[ NNSquad ] Re: AT&T testing DSL bandwidth caps

[Barry Gold <bgold@matrix-consultants.com>]

Lauren Weinstein wrote:
>   [ Of course, caps carry a definite profit center bonus as
>     compared with throttling, at least when subscribers are being
>     charged extra for exceeding the caps (and especially when those
>     caps are relatively low).  That is, a circuit that is throttled
>     but not capped will never bring in additional revenue since
>     there's no cap to exceed (and so no "excess" data to charge
>     for). 
>     Note also that currently announced caps, even for similar
>     technology and speed circuits, are varying widely from different
>     companies.  This *suggests* that perhaps their levels are being
>     set arbitrarily, or at least that's a reasonable assumption in
>     the absence of public data to explain how these caps and related
>     charges actually relate to claims of "degrading" other
>     customers' traffic.

A more charitable interpretation: the ISPs have no idea what cap levels 
will be tolerated by consumers.  So they are all guessing -- and maybe 
even trying different values from what other providers have announced -- 
in hopes of finding out what the results are.

Too high a cap will mean that they get no benefit -- traffic continues 
at rates that the ISP cannot sustain in its current business model.  Too 
low a cap will send users running to FiOS or DSL as soon as they get 
their first bill with $500 in overage fees (and there's a good chance 
consumers will refuse to pay, in addition to switching).

As for those who will simply terminate customers who use "too much" 
bandwidth, one of my friends has a lovely gesture: sitting in a chair, 
he holds one foot out in front of him.  Then gripping an imaginary 
revolver in his hand, he takes careful aim.

But of course a monthly bandwidth cap has only an indirect effect on the 
ISP's biggest problem: use of a lot of bandwidth in a short period of 
time, whether by P2P or by users viewing video. It's true that somebody 
who does this _all the time_ will also run up against the caps.  But 
somebody who does it for a few hours will degrade service for others in 
the area (and/or force the ISP to buy more bandwidth from upstream 
providers -- if not already in a "no settlements" peering arrangement), 
but won't run up against the cap.  And somebody who does it late at 
night will run up against the cap even though they do not degrade other 
users' experience.

I suspect what's happening, though, is that the business model designed 
in the 90s is simply bursting at the seeas, and all the ISPs are going 
to have to tear their hair for a while until they find a new, 
sustainable business model.  Nothing surprising here.  I commented 
earlier that finding a business model that works for 5 years is a 
miracle in this era of rapid technological change.

   [ In a truly competitive ISP marketplace, such "experimentation"
     with subscribers as the guinea pigs might perhaps be argued as
     justifiable.  But in the current constrained marketplace for
     most U.S. Internet users (as far as reasonably-priced access
     services are concerned) it's far less tolerable.  This is but
     another factor pointing at the possible need for regulatory
     intervention in this situation.  It is unfair that customer A
     forced to buy Internet service from cable company Y in one area
     may have a 250 GB cap, while customer B getting the same
     speed service from company Z in another area is facing a 40 GB
     cap -- or even 5 GB.  These are enormous differences that
     *will* negatively affect a wide range of Internet
     applications, especially at the lower cap levels.  

           -- Lauren Weinstein
              NNSquad Moderator ]