Modern Quality Assurance: prescriptive vs reactive




In traditional industrial mass production, quality can be defined as conformance to specifications. And it can be assessed upfront in a prescriptive way.

For complex industrial products this definition of quality at times can be not enough. While for digital products from the information-age, this definition is largely insufficient.







As surfaced during a discussion with Mike Cohn, nowadays quality is also:

  • Fitness for use and fitness for purpose
    Sometimes products, especially new ones, can get used in ways and for purposes that cannot be anticipated nor conceived. When people start using a product in totally unanticipated ways, will it display the same quality?

  • Contextual and subjective
    Different people can also look at a product's quality in different ways and from different prospectives. Indeed quality is in part in the eye of the beholders, depends on the person that judge it. An engineer for example could look at product’s maintainability, extensibility and evolvability while an end user could look at how well the product works. A casual user for example could look at product’s learnability and understandability, while a frequent user could look at configurability and speed.

  • Relative
    A product for example can be superior or inferior to a competitor’s product or to a cheaper product.

  • In evolution and changes over time
    The product and its users change and co-evolve together over time. Think for example at the history of tools and humanity and how these are inextricably intermingled.

    Mike Cohn: "I remember many many years ago Borland’s Turbo C 1.5 came with source code to a sample spreadsheet application. It didn’t do much but it was as good as the VisiCalc I’d run on CP/M years earlier. I remember thinking I wanted a time machine—If I’d had a time machine and that spreadsheet source code I could have gone back to before VisiCalc came out and sold the little Borland mcalc spreadsheet and have made millions. And 10 years later simple spreadsheets were so simple Borland just tossed them into their compilers as freebies. So, a “quality spreadsheet” does depend on context—neither of those would be viewed as a quality spreadsheet today. "

  • Intangible
    I find it hard (impossible?) to define quality as much as I find hard to define and measure intangible things like humour, kindness, grace or personality.

    Mike Cohn: " And good points about comparing quality to humour, grace, kindness, and such. Those are very similar—perhaps definable only in context and subjectively.:

Because of this, there is an inherent limit for how much quality problems can be prevented with a prescriptive upfront approach to quality assurance. Over that limit, investing more time and effort wont produce any improvement in quality.
Reactive approach to quality assurance consist in quickly spotting and reacting to quality problems, diagnosing and solving them so quickly and effectively that there are no or very little consequences.
Some of the learnings made using reactive approach, once acquired can then be incorporated into the prescriptive approach, for others will only be possible to spot them in a reactive way.




So in the end reactive approach is used in combination with prescriptive approach, balancing the two in a way that fits the circumstances.
As Alan Shalloway put it in a very concise way:
Focusing on quality doesn’t mean focusing on getting it right the 1st time. it means improving the system to avoid & detect errors.