As a young student, I liked the idea of patents. Here, it seemed, was a system to properly reward the sort of person I was planning to become. But gradually I came to notice that everybody I admired for their scientific and engineering achievements took a rather dim view of patents. Being practical people, they sometimes took out patents (there was money in it), but without any feeling that the patent system was really the right way of rewarding them; on the contrary, they often pointed out absurdities in it. Richard Feynman, for instance, was the “inventor” listed on the patents of the nuclear airplane, the nuclear reactor, and the nuclear rocket, and treated the whole thing as a joke.

In the area of software in particular, the opinions I’ve encountered have been unanimously against against software patents. One obvious strike against software patents is that with software, both copyright and patents can be used, giving two bites at the intellectual property apple, whereas other things get only one bite: physical devices can only be patented, and writings, video and art can only be copyrighted. Yet the arguments I’ve seen against software patents usually have not focused on that aspect of the issue; instead they have applied equally well to patents in general, even when (as frequently happens) their authors disclaim an attempt to attack patents in general. Those disclaimers, indeed, seem to be a matter mostly of modestly sticking to the area the author knows best; I have yet to encounter one that actually praised patents in other areas. I have often seen an acknowledgement that some sort of intellectual property in physical objects is essential, but never an argument that patents are really the right way to do that.

When considering the patent system, there is a basic question: why can’t the person who invents some sort of tangible thing make money the same way as an innovator who does not produce a tangible thing – that is, why can’t he make money by explaining it to the world, and making money off those explanations via copyright law? Someone who figures out how to do something without using an expensive device may be doing something more praiseworthy than the inventor of that device; likewise for someone who discovers a much better use for an existing device. Yet only the original inventor of a device is rewarded with a patent. The others must content themselves with making money off their explanations, via copyright law. (In theory one can patent new methods for using existing devices, but in practice it’s quite hard to get revenue out of that: one couldn’t sue a device’s manufacturer, but would have to sue its users, which is seldom worthwhile, as usually there are many of them and each wouldn’t yield enough damages to cover the cost of a lawsuit. There’s a minor exception to this is in medicine, where FDA rules make “use patents” meaningful, but even there it’s not easy.)

Still, there are decent answers to the question of why inventors shouldn’t just make money off explaining their inventions. One is that the audience for the explanation of a new device is relatively small; its core consists only of a few engineers who design products, and such a small audience can’t yield much revenue. But in addition to that core there is a wider audience who would appreciate a well written explanation. People who build and maintain products might not absolutely have to understand them, but they do their work much better if they do. Consumers, too, fall victim to unscrupulous marketers much less often if they understand what they are buying. Most consumers will of course just be concerned about whether it works, not how it works, but there’s still a market there. So while there is something to the limited-audience answer, it’s not compelling: the limited audience could just pay higher prices. We already see this today with technical books, which cost much more than novels do.

Another answer, and a stronger one, is that writing and inventing are different skills, and that someone with the skills to invent may be hopeless at explaining his inventions. The degree to which this happens is commonly overstated: people who are commonly regarded as technical geniuses with no verbal skills quite often turn out to not be solid technically either: they are good at producing the right sort of bafflegab to get government grants or to gather power in a corporation, but not good at producing useful results. Still, there is a real art to explaining things, and it is not acquired except by those who spend time and effort acquiring it. So the different-skills answer is a good enough reason for there to be some sort of intellectual property for physical devices.

Yet that answer reveals a great danger: it means the system is intended in large part to protect people who don’t have the explanatory skills to protect themselves. Systems for protecting the powerless are commonly subverted to act in the interests of the supposed protectors; indeed, given human nature, it is quite hard for that not to happen. The most outstanding example of our era was of course Communism, with its “Workers of the world unite” slogan. In reality hardly any of its leaders came from the working classes; Lenin, for instance, was born into the hereditary nobility. The movement is thus more properly regarded as a faction of the upper and middle classes which used that slogan to appeal to workers and which, once in power, oppressed them even more viciously and thoroughly than they were oppressed before.

The patent system, of course, can’t boast of anything like the death toll that Communism can. Indeed, I’d be hard-pressed to point to a single fatality. Its injustices are not the sort that reduce people to below the common level, but ones that stop them from rising above it. Yet it is not egalitarian, either: it rewards brains and knowledge, but chiefly the brains and knowledge of patent lawyers. The Consitution names “inventors” as the people to be rewarded by the patent system, and each patent must have an “inventor” designated. But the beneficiary is not the “inventor”; it’s the “assignee” – the actual owner of the patent, usually the corporation for which the inventor works. In the case of a solitary inventor the assignee may be the inventor himself; but for an inventor to assert his rights in court (either in bringing a patent lawsuit or in defending one) is generally said to cost upwards of a million dollars, an amount of money which few individuals can spare. With such costs, patent lawsuits are brought mostly by large corporations. When individual inventors profit from patents, it normally is by selling them to corporations or by working for a corporation that gives bonuses to employees who produce patents. In either case, though, little money is paid; the patents are valued as additions to the corporation’s patent war chest, rather than being valued as the key enabling ideas that the patent system pretends that they are. In lawsuits they sometimes are valued as key enabling inventions, but that level of money only very seldom gets back to the inventors; instead it goes to the lawyers and the corporations.

Indeed, there is currently a movement to enshrine corporate control over patents, by making life difficult for “patent trolls” – people who bring patent lawsuits without being part of a company that uses the patent in question to make patented devices. Patent lawyers have objected that this is actually the patent system working as it should and for once properly compensating individual inventors who have their inventions stolen by large corporations. In a way they’re right: that is the way the system was designed. But what it shows is not that patent trolls should be rewarded, but that the system was misdesigned from day one. The corporations who are the victims of patent trolling have generally never heard of the patent prior to receiving a demand letter to pay up; rather than stealing the invention, they invented it independently. Quite often, the “invention” is something obvious enough that the patent never should have been granted in the first place, and the trolls are just calculating that the company will find it cheaper to settle the lawsuit than to litigate, especially given the risk of an ignorant jury misdeciding the case.

To further counter patent trolls, there are various ideas to tighten up the issuing of patents, which seems to have become especially lax in recent years. Also, the Supreme Court has just cut down on the degree to which patent lawyers can pick the most favorable jurisdiction to file their suits in. But while such changes are welcome, they don’t go to the heart of the matter, which is that the whole patent system is based on a fundamentally wrong notion about how inventions happen.

That notion is that the important thing in an invention is the Big Idea – that once the inventor has the Big Idea, the rest follows easily from it. When filing a patent, the key part is the section containing “claims”: the Big Ideas that you want a monopoly on. The usual practice is to file many claims. The first claim is phrased quite generally, to try to cover all possible variations of the idea. Then, in case that claim is thrown out as being too broad, subsequent claims gradually narrow it down. It is then up to the patent examiner and the courts as to where in the list of claims they start agreeing that the invention is really patentable. To be patentable, it has to be a new invention, and one that is not “obvious to a person skilled in the art”; and much money is spent arguing over what exactly those terms are to mean in any given case.

There is one important exception to this rule of Big Ideas being patentable: mathematics cannot be patented. What exactly this means in practice – for pretty much any idea can be expressed in some sort of mathematical form – is yet another question to be endlessly litigated. If taken to its logical extension it would exterminate patents entirely, but nothing in politics is ever taken to its logical extension.

Rewarding Big Ideas seems entirely appropriate to most people, since that is how devices are commonly described: by the Big Ideas incorporated into them. “This is a helicopter”; “that is an airplane”; helicopters and airplanes are Big Ideas. But any good, experienced engineer knows that the Big Idea is just the start of the invention process; to really produce a viable product, one has to understand not just the new Big Idea but also the umpteen other Big Ideas it interacts with, some of which are engineering matters but others of which come from other parts of life. And that wide understanding is just the start of a lot of work, in which one carefully explores the interactions of all those ideas until one finds a sweet spot where something can actually be profitably made and sold. This exploration is such a difficult and time-consuming task that thorough success is hardly ever achieved on the first try; the best one can realistically hope for on the first try of a new Big Idea is to produce something that is useful enough to be worth improving. As Linus Torvalds put it,

“It was Edison who said ‘1% inspiration, 99% perspiration’. That may have been true a hundred years ago. These days it’s ‘0.01% inspiration, 99.99% perspiration’, and the inspiration is the easy part. As a project manager, I have never had trouble finding people with crazy ideas. I have trouble finding people who can execute.”

As one of his hundreds of inventions, Leonardo da Vinci sketched out a helicopter. But his drawing didn’t help in the slightest when, centuries later, people developed working helicopters; the main thing that made those possible was the availability of internal combustion engines with a high power-to-weight ratio. Making them controllable was perhaps the biggest challenge after that. Da Vinci’s helicopter drawing had a simple fan, not the complicated hub of a modern-day helicopter which varies the blade angles as they go around, enabling the pilot to steer the helicopter in three dimensions. Still, the Big Idea of “push air down, and you go up” was there in da Vinci’s work. So by modern law, he could have patented the helicopter, and if it were not for patents’ limited lifetime, could be charging royalties on every helicopter built today.

Without controllability, da Vinci wouldn’t have anything saleable, so would be a mere “patent troll”; or he could enter into a patent consortium with whoever worked out controllability and together they could charge everybody else high fees. Further companies might be accepted into the consortium, if they brought their own patents which would contribute to making a good helicopter; or they might spend much time and energy suing each other before agreeing on terms. The history of patents is, indeed, full of patent consortia arising whenever a new sort of device comes along; there was one for the sewing machine, and one for the airplane. For semiconductors, Bell Labs, the invention place of the transistor, liberally licensed its patents (under antitrust pressure from the government). In a similar vein, recently Tesla Motors announced that its patents on electric cars are free for competitors to use, under the condition that they not bring any patent suits against Tesla.

If Big Ideas, as expressed in patents, were really key to innovation, there would be nothing wrong with the people who had the Big Ideas dividing the market up among themselves; but they’re not really key. They’re how people normally talk about innovations, since they’re the part of innovation that most people understand, but the part that deserves rewarding is the part that goes above and beyond what most people understand – the part that takes extraordinary talent and dedication, and which might not happen if not for the prospect of an extraordinary payoff.

It’s not just that the patent system doesn’t reward extraordinary talent and dedication; it is positively hostile to it. Someone with the mind of, say, a Sikorsky, will in the course of solving a large problem make many small innovations – innovations which can easily be patented by lesser minds and used to obstruct him. In practice, this problem is dealt with mostly by trying to ignore it. Large corporations commonly tell their engineers not to worry about other people’s patents when doing their work. If they happen to learn of a patent they should respect it, but they shouldn’t do patent searches to try to see what to avoid. Indeed, if they tried to check the patent status of every little thing they did, besides slowing them down to an impossible degree, it would actually increase the company’s liability, since “wilful” infringement of a patent carries increased damages compared to unknowing infringement. (Patent lawyers, it seems, have not yet gotten the idea that telling your engineers not to worry about patents is in itself something that makes infringement “wilful”. Or maybe they’ve thought of it but they have some decency left.) So the engineers are mostly free to do their work, but the price of that freedom is working for a large corporation which has enough money and enough of a patent war chest to engage in long battles. The rewards which they might have gotten for their efforts in a smaller company are greatly diluted in a large corporation, and their freedom to choose a workplace that treats them well is greatly restricted.

Not everybody ignores patents, of course; there are people try to work around them, notably the people who design royalty-free audio and video codecs. But even they have their limits; as one put it:

“In fact we might as well get used to the fact right now that practically anything we want to do with an audio stream beyond addition probably is patented and I’m bracing to find out that addition is too.”

To the average engineer, who plods along on a rather humdrum task while working for a large corporation, the patent system is not an object of major complaint: he might not know what landmines it has placed to blow up a product he is working on, but there’s a good chance the product won’t get blown up, and if it does that’s the legal department’s problem. It’s the best engineers, who are moving fast in pursuit of a far-reaching idea, and taking responsibility into their own hands, who are at most risk of getting blown up. This is malevolent, yet it’s what is to be expected from a system that is supposed to protect inventors but was designed and is operated by people who thoroughly misunderstand the process of invention.

That aspect of the patent system is absent from the other main sort of intellectual property, copyright. Ever since the invention of printing, people who can write well have had a lot of political power. Today, people who can put together good videos have joined and perhaps surpassed them. Such people, acting on their own initiative, can sway many votes, and politicians know it. The historian Thomas Babington Macaulay, as a member of the House of Commons, gave a speech on copyright that was so influential that more than half a century later Mark Twain still felt the need to argue against it. (Twain found it hard to pick out a flaw in it, and was reduced to railing against it as simple-minded. Which it is, in a way – the same way that the proof of a mathematical theorem proceeds, by steps that are each simple enough to be uncontestable, to establish a truth that is not at all simple. Indeed, it is still worth reading today.)

With copyright law having considerable input from the people who earn their living by it, it is much better designed than patent law. Consider, for instance, what would happen if patent law rather than copyright law were applied to Adam Smith’s book The Wealth Of Nations. Smith is known as the great apostle of free trade; he made arguments which, considered as Big Ideas, were new and non-obvious. Under the patent system, he would have had to embalm those ideas in patentese, as “claims”; then he would have had to pay a patent fee and wait and see if an examiner agreed to their patentability. If so, Smith would then have had the monopoly on arguing for free trade for the next seventeen years; nobody else could have argued for it without paying him a license fee. Or if they did, he could have sued them. After years of legal wrangling about whether or not their writings really were arguing for free trade in the same sense as defined in his patents, and whether it really was a new thing to argue for free trade that way, he might have collected damages.

Yet after seventeen years this extraordinary control over the whole idea of free trade would have abruptly ended, and people could have printed exact copies of The Wealth Of Nations without paying him a penny. For that matter, parts of The Wealth Of Nations that did not involve the claimed Big Ideas could have been copied without his permission even before the seventeen years ended. (There are many such parts; indeed, they are the main thing that make the book interesting to a modern reader. Smith’s arguments about free trade have passed into common currency, and indeed have been improved on. But his remarks on life in his era are another matter. Much has changed since then, but much has not. Farming, for instance, is still looked down on as an occupation, yet it still commonly takes years to learn how to farm well enough to make a living at it, which is not the case for many of the jobs held by people who look down on farmers. Smith’s remarks on the American Revolution – to him, “the present disturbances” – are also quite interesting.)

Of course instead of this mess that patent law would have produced, Smith made his money using copyright law, which is much simpler. You get a copyright automatically and without any wait; there are no fees (except a modest fee if you choose to register the copyright, which is not necessary but increases the damages recoverable), you don’t have to boast about how big your ideas are, and you aren’t subject to the whims of workers at the patent office. (In Switzerland, Einstein was for a period a patent clerk, but most patent examiners are no Einsteins.) It’s simpler for others, too: if you don’t want to be sued for copyright infringement, don’t copy someone else’s work – or copy only small excerpts of it, as permitted by the doctrine of “fair use”. You can, of course, push that doctrine too far and get into trouble. But you can’t get into trouble via copyright law without actually copying, whereas with patents it is customary to be sued by someone who has a patent you never heard of.

For an even stronger example of how miserable patent law would be in copyright law’s traditional domains, consider what sorts of claims one would put in a patent if one were trying to patent the Mona Lisa. “What is claimed is: (1) a painting of a woman with an enigmatic expression…” – but the whole thought is too ridiculous to continue.

This is not to say that all is well in the world of copyrights. Retroactive extensions of the copyright term are simply theft from the public domain: with works that already are in existence, an extension of their copyright term cannot possibly “promote the progress of science and useful arts”, as the Constitution requires, but is a mere payoff to the copyright holders. (In oral argument before the Supreme Court, one of the Justices asked whether this was the point being made; unfortunately Lawrence Lessig explicitly disclaimed it, relying instead on more utopian arguments.) The DMCA’s restrictions on reverse-engineering copyright protection measures are also a sore point. The rule that a work is copyrighted even if the author didn’t bother putting a copyright notice on it also seems like a mistake. Still, on the whole, copyright law is a shining beacon of light compared to patent law.

In copyright law, it is an accepted doctrine that ideas cannot be copyrighted, only their expression. Patent law takes exactly the opposite position: ideas are primary. There is one concession to reality: a patent can be thrown out if it didn’t include a working method of actually implementing the idea. But that is only a minor concession: so long as it includes one working method, it owns all possible working methods that use the idea. While the world of physical devices is somewhat different from the world of writings, art, and video, the two worlds are not so different that their intellectual property laws should be polar opposites of each other.

An illustration of this comes from printed circuit boards (PCBs). These are the fiberglass boards which almost all electronic circuits are built on today; they have copper traces connecting components, holes and solder pads which accept those components, and usually the tersest of labels, such as “R30” or “Q3”, indicating which components go where. Occasionally I’ve seen circuit boards where one label is a copyright notice, but most have none: being physical devices, PCBs aren’t copyrightable. Yet the copyright notice isn’t entirely meaningless: it protects the design file which, on a computer, represents the circuit board. That design file is not a physical device and so is copyrightable. If a competitor were to somehow get hold of the manufacturer’s design file and send it to a PCB manufacturing outfit to be manufactured, that’d be against copyright law. The board itself can be freely copied; it’s just that one can’t use the manufacturer’s design file to do it. So the copyright notice is not 100% ineffectual, just 99.9% ineffectual: it just forces a would-be copier to do what he’d normally do and copy a finished board rather than going through all the trouble of industrial espionage to get the design file.

There is a similar situation in the case of integrated circuit masks. These are much like circuit boards, albeit in a much tinier domain: they are what light shines through to produce features on the microscopic surface of a silicon chip. They too lay out wires that interconnect circuit elements; but instead of just having labels to show which devices go where, IC masks tell how the devices (diodes, transistors, and so forth) are to be made by appropriately juxtaposing different silicon and metal features. Though technically they are quite similar, legally there is a big difference: in the 1980s, chip manufacturers lobbied for and got a third sort of intellectual property just for chip masks. An overview can be found here on copyright.gov; though it is legally distinct from copyright, it operates in much the same way. Its protection expires much sooner, though: after just ten years. Unlike with patents, one seldom hears about this law; it works well enough to not cause controversy.

But even though it’s a good law, it’s a case of special pleading: why should chip makers have their own special law that doesn’t even cover the very similar case of printed circuit boards? And since it’s so similar to copyright law, why shouldn’t it just be copyright law? The term of protection would be much longer with copyright, but is that a problem? If after ten years someone’s chip design is still worth ripping off, why let it be ripped off? And why shouldn’t copyright notices on PCBs be enforceable? Indeed, why not simply extend copyright to physical devices of all sorts, and get rid of the entire patent system?

Now, there is one big change which seems necessary when extending copyright to physical devices, which is to make it impermissible to copyright interfaces. With the traditional subjects of copyright – writings, art, and now video – there is no question of them interfacing to anything other than the human mind; but devices are often plugged togther or bolted together; and to allow a company to own the interfaces by which they are connected together is to give that company undue power. If, say, Ford designs a new car engine, it’s reasonable to give them a copyright that prevents others from just copying the engine or copying pieces of the engine, but it’s not reasonable to prevent anyone else from making spare parts that fit the engine. If you’re making a piston to use as a spare part, you have to make its diameter the same as that of the original; likewise for the diameter of the wrist pin. Those numbers define interfaces, so wouldn’t be copyrightable. But the top of the piston might be a shape that Ford put tens of millions of dollars of fluid dynamics work into; to just let people copy that shape is unfair to Ford. If a piston house wants to make pistons for a Ford engine, let them do their own fluid dynamics work. Once they get good at it, they might even do a better job than Ford did. (It might be argued that the top of the piston “interfaces” with the air-fuel mixture, but that isn’t the sort of interface that would be excluded from copyright protection, just interfaces with other parts of the machine.)

To take another example, one of the purposes patents are put to is to make so-called “tamper-proof” screws, by patenting a new shape of screw head and its corresponding screwdriver bit, and telling customers that nobody else can legally make a screwdriver that can turn those screws. This is a severe abuse of the patent system: there is really nothing “non-obvious” about making a screw head that has a five-lobed version of a Torx head rather than the usual six lobes. In practice such screws are never actually tamper-proof; the proliferation of different screwdriver bits is just a nuisance. It might be a bit hard to get the very latest screwdriver bits, but older “tamper-proof” screwdriver bits can easily be bought in sets of a hundred that contain a wide variety of bits. Would-be tamperers can also tamper with the screw itself, such as by cutting a slot in its head so that it can be turned with a flat-head screwdriver, or just wedging in a flat-head screwdriver that happens to fit well enough to turn the screw. All in all, the “security” added by this proliferation of screw heads is more of a hassle than it’s worth. The rule against copyrighting interfaces would mean that screw heads, being an interface with the screwdriver, would not be copyrightable. (The thread wouldn’t be, either – at least not in machine screws, since it’s an interface with the nut.)

There’s a similar situation with locks and keys: there are patented keyways which lock manufacturers introduce for “key control” purposes. If anyone else were to make blank keys for those locks, they’d be hit with a patent lawsuit. In most cases there is nothing seriously novel about the keyway; it’s just a different shape or in some cases a different little movable part included in the key. And again, there are plenty of ways around it; these days, “forbidden” keys can be 3D printed. (Plastic keys aren’t durable, but a criminal only needs to use them once.) For that matter, if a lock customer buys into the idea that this is a necessary feature, he’ll be replacing his locks every decade or two as patents run out. Lock manufacturers like it when people frequently buy new locks, but it’s not good for customers or for the economy as a whole. If this were really a good security feature, the thing to do would be to make a special law for it and lengthen the term of protection; but really it’s pretty pathetic, so we can let it fall afoul of the rule against copyrighting interfaces. Other manufacturers today just stamp “Do Not Duplicate” on their keys, which lacks the force of patent law but still lets honest locksmiths know that they aren’t to copy those keys.

Now, the above examples should also be uncopyrightable just on account of triviality: copyright’s doctrine of “fair use” covers such things as quoting a paragraph from a book, and copying the diameter of a piston is just as trivial a proposition, if not more so. But there are plenty of more complicated interfaces for which the exclusion on copyrighting interfaces would do good.

An example comes from software, where the lack of this exclusion has already been causing problems. We were recently treated to the spectacle of Oracle vs Google, where this was the main issue: whether the Java application programmer interface (“API”) is copyrightable. The Java programming language was originally designed at Sun Microsystems, who weren’t trying to make money off it via licensing or other fees but rather were trying to start a revolution in computing: applications written in Java, they claimed, could be “write once, run everywhere”. If programmers switched to Java, then their programs would run on Sun machines, not just on Microsoft Windows, which at the time was eating up Sun’s market. To encourage Java’s adoption, large promises were made about being able to freely use it.

The revolution failed – getting people to adopt a new language is never a quick process, and platforms are different enough that reducing the programming environment to the lowest common denominator of all of them results in an environment that isn’t great on any of them – but Java is still a decent language which saw a fair amount of adoption. Sun’s market continued to dwindle, and the company was eventually bought by Oracle, a far more mercenary company, which sued Google over Java. Google had written their own compiler for the Java language, but in order for it to be compatible with standard Java, certain files which defined the interface to the language had to be almost identical to Sun’s versions of the same files (and, according to testimony, were in fact copied). Oracle brought a copyright suit over this, and much legal talent has been exercised in wrangling over whether or not it was “fair use”. An exclusion on copyrighting interfaces would have eliminated all that wrangling, and would have ensured that anyone could write a Java implementation.

Even with interfaces being uncopyrightable, a consequence of copyright being extended to physical objects would be that many things that today have no copy protection would acquire it. There is often a fair bit of work that goes into formulating off-patent over-the-counter things, as witness the long lists of ingredients in, say, shampoos; and that work would be copyrightable. Walmart, at present, has a line of products branded “Equate”: on Walmart shelves, right next to a brand name drugstore product you’ll often find an “Equate” product which is supposedly equal to it but costs less. I don’t know if they are in fact copying the exact formulation of the branded product, but there’s nothing to stop them from doing so, and it seems like there’d be no great harm done if they were stopped from doing so. If a company wanted to make shampoo, they’d have to hire someone who had a clue about formulating shampoo rather than knocking off someone else’s formulation.

It might be hard, admittedly, to prove that a formulation was independently invented rather than being a copy. With books it’s obvious what is copied; likewise with printed circuit boards. But if there are only five chemicals in a shampoo mix, two formulators may by chance end up with nearly the same ingredients in nearly the same proportion. A similar case in existing copyright law was that of programmers writing clone BIOSes for PCs, where the programming task was so simple and directed at such a well defined goal that they were likely to produce code that looked like it was copied even if it was not. To avoid the appearance of copying, they did their work in “clean rooms” from which all knowledge of existing BIOS implementations was banned; the programmers were given just the exact descriptions of what the BIOS had to accomplish, and forbidden from knowing how existing BIOSes accomplished it. Independent invention is a defense in copyright law, unlike in patent law. Even without the clean rooms it’d have been a defense; the clean rooms made it an ironclad defense.

That sort of solution is seldom practical, though: usually the inventor of a device knows a lot about the workings of competitors’ devices, and even picked up a trick or two from them. As in today’s copyright law, there would have to be provision for “fair use” of small parts of a copyrighted work. And as in today’s copyright law, there would still be litigation over the exact boundaries of “fair use”. Even the “novelty” element of patent law would enter into disputes, though in a more practical way than it does in patent law: it’d be easier to prove copying if what was copied was distinctive and new. There would still be work for lawyers; the difference would be that they would be arguing in a more concrete framework: the question would be whether device A was a copy of device B, not the combined and more abstract questions of whether device A was an instance of idea C and whether idea C was really a proper Big Idea.

Precisely what “fair use” should constitute in each of the different industries to which copyright would be extended is a large question, probably too large for any one person to answer in advance. In traditional copyright, this has been addressed by punting the question to the courts, which have made up their own set of rules and enshrined them into precedent. That seems to have worked reasonably well. One still might fear, however, that suddenly dumping a whole lot of new questions from other industries on to the court system might end badly. In any case, the basic principle throughout should be that if someone saved himself a lot of work by copying something rather than redesigning it, that’d be copyright infringement. But how exactly that would translate into rules simple enough to be enforced is another question, and one whose answer would vary from industry to industry.

One industry, though, seems to deserve special comment: the drug industry, which puts billions of dollars into developing new drugs. At present they rely on patent law to give them a temporary monopoly to recoup those costs. But the bread and butter of the drug industry is the patenting of a newly invented molecule. The right to a specific molecule is more akin to the right that copyright confers than it is to the sort of wide-ranging Big Ideas that patent law protects. Indeed, when someone does patent a Big Idea, the industry commonly regards it as illegitimate: why, for instance, should someone who discovers a molecular pathway be granted a monopoly on any and every possible drug that would hit that pathway? To do so takes away much of the incentive for doing the hard work of drug discovery: searching for a molecule that hits that pathway, that doesn’t produce bad side effects by hitting other pathways, that is absorbed well when taken orally, that the liver doesn’t destroy as part of its gatekeeping function, and that is economically manufacturable.

At present, the amount of money that this search costs seems a bit ridiculous. The laboratory work doesn’t consume most of the money; most of it goes into running the controlled trials in humans which prove that the drug is safe and effective. (Normally in those trials the drug company provides the drug for free and pays the physicians conducting them.) Yet in a way those are part of the search: most drug candidates fail their trials. And while one can object to details of the way the FDA controls those trials, it’s basically a sane system: even if the government didn’t require it, a double-blind controlled trial really is the gold standard of medical information. Even if the overall expense of developing a drug were reduced tenfold by letting eager patients bear some of the costs of trials, there still would have to be some sort of intellectual property protection to pay for its development; the sorts of lab equipment and chemicals that are used in cutting-edge drug development don’t come cheap.

But although, as in other areas, copyright law seems more fitting than patent law, here even copyright law seems lacking. The space of chemical compounds is both more exhaustible and more inscrutable than are the design spaces in which other engineers operate. If you add a bolt to a machine, any decent engineer will be able to tell you what it does. If you add a methyl group to a drug molecule, the exact consequences cannot be forseen. If you copy half a book, it’s a clear case of copyright infringement; there’s no way it could have happened by accident. But copying half a molecule can easily happen by accident: most drugs on the market are of the “small molecule” variety, and small molecules are more like sentences than books, in terms of their information content. Two people may easily come up with sentences which are 50% the same, yet neither of which was copied from the other and which mean different things; and the same goes for small molecules. Yet while copying a sentence is generally fair use, copying a molecule shouldn’t be, since what exactly it will do if used as a drug is quite hard to determine. The principles of chemistry tell you some things, but not enough. Computer modeling of chemical interactions is in its infancy; even the simplest interactions cannot be predicted precisely. Though some tentative predictions can be made, the only real way to find out is to test with wet chemistry and then with experiments on living organisms. Indeed, one of the recent areas of advance in medicinal chemistry has been in inventing various ways to automate testing of thousands of variations of molecules.

Larger molecules (“biologics”) are increasingly used these days as drugs, but the amount of information needed to completely describe them is again on the order of a sentence rather than a book, though the “sentence” does have to be written in a different language. Biologics are usually antibodies or other proteins, and a protein can be described much more briefly in terms of its component amino acids than in terms of its component atoms. An antibody can be described even more briefly by describing only its variable region.

Under the present system, the space of available molecules is getting used up. If each patent just covered one single molecule, the rate of use would be negligible, but drug patents tend to claim not just one molecule but long lists of variations on it. And a list naming three hundred variations usually doesn’t just cover three hundred molecules: some variations are on one part of the molecule, some on others, and the patent is normally on all combinations of those variations. If three different parts of the molecule each have a hundred variations, the number of combinations is a hundred times a hundred times a hundred: a million. And from the point of view of drug company researchers, anything that has ever been covered by any patent is forever out of bounds as a drug candidate. Even after the patent expires, the molecules named in it are forever unpatentable due to lack of “novelty” and so will not repay the testing required to sell any of them.

This is a long-term worry; the situation is not to where people in the industry moan and groan about exhausting the space – nowhere near it. But they do sometimes notice that some of their drug candidates have to be thrown out as unpatentable, and the situation will only get worse with time.

A rather miserable attempt has already been made to remedy the problem of drugs not getting tested because they are unpatentable, via a law that gives companies patent-like exclusivity on old drugs in return for conducting the currently-mandated sorts of testing on them – testing that was never originally conducted because laws were looser back when the drugs were originally approved. The new law was immediately taken advantage of by vultures in the drug industry, who chose the old drugs whose effects were best known from long experience, put them through testing (which of course confirmed that experience), then used their new market exclusivity to jack up the prices by factors of hundreds, producing public outrage. But the idea wouldn’t be bad if applied to new drugs that are unpatentable because they’re listed on an expired patent. Such drugs might easily be improvements over the drug that was tested and sold under that patent. (I phrase this in terms of the present system because it could be done under the present system; one wouldn’t have to wait for the large sorts of changes that this essay is mainly about.)

In any case, it is not surprising that our intellectual property laws are poorly designed for property in chemistry, since they date from before anyone even knew what atoms and molecules are. Changing the system from patent to copyright would provide an opportunity for chemists to help decide how to protect the chemical neighborhood of a copyrighted drug. One way to do it would be for chemical copyrights to be sold on a per-molecule basis. Today’s copyrights don’t require any fee to be paid, but for copyrights on chemicals there’d have to be some sort of fee, or companies would just copyright everything, however many terabytes of disk space their applications involved. The fee should be larger for smaller molecules, since there are fewer of them, making each more valuable. This would be similar to today’s patent protection for drugs, in that companies would have to choose which nearby molecules to protect, but it would differ in that today a single patent fee covers all the molecules listed in the patent, which is however many the patent examiner and the courts will let them get away with, while here they’d have to pay per molecule, which would discourage overclaiming in a much simpler and more direct way.

When adjacent molecules were reserved in this way, they would then be released at the end of the copyright period, becoming re-copyrightable. An auction might be held, with proceeds going to the government, if more than one buyer wanted to immediately snap them up. This would be done only for molecules that remained unused and thus whose real potential was as yet undiscovered; drugs that had been approved for sale would enter the public domain after their copyrights expired. Any neighboring molecules that a drug company wanted to copyright and sell would thus be competing against generic versions of the old molecule, so they would have to be significant improvements to make any real profit.

Another possibility would be to put some of the best chemists to work formulating an exact definition of how close one molecule needed to be to another to constitute copyright infringement. They would consider not just the simple number of substitutions necessary to transform the one molecule into the other but also what it did to the molecule’s chemical properties: substituting a highly reactive group for an unreactive one, for instance, is more of a change than switching between two unreactive ones – though there are many sorts of reactivity; the debates over exactly what “similarity” meant would get deep into the theory of chemistry. But the result would be a computer program that anyone could run, which would ask for two molecules to compare and would output a number indicating the distance between the two. (It would not decide “similar enough” as a yes-or-no question; though bright-line rules are beloved by courts, they are against nature. It would be impossible to get experts in chemistry to genuinely agree on a yes-or-no rule; instead they would argue forever – or, more realistically, until politicians got impatient and forced them into false agreement. Even arguments over a numerical rule would take quite a while, but at least they could be settled by compromise.)

The law, in turn, might set a hard threshold on the resulting distance, where exceeding the threshold meant you were out of danger from copyright claims; or a gentler threshold could be used, so that one molecule could be judged a (say) 65% copy of another, with 65% of its royalties thus to be paid to the owner of the original molecule – though figuring out which parts of the income counted as “royalties” would be a puzzle, as drug companies usually make drugs themselves rather than licensing out the rights to do so.

Realistically, though, government committees to decide on big questions like this question of chemical distance never produce an optimal result, and often have produced rather miserable results. So it would probably be better to use the distance formula as an aid to the system, rather than the sole criterion: say, have the system be that when a drug was copyrighted some neighboring molecules were reserved automatically, more distant ones could be reserved with payment, and still others were too distant to reserve.

In any case, it’s not just in engineering that the Big Idea is only the start of the process; that applies to politics, too. Replacing patents with copyright would be a big change, even if done as gently as possible (which would mean not suddenly axing patents, but just stopping the issuance of new ones and letting existing ones naturally expire). Actually achieving it will be a lot harder than just suggesting it, and almost inevitably a lot dirtier. Politics is proverbially the art of compromise; a key is to have an idea simple enough that people know when you’re compromising it and by how much, so that compromise doesn’t end in hopeless corruption. After an idea has been damaged by its friends’ errors and befouled by the mud its enemies have thrown at it, if it’s simple enough it can be cleaned off, refurbished, and sent back into battle. The ideas here seem simple enough, with the exception of the suggestions on property rights for chemicals, which would be breaking entirely new ground and thus are tentative and exploratory.

Update (Sept 15, 2020): according to this article, “It is a basic provision of the law of inventions in China that the party who came up with the “idea” does not own the invention; the party that takes the idea through to practical implementation does.” Though lacking in detail, this is something like what I argue for above. So there might be a legal aspect to why the Chinese have been eating our lunch in manufacturing.