If you point out that proposal itself, my answer will be
because "100% discount of PAF for all allocations" will be
simpler than "90% discount of PAF only for IPv4 infrastructure
based evaluation".
If you point out the long-term NIR fee change, it is not a
small amount and PAF no longer fits - for example a *single*
/21 IPv6 allocation introduced us PAF of USD64K where JPNIC's
total annual payment for APNIC was approx USD200K. This is
a huge difference either in case PAF would be born by the LIR
or by the NIR. PAF used to be modest enough to be fit. but
it isn't any longer.
Am I answering?
yes, at last! so the 'complexity' is really the size of the
number, not that the rirs can not calculate a simple step
function.
while i suspect i could sympathize with the costs when looked
at as absolute bhat/dollar/dong/yen amounts, US$64k ain't bad
for more address space than all of ipv4. especially as i
suspect an ipv4 /16 will be going for upwards of US$1m in a
few years.
but now that we're down to the money, what do you think a
reasonable curve, or step function, would be? it sounds as
if stephan (and others from whom i have email) are concerned
about how it goes forward. so discussing a replacement
proposal may help alleviate some of the disagreement. then
again, it may not; some of us are just disagreeable :-).
randy