RE: [huml-comment] PC-33 -Section 4.4.6-race

[Rex Brooks <rexb@starbourne.com>]

At 8:46 AM -0800 11/29/02, Dennis E. Hamilton wrote:
>Rex,
>
>My apologies for not following up sooner, as promised.
>
>I will provide closer attention to your careful and gracious comments today,
>now that I have subscribed to the comments list.
>
>Meanwhile, I wanted to make some quick clarification.  (See the <ORCNOTE ...
>..> pseudo-markup.
>
>-- Dennis
>
>Dennis E. Hamilton
>AIIM DMware Technical Coordinator
>------------------
>mailto:dennis.hamilton@acm.org  tel. +1-206-932-6970
>http://DMware.info/             cel. +1-206-779-9430
>      ODMA Support http://ODMA.info/
>
>-----Original Message-----
>From: Rex Brooks [mailto:rexb@starbourne.com]
>Sent: Friday, November 29, 2002 06:58
>To: humanmarkup-comment@lists.oasis-open.org;
>humanmarkup@lists.oasis-open.org
>Subject: [huml-comment] PC-33 -Section 4.4.6-race
>[ ... ]
>
>However, before I focus on race, I want to answer the observation that the
>alleged conceit/illusion of saying that: "we require better fundamental
>descriptions that take nothing for granted," is at best ingenuous.
>
>This objectionable phrase occurs in Section 4.2.13 in the textual
>description of HumanGroup following upon the description of Human. The
>objection seems to imply that there should be criteria for establishing who
>or what qualifies as Human and thence HumanGroup (being two or more Humans
>gathered in an environment by our definition). This is at best ingenuous?
>
><ORCNOTE I think I've thrown you a curve.  I don't object to not taking
>something for granted.  I suggest that the specification fails to do that.
>It may be that it is impossible to do that, and I recommend that prospect be
>explored. (I looked up ingenuous because I wanted to be careful here.
>Haven't found a better word yet.  I don't mean it to be pejorative.)  />
>
>My answer is simply that is necessarily so. That is quite different from
>assuming that it is possible to "take nothing for granted." It is stated as
>a requirement that we adopt no a priori exclusionary conditions. To think
>otherwise would be to hypothesize that some acceptable test could be
>constructed to establish the qualification of an identity of Human and then
>of HumanGroup.
>
><ORCNOTE I missed this in the requirements.  Where does it state that? />

It isn't stated in the requirements. It is in the textual description 
of Section 4.2.13 in the Word, HTML or PDF Specification Documents, 
not in the xsd code. You quoted it. It might be a good idea to 
formally add it to the Requirements. In the Primary Base 
Schema/Specification we are not, for the most part enumerating 
qualifications

>Anyone who wishes to do so is certainly free to attempt that. If Dennis or
>anyone else returns with a suitable test in hand, we will give up our
>naivete and impose it, if it is shown to achieve adequate reliability and
>can be proven NOT to disallow any valid Human, or Human-authorized,
>interaction-capable software agent. THAT is an important part of what we
>mean by taking nothing for granted. The criticism implies that we somehow
>SHOULD take something for granted (in this case the universal reliability
>and applicability of some test). I submit that this tendency to make such
>assumptions (not this particular assumption) is exactly the root cause of
>much miscommunication and misunderstanding. That people do take too much for
>granted is exactly why we must do our utmost to avoid that.
>
><ORCNOTE OK, but it bothers me that there is even the notion of "valid
>HUMAN" here.  What can that possibly mean? The very phrase suggests some
>position (e.g., criteria, expressed or not).  More to the point, collapsing
>characteristics of a person or HUMAN with human-authorized and with
>non-human "characters" that somehow present human attributes strikes me as a
>universalization that is not likely to enhance fidelity in communication.
>/>
For us it just means self-assertion, or 
authentication/authorization/certification if the entity asserting 
its identity as human in an application that uses HumanML chooses to 
cite that identity.

In the context of my statement above, it means commonsensically valid 
in the sense of an actual living human computer user who chooses not 
to use any identity verification and may also choose not to offer any 
information other than asserting that the person is human, but is 
demonstrably human, or a software agent authorized by such an person 
to act as an agent of that person. It was not meant to be formally 
testable qualification, but it could be. We are actually trying to 
avoid that whole issue.
>
>race-
>
>As I said in my reply to Dennis, we included race because the term is used
>in the arena of public safety and, I will add now, law enforcement.
>
><ORCNOTE Precisely.  This is a great example. So race is attributed to
>persons in a law-enforcement context.  It shows up on firms usually filled
>in by third parties.  As a crime victim I have been asked the race of the
>perpetrator.  I get that it is used.  That is a *highly* contextual case of
>attribution of a quality that is specific to culture, practice, i.e., it is
>very situated.  To abstract this into the specification as a HUMAN trait
>strips all of that out the way it is in the specification is not going to
>improve matters. The problem is the generalization.  I sense that it is not
>your intention to be doing that.  I suggest that it is in fact what has been
>accomplished.  It is not fair to say that someone should go to the list
>discussions to know the context of this.  That is not how the specification
>will be used.  It should not be necessary to have written it to be able to
>apply it with some reliability.  Maybe the question should be how are the
>contexts of Human ML instances, uh a "human" occurrence, to be established
>and misattributions to be avoided? />
>
>[ ... ]

Good question.

>We did not include it for any reason of our own, but because we are
>attempting to deal with objective world the way the objective world is,
>rather than how we would prefer it to be. In this case, the term and the
>associated concept of racism, as discredited as they are, are also
>indisputably operant in our human world.
>
><ORCNOTE Oh boy, the objective world.  Operant in our human world I will
>grant you. />
>
>I don't think we can claim any measure of objectivity if we start to exclude
>terms of which we do not approve. [ ... ]
>
>[ ... ]
>
>I would appreciate it if someone would cast this as a motion to remove race
>from the attributeGroup physicalDescriptors.
>
><ORCNOTE You said something here about wanting to honor how people describe
>themselves (or are described).  That sounds like a perfectly straightforward
>thing to do.  Is that indeed the subject matter to be embraced by Human ML?

That's what I'm trying to do.

>That's not how it is presented.  Maybe the problem is that the scope and
>reach is stated too broadly for something that is actually relatively
>straightforward when the markup is made more situated.  I think that
>otherwise you are trying to serve too many masters, and too many agendas,
>and it will diminish the accomplishment.  I would start by coming up with a
>better term than Human Markup Language.  Also, the race term is simply a
>dramatic example.  Hair color is enough of a problem, but it is more
>difficult to recognize.  Height and weight are also difficult.  Think about
>it.  When and where and determined by whom?  There's a big difference
>between a witness's description taken in a police report and a measurement
>at the pediatricians!  By the way, are you going to tolerate non-English
>descriptive terms? The Anglo centrism is pretty pronounced. />

The anglo-centrism has been noted and admitted. I'm not going to make 
myself crazy about it by trying to be something other than what I am. 
This is what the group process is for, to note it and amend it. I 
don't mind amending this one bit. As for the other items, we have 
adopted a process of developing the language in incremental modular 
fashion, based on a Primary Base which will be enlarged through 
extension and restriction in xml schema terms in a Secondary Base 
which will then spawn more application-centric modules. As more base 
terms are iidentified as needing to be added, or our current ones 
show a need for amending, we are committed to doing that, which is 
part of the next round of work on the Requirements Document. We are 
also working on a parallel structure in RDF as well to connect up the 
resources so that application authors will have an expandable toolkit 
for making HumanML apps.

><ORCNOTE Real world example.  My niece is a single mother (presently engaged
>to be married -- are you guys going to embrace kinship relationships?  I
>forgot to look) and on the usual questionnaires, such as the one for the
>just-past US Census, she describes her son's race as OTHER and bi-racial.
>That's how she describes it.  That is not likely to be what a police officer
>or a school official writes in a report, when referring to this same
>teen-ager.  It is what my grand-nephew says, as his mother has taught him.
>How does Human ML apply to this?  />

That is more the arena of application builders. If they need more 
from us, we are committed to being ready and willing to do that. We 
have a deliberately broad scope which is intended to be narrowed by 
the use cases that develop. Trying to anticipate all uses would be 
crazy-making, but being dedicated to encouraging scholars and authors 
and software engineers from humanity's various cultures to write the 
descriptions of their cultures and for individuals to describe 
themselves is one of our guiding principles.



>Ciao,
>Rex
>
>
>--
I look forward to hearing more from you. I will respond as I can.

Ciao,
Rex
-- 
Rex Brooks
Starbourne Communications Design
1361-A Addison, Berkeley, CA 94702 *510-849-2309
http://www.starbourne.com * rexb@starbourne.com