One of the things that has struck me, looking at a wide variety of legal systems past and present, is the important role of feud as a form of law enforcement.
For some decades, I felt that the patent system did more to deter innovation (by threat of company-killing lawsuit) than to promote it. (At least outside the pharmaceutical industry.)
I eventually changed my mind. Without patent protection against low-effort knock-offs, it can be really difficult for small firms to get VC funding.
But you're correct - patent trolls are a huge problem. I was on the H.264 video codec standardization comittee. The more recent standards have been held up by fights between the dozens of patent holders who can't agree on terms - a huge problem. I suspect we need a must-license-on-RAND-terms rule, at least for patents applying to formally standardized tech.
Maybe they have less costs because they don't have to invent the thing? But much of the criticism of the patent system (at least in the software patent world) is that patents are awarded too easily, for things that contain no new innovation at all, or ideas that have occurred to someone while taking a shower, or at most took an engineer tens of hours to invent, much less than the cost of patent litigation, and not something that needs VC levels of funding.
Indeed, in your argument the small firm has already invented the thing *before* getting VC funding.
(What I mean by patents awarded too easily: the x264 H.264 encoder has 140000 lines, while I find that H.264 has been covered by some 7500 patents; that's one patent per 19 lines of code. Even if some of the patents are not for the format itself, but for particular algorithms to efficiently encode or decode it, decode it in hardware etc., these can't all plausibly be significant innovations, even if a format like this does deserve a handful of patents.)
----
Or maybe the argument is that the first company to make a new kind of product takes more risk because it's not yet clear if there's demand for the product, while knock-offs appear once it's clear that there's demand? That's a different problem than what the patent system is designed to solve.
Indeed, this argument could be relevant even when a new product idea doesn't require any technical invention (i.e. once given the goal, it's obvious to any competent professional how it can be done). AFAIUI in that case the new product idea alone isn't supposed to be patentable in theory, though again, part of the criticism of the patent system is precisely that patents get awarded even when they aren't supposed to.
All great questions. And I'm no expert on this - I'm just speaking from 40 years of observation in industry. To be clear, I've changed my mind once on this and may change it again - I could easily be wrong.
Low-effort knock-offs don't need as much funding (often dramatically less) because the first mover has already figured out a lot of hard problems - the knock off only needs to copy the solutions. A real product in a new space involves lots of choices - many of which are hard to make or require a lot of trial-and-error. This is costly (tho it may not involve any patentable invention). Patents give first movers some space to work that out.
Yes, in my model the small firm has made the "invention" and filed patents *before* getting funding. But there is an awful lot more innovation required to make a practical product than just the patentable invention. Lots of hard choices. None of which are individually patentable. But expensive to get right.
When I left the H.264 work (around 2006) there were "only" about 200 known patents. I don't know where you get the 7500 number - if that's true I'm amazed. There's a difference of course between (a) "claimed" applicable patents, (b) patents that an independent and disinterested 3rd party expert has decided are applicable (we hired Cliff Reader to do this), and (c) patents that a court has decided are applicable. Claims are easy to make. (Plus, maybe you're counting expired patents?) Also, H.264 standadrizes only DECODERS not ENCODERS (tho the standards committe does in practice release a "reference model" encoder - generally it's not very computationally efficient or something you'd want to use in a product). Even for decoders what the SDO releases is only a reference model (the standard is defined in words not code) - people build their own implementations, so maybe you're talking about some exemplar codebase.
But I agree with your implied point - it is in general way too easy to get a patent. The PTO doesn't really make any attempt to implement the "non-obviousness" requirement; of the patents I've seen that acutally issued, I'd say about 85% should not have passed that test - most competent engineers would have come up with the same thing if encoutering the same problem. I'd call anything the typical 90th percentile engineer can be expected to do, "obvious".
Whether it's a *good idea* to implement that test is the real question tho, in social terms. Is it better to issue lots of weakly-justified patents (as we do now) to effectively give startups and inventors a window without competition, or to allow independent invention? I used to feel idependent invention (soon) was a *demonstration* of obviousness. I don't know, but it doesn't seem as obvious now as it once did.
I agree that the patent system is not designed to solve this, but maybe it *is* partially solving this? And maybe that's a good thing? I'm not sure.
One thing I will say - as I've seen more over time, I'm impressed by how much of common law (law made by judges and precedent, not legislation) is ultimately wise and pro-social. Often in non-obvious ways to the naive. Maybe the PTO knows what it's doing.
For the codebase I used x264, the open source H.264 encoder usually used with ffmpeg. A fully compliant decoder necessarily contains all information the standard does, so it's an upper bound on the information content of the format; an encoder doesn't necessarily do so (it might not use all features the standard allows), though I assume it does use the important ones. ffmpeg itself has 91000 H.264-specific lines (going by file names; this seems to be a decoder).
Wow. Given that H.264 was first approved by ITU-T in May 2003 I'm amazed that more patents haven't expired (that Wikipedia list is as of December 2022).
There are some rules I don't understand that allow patent holders to extend the life of their patents; maybe that's relevant.
That's IMHO outragous. I can also tell you there is a strong *negative* correlation between the number of claimed patents in that list and the orgs that contributed most to the the technical work.
Something seems very broken here. I'm not entirely sure what.
--- Added: I was present at all the early MPEG-LA meetings re setting up the H.264 patent pool. I have a good idea how much lawyer time goes into EACH patent in the pool. And a good idea how much engineer time went into H.264's design. There is MUCH more investment in lawyers than in engineers. Maybe by a factor of 20.
Most charitably some of the patents are about stuff like how to implement hardware-accelerated decoding on particular hardware, and have been invented (or "invented") and filed later than the format was standardized. But the number still seems excessive. Also, in a similar vein, my use of the codebases to estimate the complexity of the format was also a slight overcount, as they contain some optimized assembly for multiple CPU architectures.
You kind of touch on it in your "King's Friend" post but the part you are missing in a feud system is it only works among immediate peers. You point out the Patent Troll problem but you miss the inverse, the "I'm Apple, you're some abject poverty mad scientist / small business", i.e. Apple is free to LEGITIMATELY steal your patents because they have an entire staff of lawyers sitting around doing nothing (sunk cost) whereas you can't even afford a lawyer on consignment. You might win once, probably won't, THEY will eventually win because that have nothing else to do but launch a new suit every single day until one sticks. Likewise the "I control the courts or statute", i.e. in the Federal system you see that with the US Federal Court of Claims where you effectively only bring a claim if you are willing to burn every bridge OR if you are indispensable because magically the USG will bring a criminal investigation against you for filing sometimes (i.e. not tort for tort) and likewise, the USG is immune to punitive and substantial real damages ergo no judge is going to award a $5 Trillion dollar award against the USG and even if they did, no Congress is going to authorize it's payment. Caselaw is littered federally with people that have "won" and never got paid because what the judges can't do is force Congress to appropriate funds. The problem is even worse at the state, county, and municipal levels, it's often why cities don't even care about getting sued, because they are already billions in debt from other lawsuits so they know it's uncollectable.
Feuds work in your quoted system because they are nominal peers. Whereas my feud with Musk means Musk can harm me each and every day because he's rich enough to pay the "harm" when he occasionally loses whereas I can't afford to sue him even once, and even if I could, to your point about imperfect results, he could retaliate with infinite countersuits, I can't, hence he's free to legitimately violate my patent all day every day. And if my feud is with the courts paymaster who violates my patent, you want to bet how the court will rule, i.e. "pay me or cut his own pay". Public Choice Theory.
I'll give a criminal law equivalent. My PO routinely violates my civil rights, and I don't mean that in a "every convict complains sort of way", I mean I have directly on record in writing been told if I go to church I will get violated on a "unrelated technical" and go to prison because "religion is made up hence any person that goes to church is delusion and delusion is a sign of mental illness hence prison for you". Likewise court stated "Your home can be inspected during reasonable hours" whereas my PO said "I'm defining reasonable as 24x7, complain and get violated". Now sure I can "sue" via feud system and maybe I'll even win, but the countersue is "prison for you" and that's not MAD, that's "go to jail, don't pass go" become of things like qualified immunity, professional courtesy, etc .. like even if I got my PO fired, the next one will remember that and "hook me up". Caselaw likewise if full of cases where people have whistle blew, sued, etc only be have a retaliatory "go to prison" arrest happen immediately afterwards, i.e. "you want to sue us? Ok, enjoy those proceed rotting in prison".
That isn't to say feud law doesn't work, it does as you said, but only among peers. Still most harms aren't peer related except accidentally because generally you don't purposely harm on a coin flip.
Some of both. When writing a post on something I have written on before, I do a good deal of copy/paste/edit. I'm currently working on a post on ideas we can steal from past legal systems. Most of the contents are from the corresponding chapter of my book, selected and edited. Other posts are entirely new.
But I am sure I also repeat myself in giving talks.
Non-practicing entities are a natural outcome of patent transferability. Many new inventions are infeasible for individual inventors to directly monetize. They only add value as a part of bigger systems. Inventions are made by people who are employees of companies.
So we allow patents to be sold or transferred. Typically to large companies who can implement the patents for money or to the employer of the employee.
But that also means that patents can be transferred to organizations which don't employ inventors or implement the patented inventions. And they still have the same rights for enforcement because that attaches to the patent.
One solution would be to only allow lawsuits for the value of the lost business of the inventor from that patent. But that basically moots values to individual inventors. No lone inventor is going to be able to directly monetize new eg. semiconductor manufacturing technology, allowing large companies to violate the patent for no real cost. It also would hinder future development.
Consider a company in an industry which invents technology A. This technology makes their products better in the market allowing them to attract more customers/charge a higher price/whatever. Assuming no additional invention on their part, the competition is left at baseline (or paying for licensing). Now, during the lifetime of that patent assume that the company now invents technology B which is better and wholly replaces technology A. The company now would like to be even better off than when they invented technology A as their competition is left at baseline. But if you can only sue for damages lost for the patent in-use, the competition could now freely use technology A under no threat of damages being awarded. So the inventing company would have paid for the invention of two technologies concurrently but would only get the benefits of one of them.
One problem with patent trolls is meritful but costly lawsuits (at least according to the story I hear about software patents).
The Mutually Assured Destruction you describe works not only between companies that don't violate each other's patents (but might otherwise consider launching meritless lawsuits), but also between companies that both violate each other's patents. Patents are awarded easily enough that companies like Apple and Samsung probably violate a bunch of each other's patents all the time, even inadvertently. Most of the time they don't sue each other, as figuring out the exact amount of money they owe each other would probably cost more in legal costs than whatever they owe each other on the net.
But if Apple sells a third of its patents to a patent troll, the troll can sue Samsung, while Apple still retains plenty of firepower to shoot back in the event Samsung sues it. And if they both sell patents to trolls, the result is a bunch of costly lawsuits they could otherwise avoid.
Perhaps part of the solution is for Samsung to threaten to sue Apple for patent violations not only if Apple sues it, but also if Apple sells any patents to patent trolls, either immediately, or once the troll sues Samsung? (Kind of like a nuclear power threatening to nuke another nuclear power if it gives nukes to a terrorist group, either for the mere act of doing so, or if the terrorist group uses the weapon against the former country.)
----
Another hole in the system, which you mentioned towards the end, is that it's part of the system that even a meritless patent suit imposes a significant cost on the defendant, which allows patent holders to shake down or extinguish small companies that can't afford the cost. But it's not just patent trolls that can do this, but large, productive companies too.
How were non-practicing entities dealt with in iceland? They didn't have patents (assuming), so I'm assuming that once you get sufficiently annoying, somebody just kills you?
There is someone in the Tale of Sarcastic Halli who has the reputation of killing people and never paying wergeld — Norwegian, semi-independent allied with King Harald. He's exploiting his power and reputation. Halli, who is a skald not a warrior, gets him to pay wergeld by a fraudulent claim that someone he killed was Halli's brother, using a reputational threat.
I think the usual abuse you see, as there, is not false claims but refusal to pay on true claims. But there is a berserk in Egilsaga who claims land and defends the false claim by offering to duel the representative of the owner. The only representative available is the owner's son, a youth. So Egil offers to do it instead and kills the berserk. That is an example of one of the ways of defending the rights of the weak that I think I discuss in the chapter on feud law, observed in the sagas and the Commanche.
For some decades, I felt that the patent system did more to deter innovation (by threat of company-killing lawsuit) than to promote it. (At least outside the pharmaceutical industry.)
I eventually changed my mind. Without patent protection against low-effort knock-offs, it can be really difficult for small firms to get VC funding.
But you're correct - patent trolls are a huge problem. I was on the H.264 video codec standardization comittee. The more recent standards have been held up by fights between the dozens of patent holders who can't agree on terms - a huge problem. I suspect we need a must-license-on-RAND-terms rule, at least for patents applying to formally standardized tech.
https://nerdfever.com/the-only-good-patent-is-an-expired-patent/
How do low-effort knock-offs get VC funding?
Maybe they have less costs because they don't have to invent the thing? But much of the criticism of the patent system (at least in the software patent world) is that patents are awarded too easily, for things that contain no new innovation at all, or ideas that have occurred to someone while taking a shower, or at most took an engineer tens of hours to invent, much less than the cost of patent litigation, and not something that needs VC levels of funding.
Indeed, in your argument the small firm has already invented the thing *before* getting VC funding.
(What I mean by patents awarded too easily: the x264 H.264 encoder has 140000 lines, while I find that H.264 has been covered by some 7500 patents; that's one patent per 19 lines of code. Even if some of the patents are not for the format itself, but for particular algorithms to efficiently encode or decode it, decode it in hardware etc., these can't all plausibly be significant innovations, even if a format like this does deserve a handful of patents.)
----
Or maybe the argument is that the first company to make a new kind of product takes more risk because it's not yet clear if there's demand for the product, while knock-offs appear once it's clear that there's demand? That's a different problem than what the patent system is designed to solve.
Indeed, this argument could be relevant even when a new product idea doesn't require any technical invention (i.e. once given the goal, it's obvious to any competent professional how it can be done). AFAIUI in that case the new product idea alone isn't supposed to be patentable in theory, though again, part of the criticism of the patent system is precisely that patents get awarded even when they aren't supposed to.
All great questions. And I'm no expert on this - I'm just speaking from 40 years of observation in industry. To be clear, I've changed my mind once on this and may change it again - I could easily be wrong.
Low-effort knock-offs don't need as much funding (often dramatically less) because the first mover has already figured out a lot of hard problems - the knock off only needs to copy the solutions. A real product in a new space involves lots of choices - many of which are hard to make or require a lot of trial-and-error. This is costly (tho it may not involve any patentable invention). Patents give first movers some space to work that out.
Yes, in my model the small firm has made the "invention" and filed patents *before* getting funding. But there is an awful lot more innovation required to make a practical product than just the patentable invention. Lots of hard choices. None of which are individually patentable. But expensive to get right.
When I left the H.264 work (around 2006) there were "only" about 200 known patents. I don't know where you get the 7500 number - if that's true I'm amazed. There's a difference of course between (a) "claimed" applicable patents, (b) patents that an independent and disinterested 3rd party expert has decided are applicable (we hired Cliff Reader to do this), and (c) patents that a court has decided are applicable. Claims are easy to make. (Plus, maybe you're counting expired patents?) Also, H.264 standadrizes only DECODERS not ENCODERS (tho the standards committe does in practice release a "reference model" encoder - generally it's not very computationally efficient or something you'd want to use in a product). Even for decoders what the SDO releases is only a reference model (the standard is defined in words not code) - people build their own implementations, so maybe you're talking about some exemplar codebase.
But I agree with your implied point - it is in general way too easy to get a patent. The PTO doesn't really make any attempt to implement the "non-obviousness" requirement; of the patents I've seen that acutally issued, I'd say about 85% should not have passed that test - most competent engineers would have come up with the same thing if encoutering the same problem. I'd call anything the typical 90th percentile engineer can be expected to do, "obvious".
Whether it's a *good idea* to implement that test is the real question tho, in social terms. Is it better to issue lots of weakly-justified patents (as we do now) to effectively give startups and inventors a window without competition, or to allow independent invention? I used to feel idependent invention (soon) was a *demonstration* of obviousness. I don't know, but it doesn't seem as obvious now as it once did.
I agree that the patent system is not designed to solve this, but maybe it *is* partially solving this? And maybe that's a good thing? I'm not sure.
One thing I will say - as I've seen more over time, I'm impressed by how much of common law (law made by judges and precedent, not legislation) is ultimately wise and pro-social. Often in non-obvious ways to the naive. Maybe the PTO knows what it's doing.
H.264 (alleged) patents: https://en.wikipedia.org/wiki/MPEG_LA#H.264/MPEG-4_AVC_licensors
Yes, I've counted expired ones.
For the codebase I used x264, the open source H.264 encoder usually used with ffmpeg. A fully compliant decoder necessarily contains all information the standard does, so it's an upper bound on the information content of the format; an encoder doesn't necessarily do so (it might not use all features the standard allows), though I assume it does use the important ones. ffmpeg itself has 91000 H.264-specific lines (going by file names; this seems to be a decoder).
Wow. Given that H.264 was first approved by ITU-T in May 2003 I'm amazed that more patents haven't expired (that Wikipedia list is as of December 2022).
There are some rules I don't understand that allow patent holders to extend the life of their patents; maybe that's relevant.
That's IMHO outragous. I can also tell you there is a strong *negative* correlation between the number of claimed patents in that list and the orgs that contributed most to the the technical work.
Something seems very broken here. I'm not entirely sure what.
--- Added: I was present at all the early MPEG-LA meetings re setting up the H.264 patent pool. I have a good idea how much lawyer time goes into EACH patent in the pool. And a good idea how much engineer time went into H.264's design. There is MUCH more investment in lawyers than in engineers. Maybe by a factor of 20.
Supposedly (https://en.wikipedia.org/wiki/Advanced_Video_Coding#Licensing) at least one patent lasts until 2030; this one (https://patents.google.com/patent/US9356620B2/) was filed in 2002, but only granted in 2016 for some reason, and got a ~8 year extension.
Most charitably some of the patents are about stuff like how to implement hardware-accelerated decoding on particular hardware, and have been invented (or "invented") and filed later than the format was standardized. But the number still seems excessive. Also, in a similar vein, my use of the codebases to estimate the complexity of the format was also a slight overcount, as they contain some optimized assembly for multiple CPU architectures.
You kind of touch on it in your "King's Friend" post but the part you are missing in a feud system is it only works among immediate peers. You point out the Patent Troll problem but you miss the inverse, the "I'm Apple, you're some abject poverty mad scientist / small business", i.e. Apple is free to LEGITIMATELY steal your patents because they have an entire staff of lawyers sitting around doing nothing (sunk cost) whereas you can't even afford a lawyer on consignment. You might win once, probably won't, THEY will eventually win because that have nothing else to do but launch a new suit every single day until one sticks. Likewise the "I control the courts or statute", i.e. in the Federal system you see that with the US Federal Court of Claims where you effectively only bring a claim if you are willing to burn every bridge OR if you are indispensable because magically the USG will bring a criminal investigation against you for filing sometimes (i.e. not tort for tort) and likewise, the USG is immune to punitive and substantial real damages ergo no judge is going to award a $5 Trillion dollar award against the USG and even if they did, no Congress is going to authorize it's payment. Caselaw is littered federally with people that have "won" and never got paid because what the judges can't do is force Congress to appropriate funds. The problem is even worse at the state, county, and municipal levels, it's often why cities don't even care about getting sued, because they are already billions in debt from other lawsuits so they know it's uncollectable.
Feuds work in your quoted system because they are nominal peers. Whereas my feud with Musk means Musk can harm me each and every day because he's rich enough to pay the "harm" when he occasionally loses whereas I can't afford to sue him even once, and even if I could, to your point about imperfect results, he could retaliate with infinite countersuits, I can't, hence he's free to legitimately violate my patent all day every day. And if my feud is with the courts paymaster who violates my patent, you want to bet how the court will rule, i.e. "pay me or cut his own pay". Public Choice Theory.
I'll give a criminal law equivalent. My PO routinely violates my civil rights, and I don't mean that in a "every convict complains sort of way", I mean I have directly on record in writing been told if I go to church I will get violated on a "unrelated technical" and go to prison because "religion is made up hence any person that goes to church is delusion and delusion is a sign of mental illness hence prison for you". Likewise court stated "Your home can be inspected during reasonable hours" whereas my PO said "I'm defining reasonable as 24x7, complain and get violated". Now sure I can "sue" via feud system and maybe I'll even win, but the countersue is "prison for you" and that's not MAD, that's "go to jail, don't pass go" become of things like qualified immunity, professional courtesy, etc .. like even if I got my PO fired, the next one will remember that and "hook me up". Caselaw likewise if full of cases where people have whistle blew, sued, etc only be have a retaliatory "go to prison" arrest happen immediately afterwards, i.e. "you want to sue us? Ok, enjoy those proceed rotting in prison".
That isn't to say feud law doesn't work, it does as you said, but only among peers. Still most harms aren't peer related except accidentally because generally you don't purposely harm on a coin flip.
I've read and listened to a lot of your stuff. That first paragraph I've heard/read word for word a number of times.
Out of curiosity, is it something you copy paste or did you just find a way to say it you liked and it just comes out every time you write/speak?
Some of both. When writing a post on something I have written on before, I do a good deal of copy/paste/edit. I'm currently working on a post on ideas we can steal from past legal systems. Most of the contents are from the corresponding chapter of my book, selected and edited. Other posts are entirely new.
But I am sure I also repeat myself in giving talks.
Non-practicing entities are a natural outcome of patent transferability. Many new inventions are infeasible for individual inventors to directly monetize. They only add value as a part of bigger systems. Inventions are made by people who are employees of companies.
So we allow patents to be sold or transferred. Typically to large companies who can implement the patents for money or to the employer of the employee.
But that also means that patents can be transferred to organizations which don't employ inventors or implement the patented inventions. And they still have the same rights for enforcement because that attaches to the patent.
One solution would be to only allow lawsuits for the value of the lost business of the inventor from that patent. But that basically moots values to individual inventors. No lone inventor is going to be able to directly monetize new eg. semiconductor manufacturing technology, allowing large companies to violate the patent for no real cost. It also would hinder future development.
Consider a company in an industry which invents technology A. This technology makes their products better in the market allowing them to attract more customers/charge a higher price/whatever. Assuming no additional invention on their part, the competition is left at baseline (or paying for licensing). Now, during the lifetime of that patent assume that the company now invents technology B which is better and wholly replaces technology A. The company now would like to be even better off than when they invented technology A as their competition is left at baseline. But if you can only sue for damages lost for the patent in-use, the competition could now freely use technology A under no threat of damages being awarded. So the inventing company would have paid for the invention of two technologies concurrently but would only get the benefits of one of them.
One problem with patent trolls is meritful but costly lawsuits (at least according to the story I hear about software patents).
The Mutually Assured Destruction you describe works not only between companies that don't violate each other's patents (but might otherwise consider launching meritless lawsuits), but also between companies that both violate each other's patents. Patents are awarded easily enough that companies like Apple and Samsung probably violate a bunch of each other's patents all the time, even inadvertently. Most of the time they don't sue each other, as figuring out the exact amount of money they owe each other would probably cost more in legal costs than whatever they owe each other on the net.
But if Apple sells a third of its patents to a patent troll, the troll can sue Samsung, while Apple still retains plenty of firepower to shoot back in the event Samsung sues it. And if they both sell patents to trolls, the result is a bunch of costly lawsuits they could otherwise avoid.
Perhaps part of the solution is for Samsung to threaten to sue Apple for patent violations not only if Apple sues it, but also if Apple sells any patents to patent trolls, either immediately, or once the troll sues Samsung? (Kind of like a nuclear power threatening to nuke another nuclear power if it gives nukes to a terrorist group, either for the mere act of doing so, or if the terrorist group uses the weapon against the former country.)
----
Another hole in the system, which you mentioned towards the end, is that it's part of the system that even a meritless patent suit imposes a significant cost on the defendant, which allows patent holders to shake down or extinguish small companies that can't afford the cost. But it's not just patent trolls that can do this, but large, productive companies too.
How were non-practicing entities dealt with in iceland? They didn't have patents (assuming), so I'm assuming that once you get sufficiently annoying, somebody just kills you?
There would have been no such thing as a non-practicing entity then. Maybe a beggar?
I’m more talking about a guy who abuses the court system by suing everyone all the time to make a living even if he loses a lot
There is someone in the Tale of Sarcastic Halli who has the reputation of killing people and never paying wergeld — Norwegian, semi-independent allied with King Harald. He's exploiting his power and reputation. Halli, who is a skald not a warrior, gets him to pay wergeld by a fraudulent claim that someone he killed was Halli's brother, using a reputational threat.
I think the usual abuse you see, as there, is not false claims but refusal to pay on true claims. But there is a berserk in Egilsaga who claims land and defends the false claim by offering to duel the representative of the owner. The only representative available is the owner's son, a youth. So Egil offers to do it instead and kills the berserk. That is an example of one of the ways of defending the rights of the weak that I think I discuss in the chapter on feud law, observed in the sagas and the Commanche.