A suite of standards and models can be hard to communicate. People with an urgent technical focus just want to know which tool will get the job done, and aren't concerned about how that tool fits in the toolbox. People with strategic responsibility resist the idea of adopting an alien language or structure that's supposed to describe what they're already experts in.
One of the ways of managing this problem is to bring standards and models together into an overall structure that pretty much unpacks and explains itself. Another approach is to create layers of abstraction, so that detail is organized in a hierarchy, making it more tractable. We have used both these techniques to produce the ACORD Framework.
Blogger Brandt Redd has a good graphic showing a four-level model for data standards. The contents of the levels in his model are technical. But his explanation of the leveling itself is firmly business-oriented. Redd's higher levels show increasingly “broader applicability and longevity of standards”. Reading the chart in the opposite, downward direction reveals increasing “ease of data exchange and depth of system integration”.
This is a useful perspective on the issue of standards organization. At ACORD, we deal with the dynamic between generality (for broader applicability and longevity) and particularity (for ease of implementation). However, I question the coupling of “ease of data exchange and depth of system integration”. The implication is that you get easier data exchange the more integrated your systems. While this was true once, I believe that the rise of distributed systems has led to architectures which are more collaborative than integrated. With web-based hubs and agents doing the leg work of translating between low-level formats, and service wrappers making legacy systems more amenable to new uses, integration isn't what it used to be.
As software engineering continues to develop, the tasks associated with the lower levels of Redd's model are increasingly being performed “out there”, on the network. See Redd
Comments