as forbidding us from using something composed only of basic uses of math, said basic uses of math being the very basis of the programming we do. It is no wonder we have such a visceral negative reaction to these patents.
[ link to this | view in thread ]
Re: Solutions
[ link to this | view in thread ]
Re: Re: I think we're looking at the problem wrong
[ link to this | view in thread ]
Re: My issues
Trolls feed on overbroad claims and "simple" or "basic" patented subject matter which are both inherently permitted by the USPTO with the lack of specificity required in claims and the weak obvious tests.
It would also make patents more closely match their intended purpose. With more definitive, less broad claims they provide better notice and allow others to learn from your patent. This teaching was an original part of the bargain with the public to get a patent, but has largely been lost with the non-specific claims and obscured language.
[ link to this | view in thread ]
Re: Re:
[ link to this | view in thread ]
Re: 8 suggestions to improve the American patent system
5. Also going to be difficult in practice. I could care less if the patent I worked on at company X is infringed upon and will never spend the money to defend it. The company sure cares though. Less lawsuits is a good thing, but they should be a valid recourse for infringement.
7. It's going to be impossible for this to work. Anyone working in the field competent enough to judge validity will inherently have a conflict of interest. You can either use it to screw competitors or reject them so you can ues it. If I work in the field, every patent granted effects me in some way. Some minor, some major.
[ link to this | view in thread ]
Software Patents
Eh, that has already happened. Software is mathematics, and mathematics are not patentable. However, patent offices and courts have come up with a completely nonsensical definition of algorithm which they apply to judge software as being not mathematics. Kind of ruling that Pi is exactly equal to "3".
Software being mathematics is mathematically proofable. So there is actually no ambiguity involved. And as it happens, it also dispells all these arguments that come up all the time, like "but if it's in an ASIC" and whatever, because it defines totally clear what is patentable (In this case: the ASIC as such, but not the logic of any possible program that runs in it).
Now, the only problem is to get patent offices and courts to acknowledge the scientifc truth. And *poof*, all the software patents would be gone ;)
Apart from that, there's another big BUT: Having no software patents would NOT solve the problem. All the problems discussed at this conference would still exist.
Because most of them are inherent in the matter that is being subjected to patents (namely: the impossibility to draw clear borders to other things, ambiguity of the language, and the fact that most innovation happens gradually and everything depends on everything else).
There's only one thing where these inherent problems don't exist, and that's chemical compounds. And if you look at the statistics, you'll see that _everywhere_ but in pharmaceuticals legal expenses are much higher than gains with licensing. Which makes it clear that the patent system is not just defect, it's completely kaputt, broken by design, at best useless and more often damaging for everything but chemical components.
You can't fix it. The only thing there is to do is to abolish it for everything where it does not work.
As for chemicals, there are other problems. It produces monsters like Monsanto. However, that is not a problem within the patent system itself. In that case it works as it is supposed to work.
[ link to this | view in thread ]
Re: Re: Cost of patents
[ link to this | view in thread ]
Re: Re:
I'm going to go for crade's definition - software is a set of instructions that can be transmitted entirely electronically. (Data can also be entirely transmitted electronically - all software is data, but not all data is software, therefore the 'set of instructions' part.)
[ link to this | view in thread ]
Software patents, and patents generally
I think three things are appropriate:
1. NO software patents,
2. Any action on infringement should be allowed only with a showing of USE, and that showing should be supported only with considerable evidence of use. If you aren't using it, you can't protect it.
3. Unless you want to sell our souls to the wealthy, we need to be sure small inventors can avail themselves of the lower cost fees mandated by Congress, and ignored by the USPTO.
[ link to this | view in thread ]
Re: Re: I just read the Groklaw coverage of the conference
Looks like it will probably be a long fight then. Trying to get much of anything through Congress will take time.
[ link to this | view in thread ]
Yeah, I've watched for a decade now system (copyright, patent, or otherwise) defenders show us their idea of "balanced" (TPP negotiations anyone?). You simply cannot be reasonable with totally unreasonable/irrational people.
[ link to this | view in thread ]
Re: Re: Re:
Software is a set of instructions, nothing more or less. The medium it's in isn't relevant.
[ link to this | view in thread ]
Re: Re: Re: I think we're looking at the problem wrong
It's considered a symptom because a troll cannot sue (or at least not survive a motion to dismiss) if it does not have an issued patent in the first place. Suing for patent infringement without an issued patent would be deemed frivolous litigation, and that would typically entail the plaintiff paying the defendant's legal costs.
[ link to this | view in thread ]
1. Any patent/trademark/copyright that is in the public domain cannot be, or have any aspect of it re-protected by another patent.
2. Suing for infringement requires that the suing party pay the legal expenses of the defense party. These fees will be refunded and shifted onto the defense if they are ultimately found guilty.
3. If you sue for a patent/trademark/copyright infringement and lose, your patent/trademark/copyright is immediately revoked and enters the public domain. Any pending lawsuits related to that patent are immediately thrown out.
[ link to this | view in thread ]
Re:
[ link to this | view in thread ]
Re: Re: Re: Re:
instructions is too specific, since data that the instructions operate on is also software.
[ link to this | view in thread ]
Re: Re: Re: Re:
[ link to this | view in thread ]
Re: Re: Re: Re: I think we're looking at the problem wrong
Yes, if the troll did not have any patents, they wouldn't be able to sue for infringement. But in that case they'd just find something else to do, because the patents are simply a means to an end. The business model here is "use X as leverage to get people to pay up to avoid an expensive lawsuit." Right now, X is patents. Remove patents from the equation, and they'll find a different X. Remove that, and they'll find another one. As long as the fundamental problem goes unaddressed, it just turns into a game of whack-a-mole.
[ link to this | view in thread ]
Re: Re: Which is it?
[ link to this | view in thread ]
Re: Re: 8 suggestions to improve the American patent system
Ad 4: you can license it out, but the shell company can't sue, so all you can license out is the permission to use the invention, not the monopoly, i.e. not the power to enforce it.
Ad 5: By "inventor" I meant whoever applied for the patent in the first place. If that is your company, then they get the monopoly, but they cannot sell it to anyone else. Indeed, they could sell an exclusive license to a troll and try to enforce their monopoly to protect the troll; but that could only happen if they were somehow subverted by the troll while still making products using the patent themselves (or they would run afoul of 4). This would be a lot more complicated than simply selling the patent and the power to enforce it, and nearly impossible in conjunction with 4.
Ad 7: what if panellists were screened case by case, and only the most egregious cases removed? The idea is that they decide together, and perhaps a qualified majority (2/3 in favour) should be required for each patent. If fewer patents are granted this way, then great.
[ link to this | view in thread ]
upkeep
[ link to this | view in thread ]