Template Tools
| Benoit: Data Caps Not the Right Congestion Solution |
| Wednesday, 30 November 2011 04:33 |
 Caps don’t correspond sufficiently to peak usage. “Congestion pricing” would be a far more effective tool than data caps, according to an analysis of actual carrier traffic by Benoit Felten’s new consulting outfit, Diffraction Analysis. It breaks the actual data from a U.S. ISP into five minute intervals and isolates three hours of highest traffic, the only time the issue arises. The heaviest users accounted for a much smaller percentage of the traffic in those peaks. Many of the heavy users had very little impact during peaks, presumably being polite and downloading primarily overnight. During peaks, the top 10% of users varied dramatically at different times; 40% of all users were in the top 10% during at least one of the intervals.
Economist Scott Wallsten has suggested that use of data caps, rather than some form of congestion pricing, implies that congestion is not driving the rules. Most people, but not necessarily Scott, believe that caps at current levels are about blocking competitive video, not dealing with actual congestion problems. 150-250 gigabyte caps have little effect on either revenue or congestion because remarkably few go over them. On wireless, large company caps of less than 10-20 gigabytes are rarely related to congestion problems that can’t be solved with modest investment. For a company like AT&T, spending 2-4% more of revenue for a few years would dramatically increase capacity. But in wireless Scott sees a role for price discrimination as maximizing revenue even if the actual cost of carrying the data is only minor. To over-simplify, heavier users are on average willing to pay more so charge them.
AT&T and Verizon have iPad data offerings with very low caps at $15-20, attracting users who wouldn’t want to pay the more usual rates. Tiers of say $30 & $50 allow charging the heaviest users more while offering moderate users a more attractive price. If competition is doing a good job driving down prices, all but the heaviest users might benefit. In practice, wireless competition is rarely that effective. Benoit’s data is that peaks are created when a large number of users pull more data at the same time, not when a few users pull data outrageously. The relatively ineffective caps punish punish a good number of regular customers for no good reason. Most existing mechanisms to limit customer usage apply indiscriminately at times when the network is under strain and at times when it is not. That is not only unfair but has no cost-based justification. “Customers should not be penalized for usage outside of peak hours.” Customers hate caps, I’ve observed, so they are rarely a good idea unless much higher than the current ones. Diffraction Analysis has the study for sale at a professional price . They provided me a courtesy copy of the full study which improved the accuracy of my reporting.. |
| Last Updated on Friday, 02 December 2011 02:01 |