While pondering the difficulties still faced by enterprises employing a business rule management system (BRMS) and attempting to fully enable business users, I see yet another wave of standards coming our way. Standards for rules have been attempted in a variety of ways over the years. Each attempt provided some valuable insight but none really answered the call. This got me to thinking about the how far the field has come in terms of approach and execution but not in terms of standardization.
I have been fortunate to be involved with rules long enough to see a tremendous evolution in how the technologies are labeled, presented, integrated, and ultimately utilized. We have truly seen a sweeping move across the spectrum from purely academic research platforms to a boutique niche technology to its current place as a mainstream part of everyday business.
In the days of Artificial Intelligence we had something called expert systems. These were powerful tools that allowed for some advanced reasoning techniques via various execution paradigms. The use of “expert” implied these systems could emulate the thinking of a human expert in a particular domain. We saw attempts to do more complex problem-solving here: planning, diagnosis, monitoring, and sophisticated reasoning.
From there we moved to rules-based systems and business rules. The emphasis had by thenshifted to the extent that we were no longer concerned with complex reasoning; instead, the rules served to define or constrain some aspect of business. In fact, they could be considered an actual manifestation of the business rules of an enterprise. However, we still only had what was primarily a technical implementation. Consequently, the concept of business rule management systems grew. Now we could theoretically represent the rules of business and do it in such a way that actual business analysts could manage – read, write, edit, expire, save, and deploy – via a rule repository. There were a great many successes with the integration of these systems but the reality never really seemed to completely match the hype. Business users were more involved but there was never a complete passing of the torch from IT. This hurdle – along with some possible approaches – was discussed in my last article: Why Johnny Still Can’t Write Rules.
The most recent step in the evolution is perhaps the most interesting. As business rule management systems grew, so did business process management systems. And across all of them was an explicit (or sometimes implicit) workflow. Now, when viewed together, it appears that an overriding approach has emerged that allows us focus on what we must be most concerned with: decision-making. We can think of both strategic and operational decisions, but perhaps the latter is more relevant at this point – the decisions that reflect the day-to-day operations of any business. This, I believe, more accurately represents how a business analyst thinks: not in terms of a specific rule or a process but in terms of a decision that must be made. This combination of process, workflow, rules and even analytics can make quite a powerful platform. It looks great on paper and there is terrific software for the execution of it — but can we really put this in a useable and manageable form?
This was all a rather long introduction to get us to the point of this narrative. I believe the industry has again found itself at the familiar crossroads of looking for a uniform, practical, common representation for what has come to be known as decision management. The search for standards began with business rules and we saw approaches like RuleML, XBRL, RuleSpeak, SVBR, and many others. All great work, but none provided THE answer. With every BRMS vendor using its own proprietary representation for rule execution, and their own syntax for expressing rules across various metaphors, this task became increasingly difficult. Consequently, virtually every business rule implementation became unique – not only from company to company but often within the same company. More recent efforts, including RIF (Rule Interchange Format), can potentially narrow the gap – but the gap it still exists.
Working on the consulting side of the business, I can’t even begin to count the number of times I’m asked for a ready-built library, or at least a template for rules within a specific industry. People have repeatedly asked: “Can we just have the rules?” This request is not so much due to a desire to share and standardize, but really reflects a desire to decrease the workload required to start from scratch. This is a very reasonable request but one that has really never been answered (and in many cases shouldn’t be).
And now we see yet another emerging standard for the new world of decision management – the Decision Management Notation (DMN). From what I’ve heard, this standard will be finalized and submitted to OMG before the end of the year. At that time we will certainly have much more to discuss when the individual components can be broken out and analyzed. The stated goals for DMN are twofold: 1) to provide a notation understandable by all business users and technical developers, and 2) to ensure decision models are interchangeable across organizations via XML.
While I strongly believe in the general approach of decision management and am fully supportive of doing so in a way that fully enables business users, I’m still somewhat perplexed at the constant drive for standards supporting interchange. Decision-exchange? Rule interoperability? Process sharing? Do we really need this or even really want it? To be sure, there seem to be some potential advantages including the facilitation of transparency and consistency in the application of regulations. On the flip side, do companies risk losing competitive advantages or retaining customers if their “secret sauce” is possibly shared? A number of industries are certainly pushing uniform standards. The mortgage industry has MISMO standards in place although not a lot of companies use them – yet. The insurance industry has ACORD. These represent probably the two earliest and most widespread adopters of rule/decision approaches. Will these use DMN? If they do, will they actually share with other companies? Vendors will likely do what they can to become conformant with the standard but perhaps only so they can ensure someone evaluating their software can check that box during the assessment process.
Someone wiser than me once said “when you only have a hammer, the whole world looks like a nail.” While you can probably get a board into two pieces with a hammer, it won’t be pretty and it won’t be easy – and it would be a whole lot easier with a saw. Far too often the attempt to funnel a diverse set of rules, processes, and decisions into a single “one size fits all” bottle feels like taking a hammer to a board. Perhaps we should keep the saw as well.
