[ppml] FW: Legacy /24s

[Paul Vixie]

> > (the block was later used for a root name server, and then, later, given
> > into the control of a 501(c)(3) nonprofit who operates that server, and
> > for that, i think the community of router owners would say it's worth a
> > routing slot... i'm just saying if that hadn't happened, i'd've stopped
> > using the space by now, on a "think globally, act locally" basis.)

> That is silly. We are talking about ONE group and not MANY. The item is not
> even close to being the same.

the aphorism "thing globally, act locally" means trying to deduce the impact
of global behaviour that'd be reflective of local behaviour.  for example,
think of tossing a used styrofoam cup out of a car window.  if you're the only
one who acts this way, it would have no global impact.  but if you consider
what might happen to the world "if everybody behaved this way" -- all of us
waist deep in used styrofoam cups, etc -- then it becomes possible to respond
from principle: "somebody's got to be the first person to do the right thing,
if i want doing the right thing ought to become common, so, it might as well
be me."

or think of it another way.  right now about 30000 people have swamp /24's
who would likely not qualify for them under current rules.  from a policy
development standpoint, we can do only one of exactly two things:

	1. treat them as an exception, merely an early deployment necessity
	2. treat them as the rule, something to emulate in new policy

i choose #1.  so if you have activity that can't operate in PA space, like
an anycasted root name server, then you'd better make sure that it's an
activity that the global community of "router operators" think is worth their
expense in carrying a route for you.  most of the time, these days, the "value
proposition" is in terms of compelling content or massive numbers of eyeballs,
and so RIR policy tends to favour large blocks occuping routing table slots
vs. small blocks, unless the small block has compelling content that wouldn't
be as economically viable in PA space as in PI space.

so, we really are talking about many groups, not one.  or at least i am.

note: i'm not speaking as an ARIN trustee in this message.