These are very fine comments, which I cannot respond to adequately in the time I have. Let me begin with a qualification and an amplification. I said that a drug company can get, on top of the normal 20-year patent term, up to five more years for the time during which the FDA is deciding whether to approve the drug for sale. This is true, but it is also true that there is another limit--14 years from the date of approval. So if approval came 10 years after the drug was patented, the patentee would be entitled to a total patent term of only 24 years. The amplification has to do with the opening paragraph of my posting, in which I noted too quickly that the basic economic objection to monopoly is that it deflects consumers to inefficient substitutes. This is an important point that is not intuitive, so let me explain a bit further. Suppose there are two products which are very close substitutes for one another, but one of them costs $3 to produce and the other costs $4. Then society can satisfy consumers' demand at least cost by supplying the first rather than the second product, and this will be the market outcome if each product is priced at cost (including a profit adequate to attract and retain necessary capital and compensate for risk--and such a profit, a competitive or normal profit, is in fact just another cost). But now suppose that while the second, the more costly, product is priced at cost ($4), the first, the cheaper to produce, is monopolized and its price is as a result $5. Now consumers (not all of course, or the monopoly price would not be profitable) will switch to the second product, and this will involve a social waste becase their demand could be satisfied at a lower cost by the seller of the first product. The intuitive objection to monopoly is that it gouges consumers, but insofar as consumers stick with the monopolized product even when its price goes up, their higher cost is offset by the seller's higher profit; there is merely a transfer payment. (There may be an ethical objection, but that is beyond economics.) In a second round of analysis, however, and this is relevant to the discussion in my posting, the prospect of the higher profit induces additional investments (e.g., in obtaining a patent, which is a right to exclude close substitutes), which may be wasteful. Now to the comments. One commenter pointed out correctly that my "first past the post" hypothetical example was oversimplified. The reason is that since each different molecular entity can be patented, the firm that loses the race to be first may still have a valuable product to patent. The principal SSRI antidepressant drugs, for example, such as Prozac, Paxil, and Zoloft, are all different molecules, separately patented; and because they have different therapeutic properties for different patients, each was able to command a significant segment of the market. Another commenter asked, if the 20-year patent term is too long, as it probably is, how can the copyright term, which is now the life of the author plus 70 years, possibly be justified? It can't be justified in traditional terms, that is, as necessary to induce people to write books, make movies, and engage in other creative expression, because the discounted present value of uncertain receipts 70 years after one dies (which might easily be a century after the work in question had been written) is negligible at any realistic interest rate, especially given the likely depreciation in the market value of the work. William Landes and I, however, in our recent book "The Economic Structure of Intellectual Property Law," argue that there are two possible justifications for indefinitely long copyright terms: to prevent congestion (overuse of a copyrighted work might reduce its value to nothing), which is a traditional economic argument for property rights; and to induce investment in maintaining the value of the copyrighted work, for example by producing frequent revised editions (only the revisions could be copyrighted independently). These arguments are applicable, however, only to the tiny fraction of copyrighted works that retain a substantial market value beyond a few years. The serious problem with long copyright terms concerns not those works, but the multitude of less valuable (but not valueless) works that ought to be in the public domain so that they can be published without the publisher having to engage in costly negotiations to obtain a copyright license and also so that they can be used (again without need for cumbersome negotiations) as raw material for new creative works, almost all of which build on previous works rather than being created ex nihilo. That problem could be solved even within the framework of the very long copyright term if the courts would say that it is okay ("fair use") to republish an old work if the copyright owner has failed to provide notice of his whereabouts and as a result it is infeasible to negotiate a license from him. Such a rule would give rise to private copyright registries that publishers could consult when they wanted to publish an old work, and if there was no copyright listed in the registry the work could be published without copyright permission. This argument is developed in a forthcoming article by William Patry and me in the California Law Review. Similar concerns about licensing costs are increasingly voiced by academic researchers who use patented "research tools" (such as Harvard's "oncomouse") in their work and understandably don't want to have to conduct patent searches and negotiate for patent licenses, especially if they are using multiple such tools. I am intrigued but unconvinced by the suggestion that once a drug patent expires, generic manufacturers should be allowed to use the trademark of the patented brand; in other words, the trademark would lapse with the patent. I do not favor this because the consumer and physician are entitled to know who the producer is, and the trademark is the economical identifier of the producer. But I note that a trademark will lapse once it has become generic in the sense that people are using the brand name to designate the product, not just the brand. Many trademarks have become generic in this way, such as yo-yo and brassiere--and, in this country, aspirin. The same fate may befall some of the currently popular patented drugs. One commenter works for a small biotech company and emphasizes the need for patent protection for the company's products because he expects it may take 10-12 years from patent to market. I agree that the smaller the company and therefore in all likelihood the riskier its prospects, and the less the effective length of patent protection, the stronger the case for long patent terms. I do not know whether it would be feasible or appropriate to try to differentiate patent lengths by size of firm, nature of industry, etc. The advantage of having the patent, copyright, etc. term independent (or largely so) of a particular industry is to reduce political pressure for special-deal legislation, whereby for example powerful industries would obtain longer terms for their patents and copyrights to the disadvantage of the weaker industries, which might need such terms more. Several commenters debate the significance of the fact that the big drug companies spend more on advertising and other forms of marketing than they do on research. That in itself is untroubling. Indeed, such ratios are meaningless. It is both important and costly to get information about new drugs to physicians and patients; and there is only so much that can be spent on research before the returns from further research begin to plummet. So no inference of inadequate investment in research can be drawn from the fact that companies spend less on research than on promotion. The difficult questions are the accuracy and utility of the promotional information distributed by the drug companies. I agree with the commenters who express profound skepticism about software patents (and copyrights). But that is a topic for another day.