Some quick notes, including a great deal of thoughtcrime that should be kept away from young children. Is this a great mailing list or what? As a reminder, answers to frequently asked questions about RRE can be found through the RRE web page, http://communication.ucsd.edu/pagre/rre.html Remember "push" technology? Gosh, it seems like two months ago already. Seriously, why have 1000 times more people heard of "push" technology than IP Multicast? Slowly but surely, the Internet world is talking itself into a regulatory framework to govern Internet life. The turning-point, I think, has been the fight at the FCC over flat-rate telephone service and other issues relating to the local phone companies' complaints about Internet users. Although I support flat-rate local phone service, I have to say that I'm a little disappointed with certain Internet people who envision all sorts of futuristic electronic commerce scenarios in which everyone pays for everything incrementally using micropayment systems -- what Vinny Mosco called "the pay-per society" -- but who then turn around and resist that same principle when it applies to their own use of the Internet. These folks want a la carte for everyone else, but the buffet for themselves. Along the way, we're also starting to see some realism poke its nose out from under the ideology that has long obscured the Internet community's discussions of regulatory issues. For example, in Declan McCullagh's column in the June 1997 issue of Wired (page 183), we read: the digital nation has misidentified its foe. As a rule, Washington's bureaucrats are not power-crazed authoritarians; most are reactive creatures who simply respond to demonstrations of influence and power. Bell Atlantic, PacBell, Nynex, et alia leaned hard on the FCC for access fees, and the agency reacted in its own instinctively bureaucratic way. The high tech community responded by forming its own ad hoc coalition to pressure the FCC, and thousands of Internet users chimed in to express their collective dismay. ... the real threat to netizens has come from complacent telcos and their legions of starched-collar lobbyists, not the FCC. The distinction is important, because the old rule of thumb holds true: The enemy of our enemy may occasionally prove to be our friend. This would be the most unremarkable common sense in any sane decade. Of course, the passage that I removed with an ellipsis does read as follows: Of course, the best way to win not just the battle but the war maybe to remove the commission's power to regulate the net altogether. Still, The old fantasies are still there. The fact is, however, now that Uunet and its brethren are starting in earnest to behave like an oligopoly, and now that the Internet's peering system is starting to collapse, we're starting to hear large ISP's call for regulation. The FCC, having heretofore made extra sure that nobody thinks it wants to regulate the Internet, is suddenly starting to stroke its bureaucratic beard and allow that maybe some type of regulation is required after all (New York Times, 5/12/97). What's the next step? It seems like the large ISP's -- the operators of large national and global backbones -- are moving to treat the small ISP's as customers rather than peers. Some say that these organizations are just passing along their costs in the normal market fashion, and that the protesters might as well oppose the law of gravity. Others point out that these companies compete with the smaller ISP's for customers; they describe the situation in terms of restraint of trade. In a sense it doesn't matter who's right, since even the large ISPs' defenders are effectively arguing that the Internet is inherently centralized. This argument has a lot of economics going for it, given that the Internet is pretty much the textbook definition of a natural monopoly. (Don't blame me. I didn't write those textbooks.) Once the oligopoly is in place, how long until the large carriers start merging? Maybe they'll have to, for several reasons. For example, if one carrier becomes a dominant player in the market to provide Internet service to large businesses then it can start offering quality-of-service guarantees within its own system, thus creating network externalities for its customers. Can someone explain why this won't happen? I've sat through a lot of lectures about the inherently decentralized nature of the Internet and all the geodesic this-and-that, including lectures by people who claim to be authorities on telecom economics, but I still haven't gotten anything like an explanation of why the Internet is any different in its fundamental economics from the phone system. The fight over domain names illustrates what I mean. Domain names are a lot like telephone numbers, and the fight over domain names is a Keystone Kops parody of the longstanding dispute about phone numbers in analyses of telecom competition. The level of ideology here is amazing. First the government, seized by the private-good-public-bad logic that substitutes for thought these days, decided to privatize the issuance of domain names -- by simply handing over the function to a private firm. This isn't just happening with domain names, of course. It's also happening with water utilities and all sorts of other essential services worldwide. But let's focus on domain names. Government services might be clunky and drab, but at least they are subject to some kind of democratic accountability, for example through being hassled by members of Congress when they make people mad. Private monopolies operate under no such constraints, and in any normal world we would be unmystified by the dramatic deterioration in domain-name service since the government stopped doing it. If you're going to privatize domain name service, or anything else, you need to create the conditions for a fair market. That's what telecom policy is all about. If you want a fair market in phone service, people need to be able to move their phone numbers from one carrier to another without significant expense. Likewise, if you want a fair market in domain service, people need to be able to move their domain names from one service to another. Doing this neatly and cleanly is a very technical matter, and it turns out that small details of the rules can make a big difference as a practical matter in the fairness of the marketplace. And so why is this knowledge not being applied to the Internet? Because the myth of the "vast and unregulated Internet" keeps short-circuiting our brains. So what's the deal here? Do I favor centralization? Is that why I keep making these inconvenient arguments? No, I believe in a decentralized distribution of power in society. What I disagree with is arguments that technological developments will, all by themselves, bring us that kind of society. Those arguments are, so far as I can tell, pure stipulation. And what's this stuff about regulation? Am I in favor of regulation? Regulation isn't inherently good; I do not have, as one Internet rhetor put it, a psychotic need to control people. I want a telecommunications infrastructure that can support a democratic society. In pursuit of that goal, I think hard about the interactions between technology, economics, and policy, I write down whatever conclusions I come to, and I hope that democratic societies care enough to take the effort to stay democratic. I was dismayed a while back, you may recall, to find myself on a Web page of "celebrity e-mail addresses". Since then, I have been through a whole elaborate drama. The maintainer of this page refused to remove my name from it, and has failed to respond to numerous additional requests. Meanwhile, new such lists keep popping up, mostly copied verbatim from existing lists, and I continue to receive a steady stream of requests for autographs from people who are obviously spamming every celebrity address they can get hold of. I've developed a routine: when I get an e-mailed request for an autograph, I respond by explaining that I am not any kind of celebrity, do not respond to autograph requests, do not even own a photograph of myself, and do not have a secretary or publicist to filter my mail for me. I also ask for the URL of the Web page from which the requester got my address. Sometimes the person responds with a URL, and in such cases I track down the maintainer of that page and request to be taken off it, and sometimes the maintainer actually replies and does the right thing. Mostly, though, I get no response at all. Although it would be wrong, I feel a thirst for revenge. Consider, for example, http://www.ingress.com/~wrp/email.htm , a Web page maintained by someone whose e-mail address is wrp@ingress.com. This is the one who explicitly refused to remove me from his page. If I were a bad person then I would ask you to send this guy a note and ask for his autograph. You might tell him, for example, that you're going to auction autographed pictures to raise funds for your school. Or you might tell him that he's your biggest hero in the whole world. (There would be no need to mention his name or exhibit any knowledge of who he is.) You might even attach a photograph of yourself and implore him to get you a role in a movie. You would spread these messages out, so that he received a few of them every day. That way he would know what it's like to be listed on his Web page. He's making a business out of this service -- his page includes advertising, or at least he's soliciting advertisements -- and it's not reasonable for him to profit by creating nuisances for innocent people. But his offense somehow doesn't seem like it's in the same category as spamming, so I'll refrain from further increasing the general level of unpleasantness on the net. In my message about the FTC privacy hearings, I did not mean to imply that no privacy lobby has offered comments. The comments from the Electronic Privacy Information Center are a good source of arguments for privacy warriors: http://www.epic.org/privacy/internet/ftc/epic_comments_497.html My most recent rant about Microsoft was certainly over the top in its rhetoric, but long-time readers of this list are aware of the substantive analysis that backs it up. My message was apparently posted to some other lists, whose members wrote me all sorts of poignant messages of the sort that result from lack of understanding of economics. Some of these messages included detailed catalogues of every mistake that MS's competitors have ever made. These analyses miss the point, however, until they explain in structural terms why such mistakes could lead to such catastrophic and irreversible defeats. Particularly poignant were the messages from the good and talented people who work in the trenches of MS, talking with users and making deadlines without seeming to understand their own company's strategy or the economic dysfunctions that allow that strategy to work. If all you've heard is the usual stuff about markets -- the quaint 18th century theory of supply and demand that has little to do with the high-technology markets of the 21st century -- then any claim that inferior products dominate a market will inevitably seem to imply that customers are stupid, that employees are stupid, or that big bad government is meddling. It is not so, and none of these implications has any part of my argument. Nor are MS's products uniformly bad. MS's componentization and object strategies are aligned with broad trends in the industry, and are just as competent as the competition's -- just more proprietary. The stuff about market dysfunctions is just a broad generalization; each market, and each technical feature, has to be analyzed in its own terms. And, like most people, I expect big things from the world-class people that MS has been hiring in the last year or so, particularly in computer graphics. Dollars to donuts says those big things will employ proprietary standards whenever possible, but they will be big nonetheless. The point, then, is not that MS produces bad software through any kind of character failing. Nor is the point that the market is specifically selecting for poor quality. The point, rather, is that the dysfunctions of high-technology markets often reward other things besides quality, for example the highly developed strategies by which proprietary standards come to be entrenched in markets due to the obscure wonders of network externalities. This is not the only force in the market, but neither is it a marginal force. Recent reporting on Microsoft: Ken Auletta has a longish article on Nathan Myhrvold's legendary memoranda in the 5/12/97 New Yorker (they really didn't see the Internet coming), and the 5/26/96 issue of Fortune includes a good article on Microsoft's NT strategy. Look at MS's strategy with both Windows and Office: starting with a system that enjoys overwhelming market share and whose position is reinforced by network effects, then successively adding components to that core system that compete with applications companies, compensating if necessary for the relatively poor quality of those components by drawing on immense cash reserves and a vast distribution system to cut prices until network effects take hold. Next, imagine what would happen if MS could generalize that strategy to NT and BackOffice. If you can't get the word "scalability" out of your head then you get the picture. A directory of anti-Microsoft Web sites can be found at http://www.geocities.com/SiliconValley/Pines/3334/super.html I can't say that I was enlightened by the few that I clicked on at random, but maybe others will find something of value. I recommend an interesting online newsletter called "Above the Crowd", edited by J. William Gurley of Deutsche Morgan Grenfell and described as "a bi-weekly publication focusing on the evolution and economics of the Internet". To subscribe, send a message to atc-request@abovethecrowd.com with the word "subscribe" in the body. I would forward an issue to RRE, but it's copyrighted and I haven't gotten any response to my request for reprint permission. Peter Neumann's Congresional testimony on the issues raised by the flap about the Social Security administration's Internet system can be found at http://www.csl.sri.com/neumann/ssa.html Some people are complacent about spam; their argument is that if nobody replies to the spammers' advertisements then the spammers will get tired and go away. I disagree with this argument, for two reasons. The first is that it fails to distinguish between spam services and their clients. A spam service might sell spam software or lists of e-mail addresses, or it might send out its clients' messages to an agreed number of addresses, or it might perform some combination of these chores. Anybody who's low enough to send spam is low enough to mislead potential clients with tales of instant riches over the glamorous Internet, and there're lots of naive people out there with dollar signs in their eyes. Those naive people aren't talking to one another, and I doubt if they'll spend much effort publicizing the ignominy that resulted from their adventures in spam. The second and more surprising reason is that some people do actually buy stuff from spammers. Really. I've corresponded with them myself. I ask them why, and they shrug and say, roughly, "it was something I wanted, the price was right, and the offer was right in front of me, so why not?". I guess that some people's mamas never taught them anything, so here's why not. One reason is practical: when you get an offer through spam, you can have little confidence that it's legitimate. Evidently big bad government has been cleaning up the marketplace for so long that some people think nothing of writing a check to a company they've never heard of and mailing it to a PO Box in another state. But the most important reason not to buy from spammers is that it's wrong. That's right -- it's immoral to buy anything from a spammer. Why? Paper mail advertising, while annoying, has a built-in reality check: it costs something like $0.75 to send a direct-mail advertisement, so that response rates of a few percent are needed to recover the costs. Think about it: When you buy something in response to a junk-mail letter, you have to pay for about thirty other people's letters before you get anything of value. The economics of spam are fatally different. Spam costs something like $20 per 100,000 to send, and probably much less for the big operators. (I haven't seen any formal studies of the spam market; that's just my informal impression.) If those numbers are anywhere near right, a spammer can profit with a response rate many thousands of times lower than a paper letter. The reason that spam is offensive, of course, is that it induces a cost on the people who receive it. This cost includes the time and effort of skimming and deleting the messages, and for many people it includes actual money costs for connect time and storage. Thus, when you buy something from a spammer, you are effectively receiving stolen goods: the time and effort you saved were stolen from the thousands of other people who received that same spam message involuntarily and received no benefit from it. Of course, this argument has limits. I don't think people who respond to paper junk mail are thieves in any sense worth worrying about, since the scale of wrongdoing is orders of magnitude less. Besides, contrary to what most people say, my experience has been that I know exactly where the junk mailers have gotten my name, and I find that I can control my junk mail quite well by registering with the DMA's "mail preference service" and consistently instructing magazine publishers, charities, and catalog merchants to keep me off mailing lists. I still think self-regulation in the direct mail industry has been a failure, if only because of the amount of effort needed to stay off the lists. And my life is uncomplicated compared to many people's: I have neither property nor kids. In any case, the capital required to send direct paper mail does encourage the direct mailers to maintain, on the broad average, a certain level of legitimacy. Spamming, on the other hand, has extremely low barriers to entry, which is presumably why spammers seem to consist predominantly of the scum of the earth. Nonetheless, the spammers have grown sufficiently numerous and organized that they are actually trying to justify themselves. You've probably seen some of these arguments. What strikes me is their similarity to the arguments offered by organized invaders of privacy: high-minded appeals to freedom of speech, attempts to blur the meanings of words, promises of self-regulation, and suggestions that the only real trouble is caused by those rude people who object to their practices. We need to put the spammers permanently out of business before they get big enough to hire "public affairs" departments that mysteriously manage to persuade elected officials that these arguments make sense. Once we see the first annual Cyber Promotions Public Opinion Survey on Commercial Communications in Cyberspace, or the first corporate lobbying group composed of former civil liberties activists offering "compromise" legislation in the matter, it's all over. I had an especially bad run of spam the other day. First it was the get-rich-quick scheme, then it was the pornographer, and then it was -- no kidding -- amazon.com books. Yes, amazon.com books spamming its customers with the news of its now having 2.4 million or 3.1 million or 1.8 million books, or something. Well, forget that. I've cancelled my amazon.com account, including a $100+ order I had placed last week, and moved to alt.bookstore (http://www.altbookstore.com). It has lower prices -- about 7% lower on the order I cancelled at amazon.com. Its interface isn't as refined, and they don't seem to list as many books that haven't yet been published. But they haven't sent me any spam. Upon arising one morning and settling at the computer terminal in my kitchen, I had a curious urge to look at the Christian Coalition Web site. Sure enough, the press release announcing Ralph Reed's resignation had been posted hours earlier. http://www.cc.org/publications/ccnews/ccnews97.html#resign You may recall my recommendation of the February 1997 issue of Internet World. I had never paid any attention to IW before, but that particular issue presented such a useful summary of ongoing Internet standards battles that I actually subscribed. Unfortunately, that issue turns out to have been a fluke. The product evaluations possibly aside, the level of analysis in IW has been uniformly poor. They seem to have no awareness of the institutional dimensions of Internet work, for example the need to integrate Web page design with an organizational communication strategy. They also keep repeating the half-truths and outright falsehoods that the Lexis-Nexis people spread around back when their P-TRAK system was being roasted on the net. The last straw, though, was their interview with Jim Manzi in the current issue. Jim basically asserts that he's the only guy who had thought of building infrastructure to support business-to-business commerce on the Internet, and the IW people didn't even begin to call him on this. He has the concept on a very abstract level, but he didn't begin to articulate a strategy that was going to get him there. Well, as you probably read, it turns out Jim's strategy had already been crashing while IW was in press, so that Nets Inc -- which Jim had been funding out of his pocket for a while there -- declared bankruptcy shortly after IW came out. Can we have some actual reporting in this industry? It's hard, I realize, when the whole modus operandi of the IPO artist and would-be definer of de facto standards is -- with perfect economic rationality -- to pump the old hype-o-meter. But the analytical tools do exist now, and organizations with large sums of both money and expertise (not just one or the either) do get better analyses than the readers of IW. If you should lack for either money or expertise, the good news is that you can now almost do it for yourself. From what I can tell, high technology industries can be comprehended with great sophistication through constant study and the application of precisely two sets of ideas. The first set consists of the economics of information and standards (the qualitative parts, not the analytics), and particularly the strategic consequences of network effects. I've reviewed this literature here on several occasions, and it really does explain a great deal. The second set of necessary ideas can be found in the earth-shattering books of Geoffrey Moore, "Crossing the Chasm" and "Inside the Tornado" (both HarperCollins, 1991 and 1995 respectively). I reviewed "Chasm" with bemused respect in TNO 3(5), but only in the last few months have I come to a full appreciation of it. Even forgetting the parts he got from Everett Rogers, it's still an extraordinarily original piece of work. It explains, among other things, part of what happened to Nets Inc, which just couldn't cross the chasm between the easy money of the early adopters and the hard money of the companies that want real, complete solutions to their problems. Not having taken the effort to establish themselves in strategically chosen niche markets, they had neither the positioning, the knowledge, or the cash flow to ready themselves for the transition -- a transition whose time, in the case of Nets Inc's market, has not yet come -- to the period of explosive growth that Moore calls "the tornado". (Denise Caruso's column in the 5/19/97 New York Times includes a good analysis of this, including a review of NI's competitors' strategies.) Everyone whose life is affected by high technology should read these books before it's too late. If necessary, I will personally come to your office and pound my fists on your desk until you do. When you get done with them, you will realize that you had formerly known nothing about the computer industry. Either you or Jim Manzi. John Ousterhout has an interesting rant against Java at http://www.sunlabs.com/people/john.ousterhout/scripting.html The Usability Professionals' Association is at http://www.upassoc.org/ As a frequent user of the Lynx Web browser for plain-ASCII terminals, I've found the Web becoming less and less Lynx-friendly, and in more and more gratuitous ways. Now, thanks to http://world.std.com/~adamg/we.html , I have a name for this syndrome: these sites have been dehanced for Lynx. Check out their links to particularly dehanced sites, as well as their style tips for Lynx-compatibility. It has also been pointed out to me that every child on the planet has already heard my panda joke. It has been pointed out to me that pandas are neither arborial nor marsupial. Recommended books. Isaac Kramnick and R. Laurence Moore, The Godless Constitution: The Case Against Religious Correctness, New York: Norton, 1996. In an era when some religious conservatives argue that the United States Constitution was designed to embody Christian principles, it is refreshing to read the words of the 18th century religious conservatives who argued bitterly against the Constitution on the grounds that it did no such thing. Uwe Poerksen, Plastic Words: The Tyranny of a Modular Language, translated by Jutta Mason and David Cayley, University Park: Pennsylvania State University Press, 1995. A wonderful rant against the evolution of language toward bloodless abstractions like "system" and "communication". John J. Gumperz and Stephen C. Levinson, eds, Rethinking Linguistic Relativity, Cambridge: Cambridge University Press, 1996. A smart and learned (but horribly copyedited) collection of articles that brings evidence and theory to bear in evaluating the linguistic relativity hypothesis, the controversial idea that the grammatical forms of particular languages influence perception and thought. Peggy V. Beck and A. L. Walters, The Sacred: Ways of Knowledge, Sources of Life, Navajo Community College Press (Tsaile Rpo, Navajo Nation, Arizona 86556), 1977. A sophisticated introductory textbook on Native American religion, placing traditional shamanism, the Ghost Dance, and the Peyote Religion (among other things) in historical, cultural, ecological, and religious context. Recommended records, all of which can be described as "folk". Dick Gaughan, Handful of Earth (Green Linnet, 1991). A powerful set of Scottish folk songs invoking historical memory to uphold the dignity of working people. Maddy Prior, Year (Park, 1993). A wise and gentle reworking of tradition by the leader of the modern English folk revival. Steve Earle, I Feel Alright (Warner, 1996). Excellently produced electric white-boy blues by a guy who's wrestling full-time with the devil himself. Butch Hancock, Own the Way Over Here (Sugar Hill, 1993). An excellent songwriter from Texas whose great intelligence sits in the shade of a good-ol'-boy persona. Not recommended: dc Talk, Jesus Freak (Virgin, 1995). You'll remember my recommendations of various exponents of Christian rock. Well, dc Talk's excellent single, "Just Between You and Me", has been inescapable for some time now, so I followed my politico-religious curiousity and bought the record. It turns out dc Talk is actually a bad soft metal band whose other songs, evidently recorded on a four-track in someone's basement, do not remotely resemble the single. I guess that even Christian bands sometimes feel compelled to sell out to record-label marketing tactics. The rest of this message consists of an draft note that I wrote at the beginning of the year but never finished. It's not quite right, and the writing isn't so great either, but maybe it's good enough to circulate in unfinished form... In school I learned about this cool thing called "science". Here's how it works. If someone has a theory and wants everyone else to believe it, they draw out some consequences of the theory, called "predictions", and if the predictions don't come true then the theory is wrong. New Year's Day is a convenient time for rounding up the predictions from the previous year, so let's consider one of those predictions now. You may recall that Bob Metcalfe, godlike inventor of Ethernet, predicted that the Internet would collapse during 1996. If you're reading this message then we can cautiously conclude that it didn't come true. Bob's theory concerned the economics of Internet pricing: flat-rate pricing, he suggested, would lead people to overconsume Internet bandwidth, leading to shortages and collapse; it follows that the Internet needs a pricing scheme (per-packet or congestion-based or whatever -- my arguments won't require me to distinguish between them) to allocate its capacity by market principles. So what's wrong with Bob's theory? Well, some would argue that his prediction did come true. Bob himself is quoted discussing the topic in a Reuters article today. He suggests, for example, that the long delays people experience on the "World Wide Wait" are due to Internet congestion. But those delays are often due to queueing delays in individual servers (including domain name servers), not router congestion. They are also due to the bandwidth of the user's "last mile" connection, often through a telephone line, which limits how quickly a large file can be downloaded, quite independently of throughput on the Internet proper. Imposing prices on individual servers can, and presumably already does in some cases, regulate demand for their use, but that's different from imposing prices on routed packets. He also points out that outages can be caused by wire breaks and bugs in routers, but that's also irrelevant to the issue of capacity allocation and pricing. The Reuters article then refers to the AOL shutdown over the summer and Stanford being cut off from the Internet in October, but neither of those problems had anything to do with congestion either. Furthermore, packet pricing would not eliminate congestion; it would only reduce congestion to the level that people are willing to pay for, and it would presumably make possible the creation of different levels of service quality with different prices. The article reports that "experts such as Metcalfe predict similar types of outages next year as telecommunications companies and Internet providers struggle to keep up with demand". But this struggle would still be taking place if users paid for Internet routing by the packet, given the rapid increase in demand. Is that increased demand a result of economic distortions that underprice the service? If it was then the companies wouldn't be investing all that money to provide the services. Will these experts eventually suffer the same mockery as the scientists who have been predicting for however many decades that the whole global ecosystem will collapse real soon now? I hope not. What I hope is that everyone recognizes that modern information and communications services operate by very different rules than the commodities that motivated the simple classical supply-and-demand stories in the 18th century. Many economists already recognize this, and I hope it's their thinking that is reflected in coming generations of the Internet architecture and not that the ideology of those who simply cannot stand to see anybody eat a free lunch. Of course, I'm arguing with a newspaper article here. Maybe the liberal media have tried to make Bob's argument sound stronger than it is. [And, I would note here in May, that Bob did eat his words, just as he promised.] My point is simply that we shouldn't slip into an easy equation between technical difficulties and delays on the Internet and the need to erode the practice of flat-rate pricing. The issue of flat-rate pricing has arisen on several fronts this year. Some of the regional phone companies, for example, have been claiming that flat-rate service is allowing Internet users to cause uneconomic congestion in local-loop phone systems. Computer industry types dispute their numbers. An instructive text here is Milton L. Mueller and Jorge Reina Schement, Universal service from the bottom up: A study of telephone penetration in Camden, New Jersey, The Information Society 12(3), 1996, pages 273-292. Bell Atlantic hyped this study to the skies when its results first become public a couple of years ago, presumably because it would seem to undermine one of the most common arguments for universal service cross-subsidies, namely that the poor could not otherwise afford telephone service. Mueller and Schement actually went and talked to some poor people about telephone service, and they found that these folks had some quite rational reasons not to want a telephone. The relevant reason for present purposes is that it is hard to control long-distance charges. If you're poor then you don't have any money in the bank, so that any unexpectedly large bills can cause calamities. For example, if you're poor then you probably have some relatives whose troubles are worse than yours. You're likely to feel an obligation to put these relatives up occasionally, whereupon they can perhaps stick you with a large phone bill, whereupon you can miss your rent payment and end up on the street. Much better to put your money into something whose costs you can predict, like cable TV. Okay, so much for universal service subsidies. (Maybe.) But what about flat-rate pricing? It is very commonly argued that emerging information and communications technologies dramatically lower transaction costs and thus make it possible for numerous markets to move from flat-rate to per-unit pricing, or from a free public service to a paid private service. Maybe one side-effect of this shift is that households' expenses become ever harder to predict; the poor will suffer either by having to forego those services or through the increased risk of sudden unexpected insolvency. Issues of "haves" and "have-nots", in other words, are not just matters of distribution; they may also be affected to a significant degree by pricing structure. As serious telecommunications policy people are well aware, the issues are more complicated than simple supply-and-demand stories can comprehend. Many people wonder why subscribers choose flat-rate deals that can often result in larger bills than under measured service, and one reason is that the flat-rate premium is a kind of risk hedging not dissimilar to larger organizations' use of options to hedge commodities contracts. It is also argued that flat-rate pricing effectively shifts costs from heavy users to light users, and that might actually become true if one segment begins using huge amounts of local-loop capacity to maintain Internet connections. But given the distributional virtues of flat-rate service, we should be very sure that this is happening, and that no alternatives are available, before giving in to the phone companies' decades-long campaign to get rid of it. Fortunately for most of us, and unfortunately for the ideologists, I note that the predominant direction of movement in new information and communications services themselves is away from per-unit and toward flat- rate pricing. (I don't know of any Internet server that uses congestion pricing. It would be quite practical for the Wall Street Journal site, for example, to lower its prices at night, or to post prices dynamically according to its current load.) Long-distance telephone has been moving toward distance-invariant pricing, for example, which is a good first step. AOL has also shifted to flat-rate pricing. Much was made of the first-day congestion problems that resulted, but AOL could hardly have been surprised by those problems, and must have made the decision despite knowing that they would arise. The ISP market also shows few signs of moving away from flat-rate pricing, although the fundamentals in that sector are much less stable. [And since I wrote this draft in January, Uunet has moved back toward measured rates.] I realize that this could be a bad thing: if everyone has flat-rate service then everyone has to pay the costs incurred by the average user; less-than- average users will be reticent to sign on, thus pushing up the average; this effect will reinforce itself to reduce the total number of users below the levels that would be reached otherwise, and everybody, the poor included, will suffer by losing the positive externalities (economies of scale and network effects) associated with a larger user community, including the large market that is required to make advanced applications feasible economically. On the other hand, the transaction costs involved in administering measured or congestion-based pricing raise the average bill and thus depress positive externalities as well. These are all theories too, and we need much more evidence about which theory is correct before we do anything rash. None of this answers the question of why Bob's theory is wrong. People give varying answers; my own answer (which I am sure is not original with me) has two parts. The first part concerns substitution. If bread were free then people would instantly start consuming vast amounts of it, since they can substitute bread for several other kinds of food. Consumption might be bounded in urban areas, since there's a limit to how much bread any given household is able to store and eat, but in rural areas people would feed it to animals and maybe even use it as fertilizer. Internet services, however, are not tremendously substitutable for other things. College students substitute e-mail for telephone conversations with family and friends back home, of course. But this brings me to the second part of the answer, which is that current Internet services are all structured in ways that effectively create proxy prices for Internet bandwidth. To send lots of e-mail, for example, you have to sit there and actually type it all in. Even when your e-mail is sent to many people, the amount of e-mail that you send will be regulated by the amount that people are willing to receive. This latter principle has been sorely tested lately with the spam wars, but the last I saw the online service users were winning and the spammers were losing. People's use of the Web, likewise, is heavily regulated by their willingness to sit and wait for files to be downloaded. Corporate users with T-1 lines to their desks can consume a whole lot more bandwidth than home users with 28K phone lines, of course, but the fact remains that every second they wait for a download carries an opportunity cost that somebody (either the user or the boss) cares about. Now this may all change when new applications become available. I oppose Internet telephone, for example, because it makes it easy for everyone to consume vast amounts of Internet bandwidth. Maybe when IPV6 arrives we'll be able to segment different uses of the Internet so that people who want to mess with bandwidth-intensive streaming data types can pay for it and leave the rest of us alone. I honestly don't understand the stories people tell about Internet telephone being vastly cheaper than regular telephone; when I try to understand the details, it has always turned out that they're talking about a lower grade of service. Maybe the new "push" services and order-of-magnitude-cheaper hard drives will allow people to download huge amounts of stuff easily, routinely, and automatically, just in case they want to look at it later. Maybe lots of things. My point is simply that economic stories depend in great detail on the workings of the technology, and on how the technology fits into people's lives. end