NNSquad - Network Neutrality Squad

NNSquad Home Page

NNSquad Mailing List Information

 

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[ NNSquad ] Re: NYT: Differing views on Time Warner's Bandwidth Cap Experiment

As someone who's spent lots of time internationally, there are many tiered pricing plans for DSL. All you can use are often MUCH more expensive. BUT, the entry plans are dramatically cheaper, e.g., $2-6/month in India.

So, if only a few percent of people are using up "too much" bandwidth perhaps a solution would involve congestion pricing and/or graceful degradation only for those who go above reasonable fair use caps. Is there too much complexity in this? If I want to download something huge, there should be ways for the system/network to signal "off-peak" pricing. I think we need some new protocols/tweaks/out-of-channel signalling for that. Given these are all happening on the last mile, it shouldn't be hard to program that in (in a layered, scalable, manner).

There are parallels in the power industry, where I use kWh any time of day (as a retail consumer) for the same price. If I knew, with minimal effort or better, automatically, when things were not only peak but expected to peak, I and my appliances could behave a little better (to the extent there's a win-win). Everyone can benefit if there is less peak capacity required. Of course, I need to see the numbers more (and we don't have transparency on this). This need not be a case of wealth transfers only - there can be value creation as well. In one simple world with a zero-sum-game, for every penny charged for over-usage, the average bill for others should go down commensurately.

A first Q to me is to what extent is or isn't this a zero-sum game?

Rahul

************************************************************************
Rahul Tongia, Ph.D.
Senior Systems Scientist

Program in Computation, Organizations, and Society (COS)
School of Computer Science (ISR) /
Dept. of Engineering & Public Policy

Carnegie Mellon University
Pittsburgh, PA 15213 USA
tel: 412-268-5619
fax: 412-268-2338
email: tongia@cmu.edu
http://www.cs.cmu.edu/~rtongia



Barry Gold wrote:
Lauren Weinstein wrote:
http://bits.blogs.nytimes.com/2008/01/17/time-warner-download-too-much-and-you-might-pay-30-a-movie/?ref=technology


This is beyond stupid, and I hope TW gets smacked down by its customers if and when they try this. See my earlier post on what customers want.

Item 1: No surprises. A $30 surcharge on their bill is going to result in a very unhappy user. I know what Time Warner is thinking: when the user sees a $30 surcharge on his cable bill, he'll buy a higher service tier and we make more money. Well, it's a lot more likely that the user will
a) Switch to satellite, DSL, or any other competitor he can find,
b) Write his city councilman about turning the city into a hotspot (some small cities already are)
c) write his congressman about regulating cable prices.

So the *best* likely scenario is that TW loses a customer. Other possibilities include losing a whole city of customers, or being turned into a regulated utility like phone and electric service used to be (and natural gas still is). That's a nice niche, you always make a guaranteed profit -- but your profits are pretty strictly limited.

I hope TW and Comcast wise up, because frankly I *like* the current (nearly) unregulated market. But if they keep trying stunts like this, they will feel like Wile E. Coyote when the anvil falls on him.

I can hear Brett and others asking, so what _should_ we do when we have one customer who uses 100 times as much as our average. Answer: graceful degradation. Slow him down. This can be done manually -- you notice that user X is using more than most, so you re-program his cable modem to a lower bit rate, and/or drop some of the packets destined for him. Or you can just make it automatic - the system notices when a customer is using a large amount of bandwidth over a period of time, and takes steps.

And as I said elsewhere, I would favor allowing ISPs to modify the TCP headers to reduce the window, even though strictly speaking it's a violation of the protocol -- that header is supposed to be for communication between the end-points. But until we get protocol stacks that actually _pay attention_ to Source Quench, I think this would be a good way to limit bandwidth consumption.