[arin-ppml] The root of the disagreement?

Tom Vest tvest at pch.net
Fri Jun 27 17:12:03 EDT 2008

On Jun 27, 2008, at 5:56 AM, Milton L Mueller wrote:

>> -----Original Message-----
>> Was just reading back through Ruling the Root, when I (re)discovered
>> your overview on methods for allocating goods.
>> Very illuminating -- truly. Is there anyplace where those who do not
>> have the book could peruse your chapter on "The Political Economy of
>> Identifiers" online?
> Not at the moment. Might have time to make it available when returning
> to US.
>> Do you think that IP addresses are, or should have, the same
> semantics-
>> based value that domain names currently have?
> Of course, IP addresses do not have semantics, which moots the  
> question
> whether they should have them.

In a narrow (but instructive!) sense you may be right, as an artifact  
of a terminological error on my part (apparently an old and pervasive  
one, even in protocol design). In isolation, individual IP addresses  
have no function, meaning, or use value whatsoever -- by design -- so  
the term "semantics" has no relevant application. This is actually  
another foundation of "needs-based" conservation policies: it makes  
sense to preserve a finite and (truly) scarce resource of this kind by  
establishing basic eligibility rules for would-be recipients, rules  
that simply (and only) require that the recipient is capable of using  
the resource for its intended purpose -- the purpose for which it was  

Of course, individual IP addresses do have value, and *differential*  
value, in a context/environment of other IP addresses -- which means  
that the right term is not "semantics" but rather "pragmatics". Today,  
individual IP addresses bear important "pragmatic" distinctions,  
distinctions which are widely recognized both by the users (and  
devices) that bear them, and by every other user (and device) that  
also uses, or is capable of using IP addresses.

Once, a long time ago, there were no "pragmatic" distinctions either;  
all IP addresses were uniformly public (i.e., presumed to be global- 
scope unique, with global-scope use value) and uniformly interoperable  
IPv4 addresses. However, when it was recognized that IPv6 would not be  
ready for global deployment in time, RFC1918 established a new class  
of IP addresses with different, much narrower "pragmatics" and  
correspondingly lower use value. The pragmatic distinction was so  
great in fact that an industry friend used to give regular talks  
describing the (often involuntary) ad hoc partitioning that resulted  
from their use as "evil".* But we all learned to live with it because  
the involuntary aspect of it, at least, was expected to be eliminated  
by the transition to IPv6.

However, because the old and new addressing systems are not directly  
interoperable, the "pragmatics" of IPv6 will continue to have some  
critical features in common with those of RFC1918 addresses, for as  
long as the overwhelming majority of real online resources (users,  
content. etc.) are attached via IPv4. Native IPv6 users may be able to  
seamlessly communicate with each other when they are directly  
connected to each other, but they cannot communicate with each other  
over any IPv4-mediated distance -- or with any of the universe of IPv4- 
numbered edge  resources -- unless they also happen to have access to  
some IPv4 address space.

So, just by virtue of this environmental or demographic fact alone,  
the "pragmatics" of IPv6 will compare unfavorably to those of IPv4 for  
as long as the environment is largely IPv4-based. IPv4 users know it,  
IPv6 users know it -- and that knowledge unavoidably colors everything  
they do, today (like postpone adoption) and in the future.

That's why decentralizing IPv4 allocation and subjecting the process  
to market forces is such a bad idea at this moment. Absent other  
policies (e.g., 2008-5), resource transfer proposals envision a world  
in which those IPv4 address are available only from incumbents, who  
are often competitors, and who will always be interested foremost in  
maximizing rents from those IPv4 addresses.

Like it or not, we have arrived at a point where growth is causing the  
"pragmatics" of the public Internet to change. We could resign  
ourselves to this change, and a future in which direst stakeholders  
form a semi-permanent caste system of first-class and second-class  
participants (in which some may occasionally buy upward mobility or  
sell themselves "down"), that one day might be redeemed by the magic  
of the Invisible Hand. Or we can can take steps to recover the  
original design intent, by restoring the semantic/pragmatic opacity of  
IP addressing.

>> Would you argue that the
>> value of IP addresses, and and the overall system that they support,
>> would be more valuable if, instead of being unique but only
>> unidimensionally so (i.e., globally homogeneous in all other
>> nontrivial respects), IPv4 addresses were characterized by similar
>> hierarchies of use value (memorableness, etc.) and exchange value ($$
>> $) that is characteristic of individual domain names?
> No, I think that whole chapter was predicated on the contrast between
> meaningless unique identifiers that serve a technical function (IP
> addresses) and semantically meaningful unique identifiers, where the
> semantics bring conflicts over property rights related to trademark,
> personal names, company names, geographic indicators, etc. (we are
> _still_ working those out, by the way - witness the IDN ccTLD
> discussions in Paris at the ICANN meeting).
>> Since domain names are used for human, intentional/purpose-directed
>> navigation, I can understand the argument, even the necessity, for
>> heterogeneity across those identifiers. However, since IP addresses
>> are used for automated, dynamic, algorithmic navigation by machines
>> (which are operated by humans who also/already have access to DNS and
>> other, even more reliable semantic guideposts), I'm assuming that
>> there must be some *other* rationale for thinking that they should be
>> evaluated on the same terms...?
> Yes, their scarcity value, the need for rationing.

Just restating the terms "scarcity value" and "rationing" does not  
answer the question.
The question is, given a choice, and the likelihood of having to make  
some short-term tradeoffs either way, which would you prefer -- which  
would you personally recommend to the world?

1. "Rationing" an extremely scarce resource, the few remaining units  
of which have both non-substitutable, global-scope "use value" (albeit  
for a closed/static world) and extremely high "scarcity value" (aka $$ 
$), using decentralized market mechanisms that can only produce a few  
winners and many many losers.

2. "Rationing" an extremely abundant non-substitutable resource that  
has no "scarcity value" at all, but the same global-scope "use  
value" (albeit in a open/expanding world), if necessary using  
centralized administrative mechanisms (aka verification of "need", or  
ability to use as intended) that produces many winners and no losers  
at all.

It's okay if you don't want to answer, but we have to choose one.  
That's what this policy forum is for.


*The friend is sensitive about being misquoted so here are some  
references to speak for themselves:

(indirectly related)

More information about the ARIN-PPML mailing list