Mark S. Manasse
Systems Research Center
Digital Equipment Corporation
This is all interesting and necessary, but it fails to address the problems of a market segment that I expect to grow rapidly, once it becomes at all possible: extremely low-priced walk-up transactions. The proposed payment schemes all come with fee schedules that limit transactions to be fairly valuable. In practical terms, for today's systems, and today's mechanisms for transactions, the fees come to a minimum of approximately twenty-five cents for credit-card-like transactions, and at least a penny for services that provide a level of aggregation before charging a credit card. Over time these fees are almost certain to drop, but the current performance of disks, processors, and networks suggest that these fees aren't going to come down due to economic pressures alone: a system that can only handle a dozen transactions per second per computer needs to charge fees in this range in order to be profitable--and that's the kind of transaction rate implied by systems that require non-repudiatable digital signatures. Increased performance over time will increase the transaction rate; I contend that it's interesting to look at the results of applying that to decreasing the price of transactions.
With fees in their current range, the minimum transaction is likely to remain between five cents and one dollar. For tangible goods, this is unlikely to be a problem; even penny candy costs a few cents these days! For information goods, however, the costs for manufacturing and delivery are quite small. If the overhead for billing could be reduced to the same level, what changes might take place in information delivery? Does it matter if the smallest transaction is five cents or a mil?
I contend that it does matter. At five cents, one can decompose a daily newspaper into a few sections (sports, business, world news, local news) without significantly increasing the daily outlay for information. But this level of decomposition doesn't change the nature of information delivery. At a mil, we can decompose the newspaper into individual articles, comic strips, and horoscopes.
And this information marketplace isn't limited to newspapers: individual stock quotes, background stories culled from the morgue, technical papers, historical commercial information, index retrieval, almost anything where the information production can either be automated, or where the market is large enough. An anarchic information provision world, where payment can be rendered not only for the information, but for the filtering to provide just the information that's fit to download.
A decade hence, assuming that computers (and their components) continue on the price and performance curves of the last two decades, the minimum transaction grain will be an order of magnitude smaller than it is today. The smallest transactions will be in the sub-penny range. In a decade, we will see pricing by the web page. We can unbundle the newspaper and the magazine, derive revenue from the newspaper morgue, and sell information without restricting consumers to a subscription to a small number of sources. Citation and reference by hyperlink can replace quotation. "Fair use" photocopying can be replaced by links, and the information gets to the students that care at lower total cost, and with revenue returned to the author rather than the paper and toner companies. Editors, authors, and publishers become more independent, as an editor becomes anyone who can assemble a useful list of pointers to interesting documents, and find a market for that list.
This, I argue, is inevitable. The alternative is not that information remain expensive and bundled, but that copyright become meaningless. Digital storage of information is cheaper than the paper it replaces, and your published works will be photocopied and scanned--as long as it remains overwhelmingly cost- effective to do so.
I have a clear preference between these alternatives. Authors and editors need to be compensated in order to produce quality work. A scheme in which compensation continues to flow to the creators of a piece of work is superior to one in which compensation is incommensurate with use. The remaining question is whether the technology and the will to preserve copyright will leap ahead of the technology and the will to subvert it, or not. I can't speak to matters of resolve, but I will speak to matters of technology: storage, scanning, and OCR technologies are running ahead of the pricing models for current schemes of electronic commerce. While electronic commerce will eventually realize the kinds of economies necessary to preserve copyright, it's not certain that it won't already be too late.
Therefore, it becomes interesting to look at techniques for accelerating the move to sub-penny transactions. There are a few approaches that can be followed.
First, we can look to aggregated service providers. CompuServe, America Online, and (soon) the Microsoft Network provide simple billing models for access to information. The main problem is that the information that they can pay for is information for which they have contracted, and information that they deliver. This makes it hard for smaller information providers to gain access to users, makes it hard for consumers to retain much privacy from their service provider, makes it hard for information providers to determine whether the level of use reported by the service provider is accurate, and makes it hard for consumers to tell whether they are being accurately billed. Do consumers and vendors care about the privacy and accuracy issues? Not so much, today, but a good example of inappropriate behavior by an aggregated service provider might make people care.
Or, we can look at Millicent for an alternative architecture that provides better accuracy guarantees and modest privacy guarantees, if you believe that your payment intermediary isn't snooping all of your network traffic.
To Become a USENIX Member, please see our Membership Information.