[wg-c] Re: MHSC Position Paper (unavailable), and your comments

[Eric Brunner <brunner@world.std.com>]

Continuing with the interlinear form for the last time.

>> >> Could you (in plain text form, please) expand on how Kent's
>> >> proposal is avoidably complex and over specified?
>> >
>> >As far as being overly complex, do a staffing analysis...
>>
>> Apparently you've already done a staffing analysis. Even if
>> it is a back of an envelop, please share the data.
>
>The analysis wasn't that detailed and certainly not in a spreadsheet where I
>consider it to be useful. The short answer is that it would take a sizable
>permanent staff to maintain and a fairly complex protocol between the
>registries.

The back of an envelope guesstimate would suffice, if your point can
move from proof by assertion to common quantification. Counter examples
to the assertion range from the historic SRI NIC, to modern ccTLD
registries, of course one could simply assert the contrary and leave
it at that.

>In turn, you also failed to respond to the point I made about Kent's
>solution assuming the institutionalization of the DNS. A point that I
>vehemently disagree with and a further argument against the plan. Kent's
>plan will simply not work in a non-institutionalized DNS world. It
>absolutely requires the bureaucracy to support it.

I hadn't realized you were asking a question. It appeared to simply be
a restatement of the question I asked you. You also made an assertion
which you then vehemently disagreed with, which perhaps I could respond
to if I knew what you mean by institions.

On to the issue of mutal secondaries and something else to be clarified
further below meet the goals of stability, etc. without the use of any
additional mechanisms... your claim.

>> You offered the remark that mutual secondaries (the second
>> part of your
>> remark is below) offer sufficient mechanisms to obviate the need for
>> those described in Kent Crispin's proposal.
>>
>> I asked (1st part)
>>
>> >> If secondary nameservers are a solution ... (consistency
>> of a cached
>> namespace and conflicting claims of authoritative nameservice)
>>
>> You replied:
>>
>> >Cache corruption is a configuration issue these days and not
>> germane ...
>
>I recognise that I was very terse and skipped a lot of steps. I assumed
>you'd keep up. In reference to my paper, I required that all registries make
>their parent registry one of their secondaries. This is done for a variety
>of reasons. It is also obvious that the registry would also provide it's own
>local root-server. In this way, all registries secondary for each other. The
>terse short-hand that I employed was not adequate to communicate the
>concept, I apologize.
>
>I only answered the point that I thought you were making about cache
>corruption. Which, in a properly configured system, is not an issue. After
>all ... you did ask and I answered.

I think I recognize a policy statement, something a little less insubstantial
than a claim of non-necessity with proof by assertion. 

	All registries make their parent a secondary

Please correct me if I'm wrong, but if you are referring to gTLDs in the root,
it looks as if the root "secondaries" all of the gTLDs affected by the policy
you suggest.
Next,

	 All registries secondary each other

What is the mechanism to ensure this policy is effected, and what trust
consequences do you suggest arise in consequence of such a policy?

Or,
	The existence of a local root-server in each registry, together
	with "all registries make their parent a secondary" has the
	effect of "all registries secondary each other"

I don't think the nuance is worth the bother. Your "registries" are simply
a variety of registrars, with some local cache semantics, and unnecessary
potential complications over title to the registry data.

>> I asked (2nd part)
>>
>> >> Do you expect Paul Vixie to endorse the claim that mutual
>> secondaries
>> >> solves the problem presented by two (or more) simultenious
>> assertions
>> >> to authoritative nameserver status? I don't, but I could be wrong.
>> >
>> >I don't expect Paul to endorse anything but strict
>> hierarchy, as he always
>> >has, religiously. I believe that, as he gets jerked around with
>> >f.root-servers.net, he will eventually come around to a
>> different view, but
>> >I do not presume to predict what that view might become. I
>> watch those
>> >developments with interest.
>>
>> Assuming the correctness of your assertion that cache errors has no
>> relationship to the issue of authoritative server determination in the
>> presence of two (or more) simultanious assertions by distinct entities
>> to authoritative status for a domain, if only for a moment,
>> how is your
>> answer responsive to the question of whether Paul Vixie would agree
>> with your original claim?
>
>There was a lot more implied than is apparent from a simple interpretation
>of your question. My response answered the deeper question that I thought
>you were asking.

I grant you answered a different question. I granted your premis to see
if you would offer an answer, in the belief that an answer consistent
with your premis was possible.

>> Does your meaning of "mutual secondaries" mean you wish to see WG-C
>> recommend a policy to ICANN which is inconsistent with the operational
>> practice of the top-level of the deployed system?
>
>I am not aware that what I suggest is contrary to current practice. That
>some of it is not currently being done is obvious. There is no current need.
>The idea here is to try and improve some things, or so I thought. There are
>many inadequacies in the current practice, RTFP.

You wrote "I watch those developments [f.root] with interest" and expressed
the belief that Paul's technical judgement and operational practice would
at some future time, be in agreement with your own, implicit in your own
proposal. What other impression did you hope to convay?

>> If you don't want to venture an opinion on whether or not Paul will
>> agree with the substance of your claim of sufficiency of mechanism,
>> or if you think Paul is the wrong person to ask, just say so. Either
>> answer will suffice.
>
>Sorry, I never presume to read Paul's mind. We simply don't know each other
>well enough. The deeper issue is that Paul is on-record as being against
>additional TLD not served from the root. He has cause for his beliefs which
>I will not dispute in a non-technical forum. I will say this, I have made
>modifications to BIND and have been testing them for some time. This gives
>me a fundimental understanding of the problem which few others have.
>Certainly beyond the ken of many lawyers. Partially, it is an issue of
>design philosophy and style, Paul and I differ. But, where Paul works within
>the IETF, I don't have either time nor inclination to do so. I'm too busy
>building tera-bit web-sites for a living and  having fun doing it <grin>.

I don't think this calls for mind reading, its a skill WG-C probably ought
not rely too heavily upon. What this does call for is accurate technical
judgement based upon sufficiency of experience. I'll see Paul in a few days
time and mention your ideas. Feel free to make your arguement as to the
stability issue, your proposal's strengths and Kent's proposal's weaknesses,
a stand-alone one-or-two pager, as it isn't the only thing on either of our
agendas. 

>> Next I asked a scaling question about your point that some
>> policy is not
>> required because "the customers [can] take over the registry". This is
>> the second half of your remark that the mechanisms described in Kent
>> Crispin's proposal are obviated by those you cited.
>>
>> I understand your reply to be that you are only considering
>> the case of
>> registry abandonment, where no substantive controversy could exist. If
>> so, then the issue is of little interest, whether Kent got it right or
>> wrong.
>
>I submit that it is of fundimental interest to the SLD owners of that TLD.
>In my paper I address the abandonment issue directly. This same case
>addresses the issue of the TLD registry losing their audit review.

Interest? Yes. Controlling, or even remotely related to stability (your
claim)? No. This was the second half of your assertion of non-necessity
of the Crispin position paper mechanisms.

>> If as you suggest, a staffing analysis yeilds a cause for concern with
>> the set of mechanisms described in Kent Crispin's position
>> paper, there
>> is not yet any substance to your claim that mutual secondaries and/or
>> the irrelevancy of registry abandonment are sufficient to support the
>> stated goals of that position paper -- stability, etc.
>
>Actually, the stability issue underlies the very fabric of my paper. I just
>don't presume to cast the Internet in institutional amber.

I want to return to your mention of "TLD not served from the root" (above a
few paras). Recently Brian Carpenter published a statement of the IAB on
the subject of the root. It is quite specific as to the IAB's preference
for and concern over the relationship between the root and the namespace.

If you (and this really is unclear as first you've all the TLDs stuffing
up, then daisy-chaining, and now here finally without reference to (served
from) the root) suggest that WG-C suggest this to ICANN, and WG-C and/or
ICANN accepts such a policy for TLD management, then ICANN and the IAB will
differ on an issue the IAB identifies as fundamental.

I suppose you don't cast the net in amber.

>> You did ask what legal history (pleadings) has to do with the issue.
>> Where the pleadings of both parties are in accord, there isn't any
>> issue or controversy. They are in accord on the point you
>> made (retained
>> above), and differ with your claim.
>>
>> I don't understand contradicting both DDN Management Bulletins and the
>> common legal statements on the issue you raised by both NSI
>> counsel and competent counsel for parties bringing claims against NSI.
>
>It is maybe because I don't understand how lawyer pleadings have anything to
>do with demonstrable marketing competencies and business skills.
>
>> I don't mind if you shift your emphasis, having misspoke.
>
>I didn't misspeak. Perhaps, you misunderstood.
>
>Those who can ... do (and get paid ... $well$).
>Those who can't do ... teach (at one-third the money).
>Those that can't even teach ... work in institutions (no pressure,
>guaranteed retirement).
>The rest become either lawyers, working-persons, or wine-o's.
>
>Gee, did that broad brush leave anyone out?

What it did leave out was acceptance of responsibility for having written
in error ("NSI was started by academics and scientists...") or an attempt
at proof.

Eric