Scaling ARIN proposal for small ISPs - and economic reality

[David Hakala]

-- [ From: David Hakala * EMC.Ver #2.5.02 ] --

> I certainly agree that keeping the cost of Internet access affordable  is
a
> worthwhile goal.  But I think this metaphor causes some  confusion between
the
> cost of DOING business and the cost to the  business's CUSTOMER.

OK, I'll grant you that a car dealership costs more than a car. What happens
in an industry when barriers to entry as a seller are artificially raised?
Competition decreases and consumer prices increase.

> To keep it in perspective, that $2500 is the cost to be eligible to get a
PI
> /24-/19 block from ARIN.

> As has been noted on the list, the fee scale refers to membership in  the
> registry at various levels.  It is not a cost per /24 (or /19 or  whatever
).

Your preceding two statements seem to contradict each other. The small-to-XL
fee scale most certainly *is* the price of getting and keeping  IP addresses
. Quoting ARIN's Web page: "The annual ISP fee is based on the total
allocation of address space received in the previous year. First-time
allocations will be charged a fee based on the amount of address space
allocated." If you want X addresses, you pay $Y.

Oddly, end-users pay their registration fees only once. Why? Car dealers pay
annual business license fees, and car drivers pay annual ownership fees.
Under ARIN's model, annual ISP fees must be higher to offset revenues not
realized from end-users.

Then there's another $1000 per year to have a voice in the political process
. As many people have noted on this list, that's steep enough to exclude all
but medium-large corporate entities from the process.

Last but not least, the fee for an Autnomous System Number is only $500,
plus an unspecified annual maintenance fee. It's pretty obvious that ARIN
would very much like to encourage ASes. Not a bad idea at all.

> > Since public IP addresses are a finite resource, each additional address
you
> > want should cost *more* than the last one. That's how the real estate
market
> > works. As land becomes scarce, the price of a lot goes up. Why not
license
> > IP addresses one at a time for $10 each, 10 for $110, 50 for $1000, etc.
?

> Well, maybe you have a point.  But your point is antithetical to your
highway
> driving/low-cost-of-entry model.

No; my first car costs me only $10, but adding another car to a fleet of 100
would cost much more.

> To better state my original point, the ARIN proposal is just a way to
deal
> with a cost of doing business that has previously been funded by  the
federal
> government.

Ah, now we're moving towards a cost-based fee structure, rather than some
unexplained numbers. What has the government been spending on address space
allocation?

> It spreads that cost around among address  space holders who deal >
directly ARIN.  Yes, the cost will probably  trickle downstream to smaller
ISP's.

If the cost is not inflated by markup along the way, then I see no problem.
But that is not how multi-tiered distribution systems work.

> It's a legitimate cost of doing  business and those who aren't able to
meet it will simply, IMO, fall  into the category of insufficiently
capitalized operators.

I see no legitimacy in an arbitrary requirement that one license addresses
at a cost of at least $2500 per year. No one has revealed the cost of
allocating addresses, so any suggested price is suspect.

The proffered alternative - getting addresses from one's larger competitors
- is a prescription for anticompetitive behavior.


--
David Hakala
Editor In Chief
Cyber Week
dhakala at ossinc.net
303-755-6985