Re: [apps-discuss] On "supporting the publication of this document"

[Nico Williams <[e-mail]>]

On Thu, May 12, 2011 at 10:52 AM, Dave CROCKER <[e-mail]> wrote:
> On 5/11/2011 3:07 PM, Nico Williams wrote:
>>
>> I worry that this approach will lead to an obnoxious level of
>> formalism
>
> The IETF has quite a bit of formalist, along with it's famously having quite
> a bit of informality.  When introducing rules, the challenge for us is to
> get it right.  That's why we need to be careful that our formality, here, is
> about the /nature/ of what is provided and not the form.  (We often get this
> confused, of course...)

Well, I prefer formalism for specifications, and informality for the
social aspects of standards-setting.  We could argue about both till
the cows come home.  As long it's personal preferences that we're
stating then there's no need to argue about these.  But regarding
bonafides establishment, I have fairly strong feelings here:

> For establishing credibility, the need is to say enough to show actual
> knowledge of the document, not to regurgitate a CV.

Sure, this may be _advisable_ for non-usual-suspects, and occasionally
advisable for usual suspects (namely: when necessary as a result of
participation by new participants).  (I don't really care for a lot of
tedious repetition.)  But I've grave reservations about this as a
formal part of argument at the IETF (see below).

>> Good technical opinions are good regardless of the speaker.
>
> Indeed they are, but providing enough information to know that they are good
> is the issue here.

When there's reason to doubt a statement, yes.  But it's not always
necessary.  For example, a person I rarely see posts from posted to
KRB-WG the other day to point out a security flaw in one PKINIT
variant.  The flaw was obvious once described.  There was no need for
the poster to describe their experience, nor should there have been.
You give an example where I can imagine that the poster's experience
will contribute weight to the statement, but the statement will need
to be explored in detail in such a case anyways, and I still wouldn't
really care what the poster's experience was once we're done
establishing the value of that statement:

> For example, compare:
>
>     This document specifies a technique that won't scale.
>
> versus:
>
>     I have worked on the design of a number of protocols that have become
> popular for wide-scale use on the Internet, namely xxx, yyy and zzz.  I've
> also worked on a number that failed to gain popularity, namely aaa and bbb.
>
>     Based on this experience I believe this document specifies a technique
> that won't scale.  In particular, the foo mechanism requires manual
> updating, which is extremely expensive and error prone.
I believe statements like the above can be made, and convincingly so,
without reference to personal experience.  Personal experience helps
illustrate the truth of such a statement, but the personal experience
can be distilled into a detailed rationale without having to describe
the actual personal experience.

I prefer dispassionate argument (though I admit that on occasion I
allow my passions to bleed into my arguments -- a personal failing).

Dispassionate argument, it seems to me, is best served by trying to
distill personal experience.  Dispassionate argument does not require
that one not present personal experience, of course, but I think it's
advisable to distill such experience as well, if not first.

> So, which bit of input would you expect to be more useful when making a
> decision about the documenet?

Neither.  I'd prefer:

    I believe the proposed design will not scale well for the
following reasons: <wisdom distilled from experience>.  [optional]
This comes from having observed similar failures in the field
<description of personal experience>.

Here the personal experience helps shed light on the rationale given,
but the rationale given ought to stand on its own.  Moreover, the
truth of such a statement might be obvious once the statement is made,
in which case the personal experience becomes irrelevant.  Moreover
still, if the truth of such a statement can easily be established, it
should not matter that the speaker has little or even no experience in
the field.

Put differently:  Argument by experience can be fallacious and/or
fraudulent at worst, often being just a point of view easily countered
by others' personal experience.  Then the degree to which the speaker
is a usual suspect or otherwise "important" becomes a key to giving
weight to their argument.  I don't think that would be a good thing,
particularly if we should *formalize* such a mode of argument (which
would make lambs of many participants, and wolfs of a few).

Nico
--