Melinda Shore <mshore@cisco.com> writes:
> > Does this seem like a weird position for an IAB member to take?
> > I don't think so.
>
> I think economics provides useful tools for talking about
> and evaluating this stuff, too, but I think it's pretty
> evident that you can optimize for anything you like and get
> different results. I question whether it's in this
> organization's charter to privilege the individual user over
> the good of the network. If you choose to put yourself
> behind a NAT that's possibly good for you (although I think
> it's bad for you over the longer term) and always bad for
> the people who want to reach you.
Well, my view, as I think I expressed on P-S, is that the
"good of the network" is some function of the individual
goods of the people on it, so I don't think of these as two
separate things, really. Obviously, one could take a less
utilitarian view.
> > There certainly are cases where it's appropriate for the IETF to say
> > that something users want to do is not OK. Most of those cases are
> > ones where their behavior has negative external effects on everyone
> > else. I don't think a strong argument has been made that this is
> > such a case.
>
> That last sentence left me speechless, for which I suppose
> some number of people are now in your debt.
Heh.
I think we're talking differently about externalities, which,
as you say, is always a problem with this sort of utility
analysis. Realistically, there are three kinds of utility
effects of someone choosing to install a NAT:
(1) The effect on them personally.
(2) The effect on other people who might potentially correspond
with them (a rather small set).
(3) The effect on the network as a whole, or to speak more
precisely, the effect on a large set of people who
have no relationship with the individual in question.
When I said "no strong argument" I was thinking about class (3), not
class (2), which I agree there is a much stronger argument to be had
about. I don't know of a strong argument for (3).
It's pretty clear that we have a mandate to deal with effects of type
(3). So, for instance, we go to a lot of effort to block things like
congestion collapse and DoS amplification, which impinge on all sorts
of innocent people. On the other hand, we mostly keep our hands out of
type (1), though not entirely. If people want to use RTP to listen to
Britney Spears, it doesn't harm anyone but themselves and we don't
care about that at all.
Class (2) is trickier to know what to do, since on the one hand we
wouldn't try to force people to avail themselves of new services. On
the other, to the extent to which NAT encourages people not to make
themselves available, that does decrease welfare. As you say, these
costs are difficult to account for, and I had forgotten to account
for them.
Assuming for the moment that those costs exceed the benefits, then we
would be in quite a difficult situation, since we've just described a
classic market failure where negative externalities aren't built into
price. But the problem isn't that people are stupid but rather that
they're rational. Getting them to do the efficient thing in such
situations is generally not easy.
-Ekr
--
[Eric Rescorla ekr@rtfm.com]
http://www.rtfm.com/