Spectrum Policy - What about Congestion?
Stephen Kirchner (via
Arnold Kling) points to
this debate about spectrum pricing, which has arguments from Eli Noam, Lawrence Lessig, Richard Epstein, and Thomas Hazlett. Noam argues for a spectrum commons but with congestion pricing; Lessig argues for a pure commons; Epstein and Hazlett argue for selling of spectrum as private property and letting the holder do whatever they want with it (as opposed to the current system where the FCC specifies exactly what the band can be used for).
When looking at this issue previously, I hadn't thought through the potential implications for congestion of the commons approach. Once I started thinking this through, the approach lost a lot of its appeal. Noam's proposed solution of overlaying peak pricing on a commons system is difficult to understand from either an economic or a technical perspective - who would determine the acceptable level of congestion and the appropriate pricing mechanism? Apparently the government, which seems to leave us pretty much where we are today.
Lessig notes the likely immense cost to implement a peak pricing system, such as that advocated by Noam, and compares spectrum to the Internet, which, he argues, has seen tremendous innovation due to the fact that there is no congestion pricing system. I agree that TCP/IP, the end-to-end packet-based system of the Internet, has resulted in signficantly greater innovation than would have been likely if the Internet was built on ATM (the phone companies' desired protocol). This doesn't mean that there aren't problems with congestion, and that congestion pricing wouldn't unlock yet more innovation and more effective use of what is, ultimately, a scarce amount of bandwidth. For example, my cable modem service here in Hong Kong slows down considerably in the evening, which is peak time for home use. The problem is not with the local loop but rather with their connection to the US. It is sometimes impossible for me to retrieve email from my US-based server, or to view web pages on US-based sites. Clearly Internet congestion is a problem.
One way of addressing peak congestion is for the ISP to require that everyone access the web through a proxy server. A proxy server reduces redundant downloads, but introduces an aspect of centralization that requires significant monitoring, maintenance, and upgrading, and may become a source of congestion itself. In addition, the required use of proxy servers has privacy implications, makes end-user configuration more complex, and can limit the ability of end users to run new services. A better solution would be to price bandwidth within the ISP's network differenly than external bandwidth.
Different pricing for different 'types' of bandwidth would have higher
mental transaction costs in the short term. However, there already are applications out there for decentralized distribution; widespread congestion pricing would increase the use of these and other as yet undeveloped apps, leading to rapid improvements and refinements, and eventually applications that even a non-technical user could easily install and run. This solution would increase the transparency of the network and further push innovation and control outwards to the edge, replacing the centralized (and unenforceable) rules limiting the number of computers 'allowed' to use each broadband connection, and increasing the overall quality of service for everyone. I do agree with Lessig that we are better off for not having spent the time and money on congestion pricing back in the early days of the Internet. Today though, we have the technology, it's not expensive, and it would cost-effectively encourage even further innovation and efficiency.
Coming full circle back to wireless, the fastest way to implement new types of pricing mechanisms (which will be necessary at some point, even if not just yet), is through clear private property rights that encourage lots of experimentation by private firms. If we put government in charge of the pricing mechanism, even just for peak usage (as Noam seems to be endorsing), we will end up with an extremely expensive, monolithic system, that will be built to meet the needs of the past.
Posted by Stephen Bronstein at March 28, 2003 05:59 PM