[e2e] It's all my fault
tvest at pch.net
Fri May 18 07:02:37 PDT 2007
On May 18, 2007, at 8:09 AM, Vadim Antonov wrote:
> On Fri, 18 May 2007, Tom Vest wrote:
>> Your willingness to dismiss the net effects of such considerations so
>> casually is itself a textbook example of the "What is seen and
>> what is
>> not seen" fallacy.
> Tom - you may want to look up what this term means (hint: this is a
> of an essay by Bastiat). You certainly make no sense applying it in
> My statement was merely a very simple observation that averaging,
> comparing, or otherwise aggregating non-additive quantities (such as
> personal utilities) is mathematically invalid. Meaning that all
> models based on this operation are pure unadulterated junk. From a
> logical point of view, without any political bias.
> The non-additivity of utilities is easily demonstrated with a simple
> experiment: let's say I hit you on the head with a cluebat. Does our
> aggregate utility (my increasing because of moral satisfaction, yours
> decreasing because I have a heavy hand and extensive experience with
> spanking and striking implements) increase or decrease? What is the
> measure of "social good" of this act? How does one determine it?
Thank you for making my previous, tongue-in-cheek hypothetical
example so apropos.
It helps my always-vulnerable reputation for prescience immensely ;-)
I agree with you that abstract utilities are silly, and that much in
mainstream academic economics (and many other orthodoxies) is silly.
I'm not the kind of orthodox economist (or anything else) that feels
the need to defend the discipline, right or wrong.
I guess that's why blanket assertions from any particular ideological
orthodoxy tend to set me off. For that I apologize.
In this particular case, however, I do have something that economists
lack: an empirically measurable unidemensional metric of output or
value that I can use to benchmark the aggregate results of all of the
things we're discussing. That metric is unique public routed IP
addresses, some of which are multiplexed at the edges with NAT, etc.
It's a meaningful scale, I believe, because it actually follows (both
historically and longitudinally) the distribution of "addressable
network resources" (users, devices, etc.) that are counted
independently by institutions like the ITU -- provided of course one
knows how to apply the kind of (historically evolving) conversion
rules that IP analysts in RIRs and hostmasters within commercial IP
networks actually use to do that conversion in real-world practice.
Granted, it's a tricky metric, and the continued use of pre-RIR
address resources makes it tricker, but the latter represent less
than 20% of what's in production now, and about 70 or so countries
only came online after the establishment of the RIRs (and Route
Views) so there are some questions that can be addressed with
relatively high confidence -- or at least with much higher confidence
than can be claimed in any other meta-policy-oriented Q&A context.
Being an Internet booster, I base my judgments on better vs. worse
aggregate arrangements and outcomes on what gets more Internet "out
there" faster, all things remaining equal. In this case, more ASes --
which means more "independent" utilization of network facilities --
which in the real world means greater availability/affordability of
fractional network "infrastructure" (i.e., circuits over a fiber, or
two fibers in a bundle, but not necessarily the whole trench,
conduit, and everything else) -- tends to correlate with more/faster
Internet stuff over time, all things remaining equal.
It goes without saying that "more Internet" is not a universal
absolute good. Perhaps there are countries where more Internet has
actually been harmful, or come at some price which is too high by
some other metric. I concede that my approach says nothing about such
matters. The idea that the regulations that gave rise to rapid
Internet development did violence to some abstract ideology about
pure markets and property rights won't find much purchase with me,
but I'd be happy to hear about real-world cases that others think
might exemplify the hypothetical "too much Internet" outcome...
More information about the end2end-interest