David Brakoniecki wrote a great post on "those pesky rules" last month and I just had to comment on it.? The startling finding was that at one insurance company, 30% of the rules were flat wrong.? As David says:
Given that the insurance business is really little more than sets of rules - underwriting rules, claims management rules, customer cross-sell rules - and that it is a heavily regulated business, incorrect rules are more than bad business but potential regulatory nightmare.
Bingo.? The problem with rules is that many of them are simple, but the interactions between them are not.? The resulting outcomes are not.? There are better ways to represent these kinds of solutions (constraints, heuristic search, etc.) but they require pretty advanced education to model, and most companies are looking to de-value expertise rather than invest in expertise. So simple rules are used - simple in each instance, but complicated when taken as a whole. Unpredictable as well.? The right abstractions are not available for modeling, so granular abstractions are used, and they're just not good enough.? It becomes unmanageable or inaccurate over time.
As Dave goes on to say, it just isn't realistic for the business to maintain rules without assistance from IT.? We have to get past the idea of IT or Business owning the assets of the business.? Both parties need to take responsibility for the health of the business and the health of the assets that allow that business to perform smoothly, or at all.