RE: Comments on section 3

[Peter F Brown <peter@peterfbrown.com>]

Ken:
As discussed and agreed, I copy the TC list in this next round of
comments/replies that we have had over the last week.
I've also re-inserted the (hopefully) non-contentious exchanges we had and
changes we agreed in principle, so that the whole TC can see our
suggestions. I've marked those with **. My latest comments are marked
*[Peter].

All:
The line numbers refer to the Editors' Draft Clean version of 21 Dec, not
the main changes version.

Except for these changes/comments, I am working on a substantial redraft on
"service in the context of the SOA ecosystem", as requested by Boris,
Michael and Ken; and clarifying that "service != resource" which needs to be
stated more explicitly at different points.

We still have an open question about the names of the three
views/viewpoints. I put in placeholder names of "Ecosystem", "System" and
"Process" - although I know several TC members are not happy about those
titles (If we were to align with DoDAF, the last two could/should be
"System" and "Operations", but that is maybe another debate). We do need
clear and representative names, though and expect some feedback from the TC
and members on this.

We will still have to review all diagrams to align them consistently with
the textual changes: for some, it may be better to wait until the text
settles; for others, the redrafting can probably already be done.

I plan, with Chris, to do another iteration of the whole draft once we have
this round of comments and feedback concluded, hopefully following
tomorrow's meeting.

Best regards,
Peter

| -----Original Message-----
| From: Ken Laskey [mailto:klaskey@mitre.org]
| Sent: Thursday, 30 December 2010 06:50
| To: peter@peterfbrown.com
| Cc: 'Bashioum, Christopher D'
| Subject: RE: Comments on section 3
| 
| I've deleted comments where we have agreement and just kept the ones where
I
| have further comments.  We trip upon some thorny issues I had long
forgotten
| but there is definite progress.
| 
| Ken
| 
| |
| | Line 622: text implies single purpose but could be many, e.g. Life,
| liberty, and the
| | pursuit of happiness.
| [Peter] Don't confuse purpose with goals! ;-) Purpose should be singular.
| [Ken] Interesting that I never thought of purpose as singular.  Why is
that
| so?
| For reference, I went to dictionary.com
| - Purpose: the reason for which something exists or is done, made, used,
| etc.
| - Goal: the result or achievement toward which effort is directed; aim;
end.
| - Objective: something that one's efforts or actions are intended to
attain
| or accomplish; purpose; goal; target
| I think we are fairly clear on the RAF difference between goal and
| objective. We need to be clear how purpose relates, e.g. purpose is why
the
| social structure exists, goals are overall states the social structure
would
| like to achieve, and objectives are specifics the social structure feels
it
| needs to accomplish in order to realize its goals and satisfy its purpose.
*[Peter] I don't think we need to over-analyse the difference between
purpose and goals - it is the difference between them and objectives which
is crucial.
| 
| | Line 648: continuing with previous comment, replace "its purpose" with
| "defined
| | set of purposes".
| [Peter] "reflect a defined purpose"
| [Ken] either of these is fine once we resolve the singular question.
| 
| | Line 664: in definition of constitution, missing something about rules
may
| | change/evolve and constitution may provide rules for change. See section
| | 5.1.2.2, in particular, Governance Framework and Governance Processes.
Not
| | sure what is useful to have here or to point to that discussion, which
| develops in
| | much more detail.
| [Peter] reworded to "defines the purpose, goals, scope, and functioning of
a
| social structure"
| [Ken] I like adding that there are aspects of the social structure being
| defined but be careful of your use of purpose and goals.
*[Peter] OK, I'll look at this once more
|
|** Line 680: this is suggested reordering. Consider dropping lines 
| 680-682 to after line 693.
[Peter] ok

|** Line 715: add relationship Social Structure "defines" Role to model.
[Peter] Noted. I'll only redo the models when we've stabilised the texts
 
|** Line 732: given previous text, should we say "bestow or assume rights"?
[Peter] ok
 
|** Lines 744-746: confusing. Seems you are saying too many things in too 
| few words.
[Peter] to do
 
|** Lines 755-756: unclear what trying to say.
[Peter] to do
 
|** Line 757: new model needs a bit more explanation.
[Peter] agreed. Model itself still needs some work...
 
|** Lines 762-775: Qualification depends on circumstance.
| (1) Pilot of a small plane has a heart attack and non-qualified 
| passenger only one available.
| (2) Someone needed to manage a task and no one with background or 
| training available.
| (3) Return on political favor, so given the role for which doesn't 
| have skill Such caveats are important because otherwise the text 
| states all roles are filled by "optimally" qualified individuals.
[Peter] done
 | 
| | Line 803: "or components" is ambiguous because many use term but not
clear
| | what it is intended to mean here.  Just delete?
| [Peter] I agree with you: I have been arguing for some time about the need
| to think of a "service" as something with a single feature providing a
| single function and anything that offers more as being a "composite
| service". In this sense of these terms, "service" would be enough in the
| text. I'm swimming against the tide, however, given areas of work such as
| Service Component Architecture (SCA).Could we say something like "...and
use
| others services or specific features of composite services" - or is this
| unnecessarily adding a new concept or twist? Striking "or components"
would
| certainly be easier, and we can tackle the idea of single vs. composite
| services elsewhere...
| [Ken] I'd go for striking and tackling later.
*[Peter] OK, I think that's the best approach for here.
| 
| | Line 816: reword to "structure to which the company belongs"?
| [Peter] The original is clumsy but we have to be careful: we define
| elsewhere that a company *is* a social structure, too. So I suggest "...
as
| well as the wider social structure (legal, political, etc.) that the
company
| is part of."
| [Ken] We're talking two separate issues: (1) clumsiness, (2) company
related
| to structure.  How about "... as
| well as the wider social structure (legal, political, etc.) to which the
| company belongs."
*[Peter] Clumsiness of text is dealt with, I thin, but I still don't like
the formulation of the relationship between company and social structure - I
don't think it is either 'belongs to' or 'is part of'. Let me think a bit
more on this.
| 
| Lines 828-829: should this be part of 813-816?
[Peter] Indeed, good catch. Done
| 
| Lines 830-834: move to 812.
[Peter] Agreed
|
| Lines 854-855: is definition of Commitment needed or used? If kept, it 
| needs to include the idea of intent.
[Peter] I made the same comment in my notes. I'm in favour of dropping it,
it's used in a common-sense way.
|
| | Line 903: "Ownership is a relationship ..."
| [Peter] I've reworded this as:
| "Ownership is defined as a relationship between a stakeholder and a
| resource, where someone (performing a role of the owner) has certain
claims
| with respect to the resource".
| Does this work?
| [Ken] good enough
| 
| | Line 955: instead of "right", should it be obligation or responsibility?
| [Peter] Yup: we were always taught that you cannot delegate
responsibility,
| only work...;-)
| Does this work:
| "A stakeholder who owns a resource may delegate some or all of these
rights
| to others, but typically retains the responsibility to see that the
| delegated rights are exercised as intended.  There may also be joint
| ownership of a resource, where the responsibility is shared."
| [Ken] good enough
| 
| | Lines 957-961: is this redundant or implied with what already said?
| [Peter] I'm happy to keep it.It does spell out the difference between
| ownership and right to use, which is helpful.
| [Ken] We can ask the opinion of the larger group. The extra paragraph
seems
| to drag on the obvious but I can live with it.
| 
| | Line 984: Private actions are mentioned in section 3.3. Is it
| necessary/useful to
| | introduce here? Opacity means private actions, which may be the private
| part of
| | a joint action, are not publicly visible but are likely essential in
| realizing real
| | world effect.
| [Peter] ??? We don't use the term "private action" anywhere....
| [Ken] line 1178-1179 says "...for any given action - whether the action is
| private or is a joint action...".  Also, lines 2231-2232 later say "These
| operations represent the sequence of actions (often private) a service
must
| perform..."  The larger question is what is the relationship between
Action
| and Joint Action.  If Joint Action is a type of Action, what (Private
| Action?) is/are the non-Joint Action types?  Note, this has been a
| long-running issue and there is some more history included below.
| 
| | Lines 1013-1015: somewhat redundant.
| [Peter] Worse - somewhat contradictory, and we did talk about this: what
| state is changed? That of the participant(s)? or of the overall ecosystem?
| The answer is non trivial....
| My suggested wording:
| "This implies measurable change in the overall state of the SOA ecosystem,
| although the change is primarily relevant to one or more participants."
| [Ken] Not sure the suggested wording helps.  Can we just delete and let
the
| definition for RWE stand on its own?  It doesn't really need to be in its
| own subsection, i.e. kill the 3.2.1.2 title, but if we keep both effect
and
| RWE, we need to differentiate.  I wrote something on this for work Chris
and
| I were doing recently, but I can't find it.  Basically, it said there are
| lots of effects but RWE are the ones you care about.  Chris: did I write
| this up in a high side email?  I remember thinking I caught something
| significant and would prefer not to lose it.
*[Peter] For me it is like the light bulb example - just as 'total state' is
much more than a specific state or state change, then 'total' RWE could be
equally vast, but it is specific RWE associated with specific state changes
significant/relevant/important to stakeholders that are relevant.
| 
| | Line 1026: should choreography be connected with RM Process Model?
Should
| | that replace term choreography?
| [Peter] Drop 1026-1030? It is defined (more accurately, IMO) in 4.3.4.2...
| We could extend the last phrase of the following para (end of line 1033):
| "This is achieved through communication and messaging, which in turn
| facilitate choreography and orchestration"
| [Ken] I wouldn't mind dropping, even without the additional phrase.
| However, I think we should consider something that brings in the idea of
the
| RM Process Model.  I've always interpreted the Process Model as a
| relationship between related (WSDL?) operations, but related in the sense
of
| the RM example.  Is it worth making this connection?
*[Peter] Yes, as long as we do so in a coherent and consistent way. Let me
look at this again.
| 
| | Lines 1041-1043: these seem to relate to RM possible effects of (1)
| exchange of
| | information or (2) change in shared state. Useful to make this
connection.
| [Peter] Try this for size:
| "An actor sends a message to another actor with the intent either
| .       to communicate with other actors, without impacting their shared
| state - a communicative action; or
| .       to establish joint action in order to deliver a capability or part
| thereof, which intentionally impacts the shared state - a service action."
| [Ken] No because I can communicate with you and by providing information
| thus change our shared state without a service being involved.  The idea
of
| service action arose because Frank saw Action in the RM (not necessarily
the
| Action of intent in line 994) as the message hitting the receiver and not
| what the service responding to receiving a message.  I argued the latter
as
| an interpretation of the RM.  Frank introduced the service action to be
the
| latter while trying to preserve the former as the RM action.  Counts-as
then
| showed up to relate (equate?) the two.  Emails went on for months about
this
| and I believe we eventually got otherwise distracted with no resolution.
It
| started by Dave Ellis pointing out our use of Action was inconsistent,
| especially comparing the section 3 and section 4 views.  (For the longest
| time, I argued that joint action was the only action of consequence in the
| SOA context.)  Jeff and I worked on a resolution and (I think) made
changes
| to section 4 but the corresponding changes never made it to section 3.
*[Peter] If we accept that there are states and state changes that are
significant, whilst the total state may be (incalculably) vast, then surely
a communicative action is an action whose purpose is not to deliver a RWE in
the sense of a service, even if - strictu sensu - it does cause a state
change. A bit tautological but I think it is at the heart of the difference.
Let me also compare the wording in section 4 and see if that helps us
more...
| 
| | Line 1049: should this example be "the integrity of the communicative
| action"?
| [Peter] If we are consequent with the previous change, maybe we need to
| rephrase the whole para:
| "Different viewpoints will lead to different joint actions being
interpreted
| as most important. For example, from the viewpoint of ecosystem
governance,
| the integrity of the communicative action may be dominant; from the
| viewpoint of ecosystem security, the integrity of the service action may
be
| dominant."
| Or is that way off the mark?
| [Ken] Sink hole, rat's nest, and pit of despair.
*[Peter] So, spot on the mark then? ;-)
| 
| | Lines 1050-1054: paragraph not clear what it is intending to say.
| [Peter] Agree. If we drop it, does it have an impact? I don' think so.
| [Ken] I don't know what it's saying so I wouldn't miss it.
| 
| | Lines 1066-1069: distinction between requirement and obligation is not
| clear.
| [Peter] Don't understand your point? Obligation isn't mentioned anywhere
in
| 3.2....
| [Ken] Oops, meant distinction between requirement and objective is not
| clear.
*[Peter] OK, I'll take a look...
| 
| | Lines 1074-1075: capability isn't effect but what you do to get effect.
| Suggest
| | "A capability is an action or set of actions that a service provider is
| able to
| | execute in order to provide a real world effect required by a service
| consumer."
| [Peter] Erm.... that's the definition in the RM and it's not wrong per se.
| Maybe it just needs some qualification to distinguish capability and
actual
| bring-to-bear of the capability.... How about:
| "The Reference Model makes a distinction between a capability (as a
| potential to deliver a real world effect) and the ability of bringing that
| capability to bear (in a realized service) that actually delivers the real
| world effect.
| [Ken] The RM says, "The purpose of using a capability is to realize one or
| more real world effects."  We still need to deal with the definition even
if
| we add a reference to the RM.  The capability can do something to result
in
| the RWE; the service is the use of the capability in the SOA context.  For
| the extra line, I might say, "The Reference Model makes a distinction
| between a capability (as a
| potential to deliver a real world effect) and bringing that capability to
| bear (through a service) to realize the real world effect."
*[Peter] OK. I actually like to think of service as a mechanism for
delivering capability. Once I've been through your, Boris and Michael's
paper, maybe we'll have a better wording...
| 
| | Line 1103: "participant" should be "actor".
| [Peter] I don't agree. Actor could then also include delegate - I think
that
| is wrong. But it shouldn't be 'just' a stakeholder either...
| [Ken] Line 1081 talks about actors and that was in the model you dropped.
| Also, actor is still used in line 1104.  I think including the possibility
| of delegate is actually correct.
|
| Lines 1077-1078: suggest "For a joint action to occur, each actor must 
| be able and willing to participate."
[Peter] Agreed, cleaner and simpler, nothing lost.| 
|
| Line 1103: "participant" should be "actor".
[Peter] I don't agree. Actor could then also include delegate - I think that
is wrong. But it shouldn't be 'just' a stakeholder either...
 |
| Lines 1128-1130: delete and move idea of message as content to line 1134.
[Peter] OK
|
| | Lines 1135-1144: somewhat jumbled and after reading section 3.3, suggest
| | moving essentials of semantic engagement to here. See later comments.
| |
| | Lines 1150-1158: The introduction of service action in line 1042
promises
| a
| | useful distinction but these lines add nothing.
| [Peter] It's a curious situation - we have extensive text on communication
| actions (which, although important and encapsulating the principle of
| semantic engagement, are still 'secondary' to the actual service) and
nearly
| nothing on service action - which is presumably the heart of SOA and
| delivering RWE...
| Or is the message we are trying to get across that "getting the semantics
| right can be very tricky (Comm Action) but once done, service execution
| (Service Action) is a slamdunk" ?
| [Ken] See above for the tortured history of service action. Semantics are
| important and speaking to semantic engagement makes sense.  I've never
been
| sure how much more is needed or appropriate.
*[Peter] OK, lets' see what the TC thinks too. 
|
| | Lines 1159-1216: I'll give specifics in the following comments but my
| conclusion
| | is we should remove this section but take the essentials of semantic
| | engagement and elaborate under communicative action.
| [Peter] I'd propose a sequence of sections: Communication Action, Semantic
| Engagement and then Service Action 
| [Ken] Whatever we decide, make everything clear and consistent.
|
| Line 1160: suggest "Semantics is pervasive in the SOA ecosystem."
[Peter] OK
 
| Line 1168: It is unclear why/how assertions provide the basis for 
| semantic engagement.
[Peter] " ...understanding those assertions provides the basis..." ?
  
| Line 1171: substitute "assertion" for "proposition".
[Peter] OK
 
| Lines 1172-1176: MUST have same meaning for all participants but could 
| have different impact/purpose.  If meaning is different, then we don't 
| have common basis for consistent action or trust.
[Peter] Agreed. I've drafted an alternate wording...
| 
| | Lines 1177-1180: no idea what intended.
| [Peter] We're trying to get across the idea of engagement as being
| progressively achieved...needs rewording, if used at all.
| [Ken] More, the appropriate amount of engagement is needed at any given
time
| for those circumstances.  Doesn't matter whether it is progressive or
| instantaneous.
*[Peter] Agree, that's clear
| 
| | Lines 1181-1185: this is starting to get to the point.
| |
| | Line 1187: check the RM. Is semantic engagement a process or the result
of
| a
| | process?
| [Peter] Both and neither ;-) In the RM, it is a relationship. I would
argue
| that it is a process...
| [Ken] I see the both.  It is the process of getting the understanding and
| the result.  However, if I have understanding already, I have the result
| without a process.  Good luck in capturing.
*[Peter] OK, see your point. 
| 
| | Lines 1187-1188: is sending you an excerpt from David Copperfield an
| assertion?
| [Peter] No. You could think of that as the service action. For it to have
| any 'meaning' ("what am I supposed to do with this? Eat it? Read it?
| Translate it?"), you need an accompanying communicative action which you
| then need to understand and internalise.
| [Ken] Hmm, something here seems to be making it more complicated than need
| be.  My service description says I'll send you an excerpt from David
| Copperfield if you request it.  I don't care what you do with it.  If you
| want to analyze it, you need more meaning than if you want to count the
| number of occurrences of the letter z.  Is my sending you an excerpt an
| implied assertion it is a legitimate excerpt?  Maybe, but I don't think
| that's where we need to be going.  For a different service, my message may
| indeed contain assertions ("politician X is a war-mongering idiot") for
| which significant semantic engagement is necessary for you to determine
the
| validity of the assertion.  But that gets back to needed degree of
semantic
| engagement being context sensitive.
*[Peter] So semantic engagement can come from context, implied semantics in
the service action as well as explicit engagement fostered by a
communicative action.
| 
| | Semantic engagement could be to read and understand the excerpt.
| [Peter] Not exactly. The SE would be your 'process' of understanding what
is
| expected of you... just to read it? To understand it? Etc..
| [Ken] The service shouldn't care what I want to do with the excerpt.  Do
we
| need to tie SE with objective?
*[Peter] I see your point. I'll take a look
| 
| | Would I have greater semantic engagement if I was familiar with Dickens
| and 19th
| | century London? I'd certainly have less semantic engagement if I only
| spoke
| | Spanish, but I could still send the excerpt.
| [Peter] The process of SE is easier or harder depending on the context and
| prior (mutual) understanding between the actors.
| [Ken] yes, context needs to be emphasized.
| 
| | Lines 1194-1196: good example.
| [Peter] So let's reword the preceding para so that it feeds into that
| example more clearly...
| [Ken] good by me
| Lines 1197-1198: not a clue
[Peter] It's true but not useful in the spec. delete

| Lines 1199-1216: confusing. For SOA ecosystem to be effective and for 
| trust to exist, people have to have sufficient shared understanding. 
| This is semantic engagement.
[Peter] Agreed - I've pulled this to the beginning of the section and will
review the whole section now, probably dropping in between communication
action and service action, as indicated...