What I find most interesting about all these disputes is that none of the major
players appear to be even slightly interested in settled interconnection - like
they have in the telephony world.
What I find interesting is the telephone world is now trying to end
settlements, and move more to the Internet model. The end of reciprocal
compensation, FCC forcing international settlements down, forcing access
fees down, etc.
There was a paper I saw ages ago:
that discussed some of the economics behind peering. Though it's a bit
old now, many of it's arguments hold true (hence todays CW/PSI/MFN/UUNet
conundrums). It also just begins to explore the possiblities (and
difficulties) of billaterally settled peering.
Actually I sent e-mail to Scott and Maria yesterday asking about their
paper. If they felt it still held true, or if the passage of time altered
I remember there was also an economist from Michigan who wrote several
papers about settlements and the Internet. I seem to recall he recently
(i.e. last year) spoke about the fact the Internet wasn't behaving as
some predicted. Instead of writing papers why the Internet must adopt
settlements, I believe some economists are now writing papers why the
Internet is still working without them.
Settlements seem to work best when you have geographic monopolies or
dupolies (e.g. regional bell telephone companies, regional pipelines,
regional electric grids, regional train companies). And instead of
building out your own infrastructure into a competitors region, you
split the market into semi-exclusive franchises.
Settlements don't seem to work when you have lots of competitors in
the same market (e.g. airport rent-a-car companies). You don't see Hertz
and Avis doing settlements and sharing the same fleet of cars depending
who rents more cars each day.
If we get to the point there are only 2-3 major ISPs in each geographic
region of the globe, I would expect we would see more settlements (e.g.
UUNET pays Telstra to delivery traffic in Australia, while Telstra pays UUNET
to delivery traffic in the USA) based on some imaginary metric that is
less than the cost of either of them invading the other's marketing
region. Much like the RBOCs have split up the USA into SBC, Verizon,
Qwest and Bellsouth.