The Great Exchange
shields at crosslink.net
Fri May 29 22:52:47 UTC 1998
> Michael Shields writes:
> > I think pricing based on the actual destination makeup of your
> > traffic is more fair than pricing based on the assumption that your
> > traffic is like everyone else's.
> So, this is an issue of perceived morality and not a technical issue?
> I think its only fair that everyone pay for the matchbooks they
> use. The fact that matchbooks are generally given away these days,
> instead of having people charged for them on a usage basis, is grossly
> I think it is only fair that everyone pay for the television they
> watch on a metered basis. If you watch for ten hours, you should be
> paying twice as much as for five hours, right? Its only FAIR!!!!!
> I think it is only fair that you should pay twice as much for a Fedex
> package going twice the distance, and I'm MAD THAT THEY DON'T CHARGE
> THAT WAY, DAMNIT!
> Now that we've all figured out how silly this sounds... Mr. Shields,
> "Fair" has nothing to do with it.
What the hell kind of straw man is this?
Yes, it sounds absolutely foolish, but you are attacking an argument I
did not make. Perry, I am *not* a flake. I don't think that
distance-insensitive pricing is immoral; or that it is going away; or
that backbones should adopt it; or even that I know or care what sort
of pricing models people have. All I said was that distance-sensitive
pricing is feasible.
What place it has in the market, if any, I don't know, and it is
absolutely 100% pointless to argue about it. You think it has no
place in the market, and I think it could be worthwhile for some
niches. Neither of our opinions matter; the market will figure it
out. I don't care to argue about it. It doesn't matter. The market
will figure it out.
By "fair" I don't mean "moral", I mean "pricing more accurately
reflecting costs". I am sorry if you see this as a loaded term.
In fact, "fair" is exactly the word used by connect.com.au, the first
Australian ISP whose web page I looked at, to decribe their three-
tiered pricing model -- intra-network, web cached, and external.
This sounds to me like a coarse attempt at distance-sensitive pricing
for an area of the world where bandwidth *is* a major cost.
So, is it technically viable or not? I posted a method that I think
will work, using existing netflow data, i.e. no significant extra
router load. Is something wrong with it?
More information about the NANOG