> As a discrete set of alternatives start to emerge, I strongly suggest quantitatively modeling out the impact of each one and revisit your Setting — specifically the why, the optimization function. It can be very hard with ambiguous decisions to get down into the numbers, but it’s very valuable to do so. Figure out your value metric, your success criteria.
In software engineering decisions, estimating the superiority of an alternative (eg: some performance measure) can be a long process depending in the level of detail required. Napkin calculations can be done in minutes but building a prototype can take as long as delivering a working product.
When the assessment of alternatives is abridged, confidence in the decision is compromised. This may be one reason (not including changing requirements) why agile methods are popular — they absolve stakeholders of committing to an alternative up front.
Douglas Hubbard’s “How to Measure Anything” discusses some ways to assess the information-value of improving estimates. His approach can help strike a compromise between doing napkin calculations and building fully-functional systems.
In software engineering decisions, estimating the superiority of an alternative (eg: some performance measure) can be a long process depending in the level of detail required. Napkin calculations can be done in minutes but building a prototype can take as long as delivering a working product.
When the assessment of alternatives is abridged, confidence in the decision is compromised. This may be one reason (not including changing requirements) why agile methods are popular — they absolve stakeholders of committing to an alternative up front.
Douglas Hubbard’s “How to Measure Anything” discusses some ways to assess the information-value of improving estimates. His approach can help strike a compromise between doing napkin calculations and building fully-functional systems.