Generative Methodology thread

Thursday, January 26, 2006

Communication to part of the SOA Blueprint Technical Committee (at OASIS) à [144],

First replies to this communication are edited and first names are used because we are interested in concepts more than individual origins - as part of the AIC (appreciation, influence, control (www.odii.com ) process development methodology and the theory between the BCNGroup Glass Bead Games.

Three individuals on the TC (technical committee) were on the To: list: they all responded. I make some foot notes to their comments.

Paul,

Very interesting points. Something I was hoping to accomplish with the proposals I made a while ago.

Great feedback, I would be interested to hear a bit more of the ideas you have.

- Dan

Paul,

While I have to admit that some of the terms here sailed over my head as a humble implementer of SOA, I think that the formalism elements you are talking about would definitely help around the SOA Adoption Blueprints, the challenge around formalization of definition both in terms of the levels at which blueprints occur and in the relationship between different artifacts (one question here is how your work would view things like WS-CDL as a formalization of Mill's canons). [1]

The complexity of the SOA Adoption Blueprints charter is one that comes down to a simple question posed by (again) a much smarter man than me

"If you can't explain it simply, you don't understand it well enough". - AE

What is needed (and I think generally accepted on the TC) is a clear and formal definition of the problem domain(s) we are looking in language that is understood by people who need to apply SOA in their field.

If I understand your work correctly then it actually represents the separation in the TC between the technical adoption blueprints (potentially First School as they are at a objective level) and the business adoption blueprints (almost certainly second school as they are at a subjective level) which is part of the challenge we are currently grappling with. [2]

I've cc'ed in Miko (as the chair) who is almost certainly better positioned than I in defining how this would fit. The challenge I see is putting these elements in simple terms so they can be adopted generally. [3]

Cheers

Steve

Hello Paul,

I think you are certainly talking about some of the right elements of formalism that are lacking in the space between an RM and implementation. [4]

I would be interested to look at how your ideas are concretized into blueprint form, so you have any examples of such? [5]

Thanks,

Miko

Much of the points Paul brings up remind me of the way you approach creation of painting/detailed drawing.

The best approach I have used was to start with points of the subject that you where drawing (blueprint A so to speak) which gives you a rough idea of where the pieces of the painting will develop allowing for iterations without changing a lot of the detail. This approach is followed up by a contour study of the subject within the point layout representing a second theory/blueprint to allow you to get to the detail a bit faster.

Once both techniques are used you beginning to see more of the subject coming through which you than add the detail and the corresponding tones. This can represent the invariance within the drawing analogy.

The idea would be to sketch a and b rather than show the detail [6].

- dan

[1] I will review the W3C’s WS-CDL, but the W3C has been good at top down descriptions of what a standard should do, but has not taken the observational position that J S Mill did, that the “secrete” Soviet (four decade) project in applied semiotics did.

[2] The First School rushes to continue a pretense that logic (and the term “logic” no longer has a single meaning) can be imposed on what are being called semantic models. The technical adoption should come second, and the loosely defined Blueprints should be developed without the technical adoption issues in mind. What might happen is something called “bypass”; where once the full set of Blueprints is developed (say 50 – 100) then everyone sees a technical bypass.

[3] I agree that the level of confusion in the market is not simple; however in the kind of complexity theory my group represents we see complexity as the simplest explanation of certain aspects to reality. Complexity is simply where there is more than one interpretation for the same thing. Whereas computer science and formal systems avoid natural complexity (and actually confusing the term by associating meaning to the phrase “computational complexity”) human being use natural complexity all the time in everyday life.

[4] This is also the conclusion that my client, a Canadian firm, come to; ie that there is space between the Reference Model and implementation.

[5] A blueprint should be developed by a domain expert, without considering the technical issues involved in implementation (for more on this I have composed à [146]

[6] The notion of stratification is that there should be no relationship between the set of primitives and the set of compounds… like atoms and molecules. The first set is to have a set of molecules to talk about, and then the second step is to discover what are atoms.