RE: [humanmarkup-comment] As If Article and thoughts about attrac tingattention to our cause

["Bullard, Claude L (Len)" <clbullar@ingr.com>]

That makes sense.  Remember that a lot of the negative 
reactions are based on systems that profile, that gather 
personal information, etc.  Because we separate concepts 
of Human and human and make it explicit that HumanML 
characteristics are not necessarily personal, one can 
use it to create profiles of cultural members without 
ever pointing to an individual.  That should be clear. 
It is how most behavioral systems for establishing 
patterns of behavior (eg, link analysis) work anyway. 
Those who don't like or want HumanML based on that 
paranoia should be more aware of how that is done 
already.  Behavior systems as such have been used 
in public safety since the 1960s.  To couple to  
identity systems, biometrics, not culture, is used. 
Cultural systems (eg, gang modules) are abundant 
for detection of early emergence.

What one should be clear about is that HumanML, per 
se, is not an identity system.  It can be applied to 
profiling but that is not something unique to HumanML.

The problems of online identity are enormous.  XML.COM 
has an article on that this week which I recommend to 
this group.   These are definitely issues to be resolved 
elsewhere, but in practical terms, a shared unique identifier 
is all any two systems need for loose coupling.

len

-----Original Message-----
From: Rex Brooks [mailto:rexb@starbourne.com]
Sent: Thursday, April 18, 2002 10:40 AM
To: Bullard, Claude L (Len); humanmarkup-comment@lists.oasis-open.org
Subject: RE: [humanmarkup-comment] As If Article and thoughts about
attrac ting attention to our cause


Hi Len,

I don't think we should make any attempt to interpose HumanML as an 
identification verification or authentication tool at the base level, but 
we should allow for secondary enhancements, such as individual profile 
information under the individual's control, for extra depth of 
identification for specific purposes--which would be determined by 
application authors.

What that means for the base schema is that it should have a mechanism for 
tying into whatever authentication/certification systems become widely 
used. I would say that this is covered in the requirements already and 
would not need a special addition to the HM.Requirements document at this 
point, although later on, when the arena shakes out some winners in the 
horse race currently playing out, we may want to revisit the requirements 
with specific language for support of the dominant id systems.

I think what you already have in the toolkit may fit the bill adequately, 
although I haven't gotten the time to go back and look since I went 
straight from building a Proposal to use HumanML in a Joint Project between 
Humanmarkup.org, Inc and the company where Rob Nixon works for DARPA STTR 
program to these meetings of the WSIA TC. I wil try to probe deeper over 
the weekend to make certain that our requirements are covered.

There are a wealth of applications which will be using individual profiles, 
which is the primary reason I am even in this TC, so I can ensure that our 
preferences profiles and voluntary behavioral profiles can be immediately 
incorporated into the Web Services that get codified here. (Yes, I know 
this is a misnomer.)

Heads Up Alert: I have already coined the phrase "Multi-Model Transport 
Protocol" and used it as possible solution for a small part of the proposal 
we just finished. It calls for a new protocol incorporating some 
functionality of http and smtp or messaging--a piece of the puzzle that 
everyone dances around but doesn't want to face. We have to do something 
for intermittency, asynchronous messaging that completes interrupted 
session transactions, allows for persistence, and has security--one way or 
another.

Ciao,
Rex

At 10:06 AM 4/18/2002 -0500, Bullard, Claude L (Len) wrote:
>What are the positions of contributors on the question
>of whether or not HumanML applications should provide
>"identity", that is, directly support identification
>processes for real human users?
>
>It seems to me that it should not... directly.  I suspect
>a lot of the bewilderment over HumanML and much of the
>consternation starts there.  If we eliminate that,
>of the proposed secondary applications, which would
>be affected?
>
>len




----------------------------------------------------------------
To subscribe or unsubscribe from this elist use the subscription
manager: <http://lists.oasis-open.org/ob/adm.pl>