The Patent — The Externality, Plate ix

The recipe can cross the boundary without leaving its first kitchen. A second baker who learns it does not diminish the first baker's knowledge, though he still needs flour, an oven, and the skill to use both. This is the strange economy of an idea: its use can be shared without dividing the idea itself. Whether that use is permitted is another question. A secret can be withheld; a legal right can restrict what others may do with disclosed knowledge. Neither changes the fact that two people could use the same instruction at once. The distinction between nonrivalry and exclusion matters because the patent works on the second property. It draws an enforceable boundary around certain uses of an invention, seeking to obtain through controlled access a reward that the idea's physical nature would not reliably collect.

Finding the recipe can be costly even when copying it is cheap. There may be failed experiments before a useful result, and the investigator cannot know the result in advance without already possessing part of what the inquiry seeks. Arrow placed uncertainty and incomplete appropriability at the center of this problem. He also identified the difficulty in selling information whose value a purchaser cannot judge until it has been disclosed: learning enough to value the offer may convey much of what was for sale. These difficulties can separate the return an inventor receives from the benefit the invention creates. They do not imply that invention always stops without a patent. A first mover may sell a useful product before rivals arrive; a researcher may be paid by an institution; a discoverer may value the discovery itself. The question is which worthwhile inquiries receive adequate support under the arrangements available to them.¹

1. Kenneth J. Arrow, “Economic Welfare and the Allocation of Resources for Invention” (1962), pp. 614–619, on information, uncertainty, and appropriability. The final examples distinguish possible sources of reward; they do not claim that any one supplies sufficient incentives in every industry.

A simple invention shows why rewarding discovery and encouraging use can pull in different directions. Suppose a product, once invented, can be made at a constant cost of two per unit, and buyers' marginal willingness to pay is ten minus quantity. Competitive pricing at two would yield eight units and a total surplus of thirty-two before the cost of discovery. An owner able to set a single monopoly price instead chooses four units at a price of six. Operating profit is sixteen; consumers receive eight; together they receive twenty-four. If the invention costs twelve to obtain with certainty, that monopoly profit can cover its cost and leave four. Under the additional assumption that competitive copying eliminates every source of revenue above production cost, the inventor cannot recover the twelve. Exclusivity can therefore finance this invention while leaving eight of potential surplus unrealized once it exists. The omitted units are useful, but the permission to restrict them may have helped bring the product into being.²

2. An original one-period example: p = 10 − q and marginal cost = 2. Monopoly maximizes (8 − q)q at q = 4; profit = 16 and consumer surplus = 8. Efficient q = 8 yields surplus = 32. After a fixed invention cost of 12, total surplus is respectively 12 and 20. There is no uncertainty, price discrimination, alternative funding, or later period; the example does not identify an optimal patent term.

⁂

The comparison depends upon the moment at which it is made. Once the twelve has been irretrievably spent, excluding a buyer willing to cover production cost sacrifices a possible gain. Before it has been spent, the prospect of receipts can determine whether the product will exist at all. A rule chosen in advance must face both moments. Giving more power to exclude may increase an expected reward, but the useful question is whether that increase induces enough additional invention to justify its effects on use. Existing discoveries do not become additional merely because their owners receive more. Nor is every protected invention a monopoly over a whole market: close substitutes may leave its owner little room to raise a price. Duration, the range of uses controlled, and the alternatives available to buyers therefore concern different parts of the incentive. The two dates alone establish no uniform answer to how much protection a discovery needs.³

3. The comparison is between incentives before discovery and allocation after it. The numerical example deliberately grants monopoly pricing power; its results cannot be assigned to every patent merely from the existence of a right to exclude. No claim about a jurisdiction's current legal term or standard of patentability is made here.

Ideas also enter the making of further ideas. The first invention may be valuable partly because someone else can improve it, and the second inventor may need access before investing in that improvement. Green and Scotchmer studied how the profits from such a sequence are divided. Joint profitability is insufficient if the division leaves one indispensable stage unable to cover its cost. Licensing can connect the stages, but the time of the agreement matters: a bargain made before the later inventor invests faces a different situation from one made after that investment is sunk. Their conclusions depend upon the contracts and market structure assumed. The broader question remains useful outside the model. A reward attached to the first step can support the route to the second, while a demand upon the second can leave too little incentive to take it. The value created by a sequence has to sustain the people whose participation the sequence requires.⁴

4. Jerry R. Green and Suzanne Scotchmer, “On the division of profit in sequential innovation” (1995), pp. 20–22 and 30–32. Their distinction between agreements before and after investment is a model result with stated contractual assumptions, not a description of all licensing practice.

The reward can also be separated from a charge upon each use. A research grant pays for an inquiry; a prize can promise payment for a specified achievement; a purchase of rights can compensate an owner while allowing broader access afterward. Each proposal relocates a problem. Someone must decide which inquiries deserve support, how a claimed success will be recognized, or what price to pay for rights whose value the seller may know better. The treasury's payment draws on resources with other uses, and an apparently precise target can reward the measured result while neglecting the purpose behind it. Arrow's discussion of alternative research institutions put the allocation of funds and incentives for their effective use together. Opening access is a valuable aim, but funding discovery still requires an institution able to distinguish an achievement from a promise and a worthwhile inquiry from an attractive application.⁵

5. Arrow (1962), “Alternative Forms of Economic Organization in Invention,” pp. 623–625. The grant, prize, and rights-purchase comparison is this plate's application of the distinction between financing discovery and charging for use; Arrow's discussion supplies the underlying questions of resource allocation, uncertainty, and incentives.

The patent is therefore a boundary drawn around a thing that need not be diminished by sharing. Its purpose can be understood only by looking beyond the fence to the work the prospect of a reward may call forth, and beyond that work to the uses the fence may prevent. A successful arrangement has to sustain discovery and leave room for what discovery makes possible. Neither the ease of copying nor the cost of a laboratory, taken alone, settles the balance. The invention may also gain value through the number of people who adopt it: a method becomes a standard, a device gains compatible partners, a language finds more speakers. That introduces a further consequence of one person's choice for another. The Network follows the benefit that grows when use itself brings more people within reach.