Eric Maskin on Patents

From a letter to the NYTimes:

… in the software industry, progress is highly sequential: progress is typically made through a large number of small steps, each building on the previous ones. If one of those steps is patentable, then the patent holder can effectively block (or at least slow down) subsequent progress by setting high license fees.

Moreover, like any other monopolist, it has the incentive to set such fees.

Thus, in an industry with highly sequential innovation, it may be better for society to scrap patents altogether than try to tighten them.

Cambridge, Mass., Oct. 8, 2012

The writer, a professor of economics at Harvard, is a 2007 Nobel laureate in economics.


He obviously doesn't know what he's talking about.

It's an interesting question though. Should there be a difference between taking someone's idea (as in really doing it, not just tricking a court...ahem...Apple) to sell it versus building on it to create a further advancement.

The world's largest business software company is completely content to be headquartered in a country that does not allow software patents.

Because they infringe on them?

Sure, if one accepts American law as definitive.

SAP, as a German/EU company doesn't, and its world wide sales reflect that perspective. Much like American companies continue to dismiss EU laws concerning privacy - well, at least until the EU fines them, that is (in the case of SAP, it is pretty clear that patent trolls have a hard path to follow, SAP having most of the prior art involving ERP software due to ti inventing much of it - along with numerous other German software companies that compete with SAP). However, try as hard as they do, American companies still can't convince the world to go along with their perspective - the failure of ACTA and the rise of the Pirate Party in Germany being a good example of democratic pushback.

Wishing for something to be true does not make it so. Especially when it is odiously self-serving. Just ask Redhat - or better, ask Alan Cox, who has told me he will never travel to the U.S. due to its ludicrous laws.

What country would that be? Germany? I think you are wrong, if so. Furthermore, yes, big companies often do not like patents because they can be disruptive to entrenched interests.

Germany doe not allow pure software or business process patents - Amazon's one click patent has no standing in Germany (or the EU, for that matter).

Wasn't obvious to me.....

What's also interesting (in the world of bad patents and especially "obvious" patents) is that if you're you can profit from a patent, there are actually incentives to price the licensing higher rather than lower. If you had a really good way to develop a product you might not need the monopoly to be successful.

The problem with software patents is that, unlike drugs or lots of other products covered by patents, the end software product incorporates a lot of elements that are subject to patent, whereas a drug or other product will typically have been developed with other patented devices but will not itself contain lots of patentable elements. Thus, while a drug discoverer will pay for ideas used in the discovery of drugs through the purchase of equipment used in the drug discovery process, it will not be forced to pay a licensing fee for each pill it sells. A software producer on the other hand must actually incorporate many patentable elements in its final product to make that product the best product it can be. The same problem that arises in software also arises in lots of complex hardware, which will often incorporate a lot of different patentable elements. The more patentable elements a final product has, the more difficult it will be to bring that product to market. As products have become more complex, it has become harder to navigate the patent system to bring them to market.

'the end software product incorporates a lot of elements that are subject to patent'
Well, in the U.S. - other countries do not allow software patents at all.

What countries are you referring to? The EPO may say that, but the truth is much more shades of gray. I would say that software patents do get granted by the EPO. There is an inherent problem in distinguishing software from hardware. Frankly we might well be better off without software patents, but they are harder to define than you might imagine.

More generally, idea patents are the problem (including, e.g., patents on business methods). They differ from traditional patents in that the inventions they describe to not involve the taming of natural laws and, therefore, typically don't require experiments. That's a pretty clear definition as far as patent law goes.

The ecomomics of this are a bit different than the post suggests.

Think of it this way: Let's say I have a blocking patent that prevents you from using your patent, a patent which either improves my patent, or one which creates an entirely new space for my patent to be exercized.

Now, you can think of this as a bargaining game betweeen two bilateral monopolists (hint: indeterminate solution, with the result being a bargaining process) or a cooperative equilibrium (both parties form a joint venture, where even the original blocking holder invests in the improvement firm, or takes back an improvement license).

If you want to learn more about the real world--cross licensing, joint ventures, exclusive and non-exclusive licenses, and contingent royalty rates--go to law school or look up a book on licensing economics.

I am disappointed that his blog takes such a simple view of patents and makes patents and transaction economics a bogeyman. If you want to study the real world, look up some books by Bill Kerr, a friend and economist I've used in patent antitrust litigation. He's written extensively on patent valuation, licensing mechanisms, etc, and even published a book used in law schools and by practitioners on these subjects. Here's a link:

Apple just pantsed the courts.

We don't have a simple view of patents, we have a simple view of government.

These are all expensive workarounds to the problems engendered by the patent system in its current form. Cross-licenses often take 1-2 years to complete if a match can be made at all.

You are not confuting Maskin. You are hinting you can price patents, find a workaround. Maskin is saying that this come with a cost. Do you think it is free for a small software house to bargain a non-exclusive license with Apple to improve their products? It is not.
If the transaction for a new program involves 10 firms, you are going to have 45 different couples of firms that have to agree on the conditions of use. Is it impossible? It is not free.

Is the system encouraging innovation overall? That is an empirical question I would like to see addressed. But theoretically, your "it is too simple" confutation does not dent any important part of the argument.

Obviously it was meant for Bill

Guys, think for yourself a minute.

First, because something is patented doesn't mean it has market power. There are substitutes, competing technology, etc.
Second, the issue of a patent with market power is the equivalent issue of a firm with market power on a product deciding to sell, or not sell, to someone who wants to distribute that product. It's the same economic issue. The firm can decide it doesn't want to sell its engine, for example, to another car manufacturer, even though, if it did, there would be a better car as a result; or, it could sell the engine, and negotiate a price for it that capture most, some, a little, or all of the improvement--depends on bargaining, alternative uses, etc.

Third, for the person who says--gee, golly, there are transaction costs and valuation issues in patents--step back and ask yourself: all market transactions have transaction costs--patents are no different, as you saw with the negotiation of the engine example above.

Bill, aren't you a lawyer? If so, there's bound to be some conflict of interest here.

Are there any lawyers that hate patents?

One man's transaction costs is another man's fees ;)

If we wanted to reduce transaction costs, the patent system could embed a licensing cost for each patent to remove the vagaries of the court system. Download your licenses off iTunes. And if the patent owner must sell then that eliminates the monopoly stonewalling. The court becomes the mechanism for punitive damages for deadbeats.

Rahul, Ask yourself this question: If lawyers deal with contracts (of a non-IP nature-=ie, buying and selling things) is that any different than buying, licensing or selling. If your argument is that lawyers are involved in transactions involving property, and therefore have conflict, do you think you might be going to far. Is that really analysis, or baiting for want of having a substantive argument to make. You can address the merits.

An argument from an interested party ought to be viewed with some degree of skepticism. Doesn't mean that a lawyer's argument is axiomatically bogus; just that one ought to be careful about separating his professional interests from those of society at large.

That's all I am saying. What's wrong with that?

Would you believe a car salesman trying to convince you that people ought to buy cars more often?

Rahul, I represent an occaisional plaintiff, but much, much more often, represent defendants, or, as I call them, the wrongfully accused. Your bias theory would go the other way if you primarily represent defendants. I should be railing against the system under your theory.

Just as when I deal with the government and think they are making a mistake in investigating a client or transaction I might be tempted to be anti-government, but I'm not. I can see the benefit of enforcement, and believe that, over time, government and judges do or try to do the right thing. Or, perhaps I'm just lucky when they do.


Your bias theory would go the other way if you primarily represent defendants. I should be railing against the system under your theory.

Not at all. My theory says you'd still like patents. The more pervasive a patent system the more defendants to defend! To clarify the point, in an extreme, no-patents world you'd have no cases at all, no matter if you are a defense lawyer or not.

It's a win-win for lawyers. That's precisely the point.

Bill, of course you are not an economist nor a software developer. Software patents are not "any market transaction": they are monopolistic transaction, and any of them is a loss for the total welfare. There are many occasion you could want to avoid using a patent to remove competition. This is optimizing for the owner but not for society.

In heavy manufacturing, where you have a few (big) actors, big projects and high R&D spending, those transactions cost could well be worth the benefits of IP. In software a lot of small actors can be simply removed from the game to avoid entrants and competition. If you have ever tried to start a software house with two friends and you find yourself (without knowing it) infringing an Apple patent you will notice how they are not "any market transaction".

By the way, Hayek (not exactly an enemy of the market) was arguing for a reduction of IP in the 40s on a very similar ground.

"The problem of the prevention of monopoly and the preservation of competition is raised much more acutely in certain other fields to which the concept of property has been extended only in recent times. I am thinking here of the extension of the concept of property to such rights and privileges as patents for inventions, copyright, trade-marks, and the like. It seems to me beyond doubt that in these fields a slavish application of the concept of property as it has been developed for material things has done a great deal to foster the growth of monopoly and that here drastic reforms may be required if competition is to be made to work. In the field of industrial patents in particular we shall have seriously to examine whether the award of a monopoly privilege is really the most appropriate and effective form of reward for the kind of risk-bearing which investment in scientific research involves."

I think looking at the situation now he would write a different book on acquired rental positions and enemies of the market.

PS: I am really starting to believe you are a lawyer and, as it is known, a lawyer opinion does not count.

Bill won this thread. Your anti-patent bigotry is obvious, as is Alex T's. TC has a better view of IP--it can come in handy, and as the Solow equation says, technology innovation is the only way GDP grows in the long run. Encouraging patents rather than trade secrets (which is the opposite of patents) is the way to go. As for "Big Pharma only needs patents", this is a lie. As for "small innovators will be sued by Big Software if software patents exist" this is another lie. With patents, David can level the playing field against Goliath. In fact, Big Software fears David much more than the other way around--if David is not a patent troll. Sure, we can make patent trolls go away (for example, adopt the English Rule, which means the losing plaintiff pays for the defendant's legal fees), but let's not throw out the baby with the bathwater and abolish patents like most economists wish. Hayek and Maskin included.

So, you missed the part where a ton of people consider patents anti-innovation, then.

Take Bill's comment about finding workarounds for example. This doesn't work depending how how incompetent the court is. If some discovers, say , a gene (or large buttons on phone screens that are simply getting bigger as a matter of obvious evolution) and then patents it, there is no workaround or substitute. They simply can squat on this until someone pays them.

Oh, and it just occurred to me that you and Bill are the ones with the simplistic "patents good!" view.

Alex posted something specifically about software patents. It is not some novel concept that government should be extremely careful granting property rights or with a patent system that it could either under/over or wrongly do it.

Ray, you are discussing a different thing. In particular you are attacking a scarecrow. You do not have to convince us that "technology innovation is the only way GDP grows". Did someone say the opposite?

The argument you have to confute is that IP have both costs and benefits toward technology innovation. In these cases a proper economist go to the empiric analysis, not to ideology. A different way to say the same is that both a regime without IP and a regime with everything covered by IP are not optimal. There is therefore an optimal point, in the middle, with respect to technology growth. Looking at the international debate and at some empirical sketches I think that one side of the argument is over represented with respect to its real weight. You do not think so, and it is perfectly fine. However, since it is easy to see reasons for a less then optimal policy looking at rent seeking and conservation of monopoly power, there is an obvious need to do some empirical research.

By the way, shouting "anti-patent bigotry" when someone asks for empirical validation is another reason to search.

"technology innovation is the only way GDP grows in the long run"

One can accept this without concluding that software patents are necessary. Google is largely built on open-source software. They have patents on their search algorithms because they can get them, but Sergey and Larry would've built Google anyway.

Bill is a lawyer, as am I (a patent lawyer).

I guess I am not sure what your point is. It's property, it can be bought and sold and licensed, yes that's true. That doesn't mean it's a good thing.

Actually, in the software world, there are often not any substitutes. Software patents tend to be broad patents on entire classes of ideas, not specific implementation patents. All of which means that the first company to patent an idea, even if they don't have a working implementation, or the first company to patent an idea with "on a computer" or "on a cell phone" added in, will have coverage over every single implementation.

To go with your car analogy, a software patent is the equivalent of patenting "an automobile with an internal combustion engine". Then the firm goes around suing every auto manufacturer, regardless of the engine used, whether or not the patenting firm has ever actually created an engine.

This is, obviously (at least to software engineers), in direct contradiction to the original intent of the patent system, which was all about protecting specific, working implementations.

Unfortunately, software patents in the US are controlled by the Federal Circuit Court, which has so far resisted all attempts the Supreme Court has made to rein them in. The patent office itself is strongly incentivized to permit everything and let the courts sort it out. The courts, in a seriously expensive disconnect, presume that the PTO has actually done the work to validate patents. Congress, of course, is completely ineffectual even where it isn't captured.

Can't speak to other industries, but the software industry is having huge issues thanks to the messed up US patent system.

"in [almost any] industry, progress is highly sequential: progress is typically made through a large number of small steps, each building on the previous ones. If one of those steps is patentable, then the patent holder can effectively block (or at least slow down) subsequent progress by setting high license fees."

People like to use pharma as an example where there are no patent thickets. This is no longer the case. Take a look at all the biologic anti-rhuematic drugs (Embrel, Remicade, etc.). These drugs really are miracle drugs that took decades to develop, and the IP around each of these is layers deep, and without any one of the developments, there is no drug. Don't even get me started on genetic testing and drug interaction. This is very different than a world where one easily synthesized chemical compound is administered to fight a disease.

In a world where information sharing is the norm, determining the allocation of economic rewards on ideas and invention gets a lot more difficult.

There was a time when voluntary disclosure encouraged by patents was of large social utility in expanding the pool of knowledge.

This utility is fast declining. I don't need a company to disclose recipes; a GC-MS will do most of that for me. Reverse engineering has become so good, patent disclosures (not that they are ever forthright) are almost superfluous.

The patented technologies in this area have nothing to do with recipes, and many relate to processes that you can not reverse engineer.

The basic process is that you take a cancer cell, remove the nucleus, and reprogram it so that it does two things: replicate quickly and spit out monoclonal antibodies that attach to a certain protien (Tumer Necrosis Factor Alpha), which in turn interrupts the inflamation process for many people who suffer from Rheumatic disease.

There are at least 5 earth shattering scientific breakthroughs involved in "discovering" these drugs. Without disclosure and cross-pollination, which likely would never happen without patent protection, these drugs are not around today.

There is abuse of the system, and for every example of benefit, there are 3 examples of rent seeking. However, the protection that is offered to people does encourage disclosure and collaboration (beyond simply patent disclosure). Throwing the system away is not the answer, in my opinion.

Of course transaction costs are the bad part of things.

Also, you can never create a property right for everything, so at some point when you have created patent power for almost everything, who gets left out?

The net benefit of intellectual monopoly protection depends on the extra innovation from monopoly rewards, minus the costs of enforcement, including the costs of negotiation and substitute testing (plus substitute discoveries often adding benefits).

The "info revolution" is partly about how the costs of enforcement of intellectual monopoly rights now exceeds the benefits, so there is a net social loss from the patent system, as compared to a non-patent system.

In pharma, especially, a series of prizes (including significant tax credits) could generate similar, and possibly greater innovation. In SW, open source is already viable, and would become more so.

There is no social need for more entertainment subsidies for either DVDs, books, nor sw games. Advertisement and fame creates enough cash and incentive, with no enforcement costs.

As an example of a patent's ability to stifle progress, consider instant cameras in the US in the 1980's. Polaroid's was fairly good. Kodak came out with a much better one (the colors were brighter and truer and never faded, even after decades). Polaroid attacked Kodak for patent infringement, and that was the end of instant cameras with high-quality color. I'm not saying that Kodak didn't infringe any patents; I have not studied that. I'm merely pointing out a real-world example of a patent that killed progress in an area.

When Maskin writes "sequential" he really means "incremental". Other than that, I'm very happy to see an IT outsider who understands the damaging effect of idea patents on the software-development process.

No patents! Everything should be trade secrets! I don't see how failing to disclose the small steps necessary for anyone to innovate could POSSIBLY fail to help innovate!

Dupont does not patent all kinds of things - it gives them a much longer reash in the market.

'US chemicals giant DuPont has won a 20-year global ban on a Kevlar rival product made by South Korea-based Kolon industriesafter a court found Kolon had stolen its trade secrets.

“The trial court ordered Kolon to not produce, market or sell any para-aramid fiber products, worldwide, for 20 years,” Thomas Sager, DuPont’s general counsel, said in a statement Friday.

“It also permanently enjoined Kolon from using any of the trade secrets it stole from DuPont,” he added.

The ruling, issued late Thursday by a US federal court in Richmond, Virginia, followed Kolon’s trial loss in September 2011 when a jury found it had stolen DuPont’s trade secrets on the manufacture of Kevlar aramid fiber.

Kolon, which produces and sells Heracron aramid fiber, has been ordered to pay more than $920 million in damages and interest.'

If Dupont had bothered with patenting, then the verdict would have been different -

'After lengthy analysis, the court held that the eBay standard, which requires irreparable harm and unavailability of an adequate remedy of law for an injunction in patent cases, does not apply for a violation of the Virginia Uniform Trade Secrets Act (VUTSA). The court instead applied Virginia state law, holding that because DuPont had proved a violation of the VUTSA, it need not prove irreparable harm or the lack of an adequate remedy at law. Since nearly all states have adopted some version of the Uniform Trade Secrets Act, DuPont will be useful to trade secret owners outside of Virginia and across most of the United States.'

A side note - the Korean company will almost certainly not stop making its product, since a sweeping world wide injunction from a Commonwealth of Virginia judge will have little to no effect on countries not subject to American law - which, when one thinks about it, is all of them. It is a reasonable bet that the Korean company will continue to be able to sell its product to the Korean Defense Ministry, for example. Much the same way that American companies will continue to sell wine they produce as 'champagne' within he U.S., though it is actually illegally labelled according to international trade laws.

A problem I have with this area is that it all sounds good. Yes, yes I think it's absurd that Apple tries to claim the rectangle is an 'innovation' they all sounds like it makes sense. Big innovations can't happen because every time you take a few small steps, out of the woodwork comes someone who put a patent on every comma and period making it impossible for you to move without paying everyone under the sun off....but....

Why, though, does it seem like there's no shortage of software innovation. I mean look at 2008 versus 20012. Of all the products to claim innovation has been inhibited, you really think cell phones stand out as a clear example?

So when the real world doesn't seem to square with the model, maybe the model might be wrong. Perhaps annoying patent claims might be spurring innovation on the macro level. Consider the humble Blackberry with its built in keyboard. Perhaps Apple leaped over that to the keyboard that appears on the touch screen only when needed and otherwise leaves you with more screen space *because* of concerns that if they tried to only tweak the built in keyboard idea they would be hampered by patent suits. So instead trying to get marginally closer to the peak to the local mountain, they took a shot at a totally different mountain and discovered it dwarfed what passed for a high summit in 2008.

Maybe the overreaching patent law has had the bonus of preventing a 'lockin' of one type of technology where all the big players are trying to just make a slightly different hamburger rather than doing something really outside the box.

Why not grant perpetual patents then? Should ensure even more diverse hamburgers!

Why not? Your priors are showing again Rahul. What do you have against perpetual bonds? Do you think they are a bad idea? Yes bonds. Think about it. Perpetual patents would also not necessarily be a bad idea. It would encourage innovation, at a small cost to society, but with tremendous gains. Think of this thought experiment, you anti-patent bigots (reading this thread): for pioneering inventions, a perpetual patent is granted. For stuff like safe nuclear fusion, extending life for 100s of years, cure for the common cold, flying self-steering cars. Do you think that these inventions would have "inevitably been invented" *without* the perpetual patent? If so, "perpetual patents are bad". But if these inventions would not have been invented "but for" the perpetual patent, then these patents are good. Another mistake you make (among many): assuming that somebody who comes up with a pioneering invention is 'blocking' innovation. Wrong. Before that person came up with the pioneering invention, nobody knew it existed, so again, unless you are from the 'all innovations are inevitable' school of thought, and that's a 'prior' assumption of yours, no patent ever blocks since there's always a substitute, even if that substitute is doing without the invention behind the patent. No I'm not a patent lawyer but have worked with them.

For stuff like safe nuclear fusion, extending life for 100s of years, cure for the common cold, flying self-steering cars. Do you think that these inventions would have “inevitably been invented” *without* the perpetual patent?

Where we don't have reality we have to fall back on models and do our best. We don't have a history of perpetual patents (although I suppose one could argue that Sony Bono's copyright extension act to life of the author plus 75 years or whatnot. Essentially this means Gene Rodenberry enjoyed the knowledge that his concept of 'Mr. Spock' would be of economic value to him for his entire life and likely be able to support his great grandchildren when they are around 75 or so as well. Would Mr. Spock have been better written had Gene had the benefit of extended copyright laws back in the 60's and 70's? Would it have been even more well written had he enjoyed a perpetual copyright that could theoretically have entitled his descendants 1,000 years from now to royality checks from Star Trek reruns?

I think the nature of present value calculation reduces the benefits of the 'far out years' to nearly zero making the added incentive of a perpetual patent marginal at best. It does, however, give us the danger of real innovation inhibition. Right now if a silly patent does happen to get recognized in law, at the end of the day it will eventually expire opening up the market. A really bad idea, though, in a perpetual patent world, would never go away (such as Apple 'owning' all rectangles).

Remeber the time when Blackberry was the only device company that understood how to do the push email service? Everyone else had to rely on pop-mail pulldown, which eats up batteries, is not as responsive, and the constant checking is a drain on bandwidth.

Guess what, now everyone is focused on the touch-screen versus the keyboard. But where would smart phones be today if Blackberry hadn't perfected email push?

Blackberry is getting crushed in the marketplace. However, it still has some juicy patents in its portfolio related to sync and email push. And these are software patents. And they are sequential to the process of innovation. At some point, the value of enforcing these patents is going to be more valueable to RIM shareholders than the protective value that they provide against Apple, etc.

How is Apple going to feel when it is on the wrong side of the copying argument? Is Blackberry wrong for enforcing its patents related to a market that it generally helped create? Should society completely ignore the value the Blackberry brought to market for perfecting email push?

There are no easy, broad-brush answers here.

The possibility of applying for a software patent also has a positive effect on which inventions and innovations are made public. Here’s a little anecdata from my career:

Back in the 80s and early 90s when software patents were first becoming part of the scene for high-tech companies, I was the R&D department for a smallish software company (sales were a few 10s of millions per year). Before software patents were on our radar, *everything* I did was protected as a trade secret. Management’s answer when I wanted to publish was always “No! That’s a trade secret.” We never published anything.

After software patents arrived, that changed completely. Either (a) we applied for a patent and then published (not too frequently, as the cost of pursuing a patent looked high to our management), or (b) we published quickly to establish prior art before someone else could patent our innovation out from under us. Either way, the entire output of our (small) R&D department went from completely hidden to completely open.

Many folks I know who worked *in* *R&D* elsewhere in the software industry tell me they saw a similar pattern at the time. A significant increase in publication of results and inventions that would otherwise have been held as trade secrets, plus a modest increase in the number of patent applications.

It’s only anecdata, but it was uniform across the industrial research groups I had contact with. Other portions of the software industry may have seen other effects. Your mileage may vary. Insert generic disclaimers here.

What piece of software cannot be reverse engineered today? The utility of voluntary disclosure is not as large as it seems.

Also consider the fact that a large chunk of the core work on Unix, C etc. was done in the pre-SW-patent area and was yet widely disseminated. Xerox, Bell Labs etc. were publishing prolifically far before Software patents were on the scene.

I'm not doubting that the quantum of publication increased; I'm merely doubting its social utility.

Comments for this post are closed