Is a “loose data standard” a contradiction in terms?
The idea of “loose or even ad hoc” standards for metadata pops up in an interesting post by Charlie Greenberg. One of his fictional examples has the term “Retired Employees” defined in different places as “age >65 years” and “age 65-79 years”. This would lead to different counts in different parts of the organization – and perhaps even truncation of benefits to some people of 80 and over. Is that a loose definition? I think it's a case of a faulty standard. Because it's not standard.
But I think the idea of loose standards does have some merit. This is where data items are optional. The optionality of data items, and the logic which can group certain optional items together, is an important aspect of the business. For example, if you're insuring home contents, it's optional for people to have a bicycle. But if they have a bicycle, you want to know whether it's the kind of bicycle people like to steal.
Creating standards around optional items needs the kind of looseness engineers call tolerance – or, less formally, wiggle room. This is why we talk about standards being robust and resilient, rather than rigid. They need to be created with flexibility in mind. INFOSEC
Comments