Ultra Structure

This is condensed from a Jan 11, 1995 posting to the DigitalkCompuserveForum?, by Hal Hildebrand.

Hal says in 1995...

"Ultra-Structure: A Design Theory for Complex Systems and Process" Jeffrey G. Long and Dorothy El Denning. Communications of the ACM, January, 1995.

"...I don't have the article here, so I'm going off memory, and it's a dense article...Anyways, the upshot of the thing is that with a small number of rules (<70), you can completely describe extremely complicated systems (such as a major corporation)...Very powerful, and applicable to just about any system, not just business...gives the surface structure the richness we observe of systems in the real world...Anyways, it is quite cool, and I can see this as having a large impact on the things I do... Recommended reading. -Hal."

http://www.acm.org/pubs/citations/journals/cacm/1995-38-1/p103-long/

REVIEWS From Computing Reviews Elizabeth Ann Buschlen Unger

Those interested in the design of complex systems, or for that matter almost any dynamic system, can benefit from the premises of this excellent paper and the resulting work. One underlying premise is that our current conceptual structures, such as objects, work well for static systems, but dynamic systems require the capture of the complex semantics primarily in data rather than processes. Founded on Chomsky's transformational grammar, Ultra-structure is founded on two essential hypotheses, ruleforms and complex operating rule engines (COREs). Processes are the result of executing a series of rules, which may change. Constraints on rules are defined by ruleforms. COREs of less than 50 ruleforms have been found sufficient to represent all rules found among systems sharing common structures, such as institutions.

The concepts and the model system are described in an easy-to-read tutorial format illuminated liberally with examples, such as pricing rules and general ledger processes. A case study is described and the results are presented. Comparison with another design method for complex systems demonstrates the effectiveness of this approach in one case.

This well-written, clear, and meaningful paper should be required reading for students of software engineering and all current system designers, especially those working on dynamic systems.

I find it surprising that it takes so many rules! From my reading in EmergentSystems I would have guessed a much lower number. even quite complex behavior (e.g. insect walking) can be described in very few (6? 8? memory fades...) rules. -- MikeMorris

I think that a large business is much more complex than a typical insect. Can an ant fill in a Schedule C (Profit and loss...) Tax form?

Back to UltraStructure! A very provocative paper. But can everything be described by tables of data? Could Codd have been right? And can every business be described by a collection of tables with different data but the same structure? A common conceptual model for all commerce? ERP!

--DickBotting

It almost seems like the paper is saying you get more milage if you don't try to describe everything by tables of data. By having multiple, heterogeneous tables, and also by having glue code, you have a place to put different types of rules as you find them, but you don't have to worry about forcing all logic into rules of the same shape. --StanSilver