Small comments
All,
The
attached is a revised trust draft that incorporates pieces of Frank’s text and
items for continuing discussions. I specifically added a section on
trust related to a Delegate, and I included ideas Frank included in his
write-up. I specifically did not include Goal Adoption because I think it is
unnecessary and confusing as currently described; my response to Frank’s
Additional Comments email goes into my rationale. See some related
comments below.
Note,
I modified the Trust & Risk diagram (Figure 2) to include using Reputation
for assessing Trust and Risk because otherwise Reputation is simply defined
but not used.
In
specific response to Frank, see <KL> as follows:
<Frank_email>
Some
comments...
1. The diagram should say
that an actor participates in a joint action. You can’t participate in an
action; you perform one.
<KL>
Agree
and Figure 1 now contains two variants from which we can
choose.
</KL>
2.
The definition of willingness is somewhat circular.
Wikipedia
uses this definition:
Willingness:
the state of being willing.
Willing: Ready to
do something that is not (can't be expected as) a matter of
course.
Ready: Prepared for
immediate action or use
However,
I think that we mean something slightly different:
Willingness:
an internal commitment to participate in a joint action.
<KL>
The
Trusting Actor is willing to perform an action that is usually expected to be
part of a Joint Action, but this can happen independent of (and certainly,
prior to) Willingness on the part of other actors.
Still
needed: text elaborating relationship between Joint Action and
interaction.
</KL>
There
is another point that I think is important for SOA:
There
is a presumption of willingness based on participation in joint actions. The
fact that an actor participates in a joint action may be taken as evidence
that the actor was willing to do so. I.e., we do not try to model coercion in our model of
willingness. This evidence of willingness ultimately becomes the foundation
for non-repudiation: evidence of participation is evidence of willingness,
which in turn is evidence for non-repudiation.
<KL>
My
initial inclination was to agree with this, but I have reservations. Phishing
is an example where participation is not willingness in terms of the actual
RWE. Also, willingness is only explicit for RWE known to the Trusting Actor
and not necessarily for RWE that is not publicly documented or otherwise
known. Thus, the connection to non-repudiation is tenuous.
</KL>
3.
The Trust and Willingness diagram draws a trust relationship between actors. I
think that they are both inherently ternary relations: trust about some
action/outcome. I think that actors assess the evidence to determine their
stance to risk and trust.
We
have tended to focus Real World Effect on the effects of service actions: I
ask you to do something. But reputation may necessarily have a much broader
basis (hearsay, government intervention, etc.)
<KL>
Agree
and conveying that idea is certainly intended.
</KL>
4.
I do not think that we need to go into chains of trust. For the same reason
that we don't do much of service composition.
<KL>
That was
in your proposed text. I tried to incorporate this more fully in the
latest revision. See the comment that I included in the text. The
range of discussions and examples point to the need for three distinct
write-ups: (1) a concise model for trust, (2) a separate discussion of how
trust affects interaction, especially composite interactions, and (3) more
detailed discussions about the processes and mechanisms involved. (1) is
what I am working here, (2) should probably be an addition (if necessary,
after PR2) to the interaction write-up, and (3) are separate documents outside
the scope of the RA but hopefully elaborations that can build on the RA
foundation.
</KL>
5. Consequences
of Assessing Trust and Risk repeats earlier stuff.
<KL>
Moved
to section 3.x.x.1 and condensed.
</KL>
6. Trust and SOA: cut down
and move to the beginning.
<KL>
I
didn’t move this because I think it requires the discussion of trust and risk
before it makes sense. However, I’m open to specific
suggestions.
</KL>
</Frank_email>
Hopefully,
this will enable us to move forward.
Ken
---------------------------------------------------------------------------
Dr.
Kenneth Laskey
MITRE
Corporation, M/S
H305
phone: 703-983-7934
7515
Colshire
Drive
fax: 703-983-1379
McLean
VA 22102-7508