Less Checking, More Correction

This morning when I went through parts of my actuarial study notes on modeling, I came across this statement:

"The modeling process often has a demanding deadline and a temptation could arise to accept the results from the model without performing a rigorous validation process. It is imperative that sufficient time be allotted to validating the model and subsequently reconciling the model results to emerging results."

This statement has rightfully highlighted the importance of validating & reconciliation exercises during modeling process - in fact, this statement is also true for other actuarial works as well as system implementation. Theoretically, we know that we should allocate adequate amount of time & resources to check/validate the model/process we have built up and reconcile the results the model/process has produced. Unfortunately, in real life, not many of us do what the statement has suggested in practice.

"We have to forego vigorous checking & validation due to tight deadline" - does this statement sound familiar to you? Yes, time constraint always appear as one of the major reasons (or excuses?) of not doing adequate checking & validation exercises, which are always viewed as unpleasant burdens that do not benefit the entire project much - sometimes, these exercises may be viewed as obstacles to deter from completing a project (especially when the project is significantly over-due). To make the matter worse, when the management starts to get uneasy with the project implementation team ("PIT") for missing deadlines or reporting recurring issues, the management's pressure will "encourage" the PIT to sweep the potential issues under the carpet so that they can can report good news to the management - to suppress the potential issues from emerging to the surface during the project timeline, PIT will definitely cut the amount of checking & validation to be done (how on earth we can we detect a problem if we do not carry out checking & validation?).

Unfortunately, the less checking & validation work we do during the project timeline, the more corrections we are going to do in the future. "Never mind, we can do enhancements later and schedule as parts of Day-2 exercise" - ehhmm, I don't mean to harsh, but are you sure that the "Day-2 exercise" will ever come? As what I have learned from some articles on Deming Cycle around 5 years ago (Manager Today, ISSN 18132391, May 2007 Issue), corrections are always much difficult comparing to do things correctly at the first place. Try to imagine if we promote the unit creation & deduction process which contains errors to production, how much efforts & resources to be put in do rectify the erratic data generated in the policy admin system? How many stakeholders will be affected by the errors which by right can be detected earlier? Unless we perform a thorough checking & validation exercise, we will not know how we rectify the error we encounter today will cause another error tomorrow.

Of course, we should not over do checking & validation until it becomes too tedious and consumes too much resources that we should spare for other work. In short, we should strike a balance between theory and practical.