Hi folks, I a m bit late to this discussion
but I have some 'meat' for Ken to thing about as well.
Frank is absolutely right IMO that P2 might not know about P1 intents. Even
more, P2, in presence of intermediary P3, might even not know about existence
of P1, the same relates to P1 with regard to P2. This concept of unknown
counterparty is used in other patterns such as Pub/Sub and Mediator, for
instance. I have two points to talk about in this context.
1. I find that analogous situation exists for service contract when service
consumer and service (service provider) communicate via infrastructure intermediary
like ESB. That is, since service contract is supposed a) to b set between
consumer and provider, and b) to contain agreed business functionality, RWE and
related policies, the intermediary must become transparent in the service
contract. If consumer has to invoke end-points that belong to the intermediary,
this communication information must be in the contract. Otherwise, when
consumer communicates with the intermediary that shields or hides the service,
the contract has to be set between the consumer and the intermediary. In the
latter case, the intermediary takes all business related information about the
hidden service and acts as service agent trying to solicit consumer's interest
and to 'sell' hidden service to the consumer.
Following this logic, I can say that the trust between consumer and service is substituted
by the trust between consumer and intermediary if the latter hides the service.
This trust is associated with the obligations taken by the intermediary on
itself with regard to providing for business functionality and RWE that
originally were offered by the hidden service. We cannot say 'on behalf of this
service' because such intermediary disallows the consumer to know the service.
To be fair, I have to say that not every intermediary hides service from
consumers. For example, an intermediary, which provides semantic translation
between consumer and service, acts 'in clear' and does not hide interaction
participants.
2. To Ken's note - if we consider the mode, which includes a service-hiding-intermediary,
we cannot say that P2 intent relates or caused by P1 intent, that 'the intent
of P2 will lead to effects that P1 would consider' because P1 has no clue about
P2, i.e. no trust is needed between them.
The entire situation with an intermediary becomes tricky in the area of the
service contract and linked consumer-service trust. I see the solution only in
the mutual multi-party trust model where P1, P3, and P2 are a) aware about all
others; b) have trust too all others thought particular trusts or degree of
trust may differ. For example, in the process of payment by a Debit Card, the
card holder trusts the merchant that it would take only agreed amount of money
and take care about the rest of the payment procedure; the merchant trusts the acquirer
that transacted payment would be passed to the card issuer reliably and in
timely manner, and that the acquirer would arrange the payment to appear on the
merchant account somehow; finally, the account trusts the card issuer that
transferred transaction would result in the deposit of specified amount of
money on specified account (that belongs to the merchant, which is unknown to
the card issuer).
I hope this will help Ken 'to rethink' trust related links and dependencies.
- Michael
----- Original Message -----
From: "Ken Laskey"
To: "Francis McCabe"
Cc: "OASIS-RA RA"
Subject: Re: [soa-rm-ra] my voiceover for trust slide 13
Date: Wed, 18 Feb 2009 17:45:28 -0500
I'm fine with most of this, including the two modes. My assertion is that in either mode, P2 has its own intent shaped by its own (or its domain's) goals, objectives, responsibilities, and constraints. Trust on the part of P1 is the attempt to accurately calibrate the likelihood that the intent of P2 will lead to effects that P1 would consider adequately furthering its intent. If P2 was a bonded courier, it would act in the ambassador role and its being bonded would lead to a high degree of trust that the package will be delivered to P3.
Now, having just written that, I would need to rethink where I am talking about confidence and how it relates to trust.
But not now :-)
Ken
On Feb 18, 2009, at 5:30 PM, Francis McCabe wrote:
The way that we ended up discussing this was that there seemed to be two distinct modes in which 'intent' is 'transmitted' to a third party: ambassador mode and facilitator mode.
In ambassador mode, the middle party's responsibility is limited to one of communication: ensuring that the intent of Participant 1 is faithfully communicated to P3. In such a scenario it is not even necessary for P2 to be aware of P1's intentions: its expression may be encrypted for example.
In facilitator mode, P2 adopts the goal for itself and communicates with P3 as needed in order to achieve the adopted goal. In this case, clearly, P2 must 'understand' in the intent in order to be able to adopt it. In addition, the interaction between P2 and P3 need not be directly about P1's intent but some derived portion of it.
Frank
On Feb 18, 2009, at 2:17 PM, Ken Laskey wrote:
While I don't disagree with the text on the slide, I would emphasize the following:
Each participant in an ownership domain has goals, objectives, responsibilities, and constraints that are shaped by that domain. When a participant in ownership domain 1 interacts with a particpant in ownership domain 2, the participant in 1 conveys the intent that motivates the interaction. The particpant in ownership domain 2 needs sufficient semantic engagement to understand the intent of participant 1. Also, the participants need (to a varying extent) to be able to establish the identity of the other particpant. Participant 2, upon receiving and, to some extent, understanding the intent of participant 1, creates his own intent that (1) reflects what was received and understood from participant 1 and (2) what can/should be done within the goals, objectives, responsibilities, and constraints of ownership domain 2. The alignment of the intents of the participants depends on the other items that went into forming the intent. Trust on the part of participant 1 for participant 2 is that participant 2 will do things that result in the intent of participant 1 being satisfied. The accountability on the part of participant 2 will determine whether the trust of participant 1 was met, not met, or possibly exceeeded. The assessment of accountability will likely affect trust in the future.
The diagram also shows a similar interaction between participants in ownership domains 2 and 3. The interesting case is when this interaction is needed in the context of something needed to support the participant 1/2 interaction. This may require an implicit or explicit delegation of intent and/or authority in the communications chain. Discussion of the chaining will be put off until a later time.
Ken
------------------------------------------------------------------------------------------
Ken Laskey
MITRE Corporation, M/S H305 phone: 703-983-7934
7515 Colshire Drive fax: 703-983-1379
McLean VA 22102-7508
------------------------------------------------------------------------------------------
Ken Laskey
MITRE Corporation, M/S H305 phone: 703-983-7934
7515 Colshire Drive fax: 703-983-1379
McLean VA 22102-7508