This article originally appeared on the BeyeNETWORK
I very much enjoyed your article on business metadata, and I very much agree with you – the problem is as slippery as a trout in a mountain stream. I would like to add some points to what you said.
In my early days in computing (back in the 1960s), I lived with the delusions of our technical abilities to solve problems. While I must admit that the problems we face are partially technical, the most important problem is the people issues.
In government and most companies, I get paid for doing a job and making my manager look good. I do not get paid for helping others – in fact, I may get penalized for spending too much time helping others. From a metadata perspective, we can only get good business metadata if we sit down and cooperate, and that takes time. Also, if we cooperate, I may have to go back and change some of what I have previously done – and no one likes to pay for what they perceive of as "worthless rework."
Another factor is ego. After all, "I understand my data and no one will tell me otherwise. The fact that my data is not in consonance with your data simply means that you are wrong and must look at things my way." Even when a congressman holds up two reports with glaring conflicts from the same agency, little seems to happen after the news show.
Also, even when we create metadata, we seem to be somewhat averse to the use of adjectives and other qualifications on the term we are defining. As you phrased it, is it existing customer, new customer, potential customer, et cetera, et cetera, et cetera (as the King of Siam would say).
If senior management’s bonus were tied to cooperation and that cooperation was actually documented and measured, we would make real progress. Until then, there will be our usual collection of fits and starts.
From a cynical perspective, in the good old days, we had data dictionaries. These didn't work well because we really didn't do the job we were supposed to do. So rather than admit we did a poor job, we then sold the industry on metadata. That suffered the same fate, and we went on to semantics, which did no better. The latest word seems to be ontology. I think you can guess what I think of its success.
Technical metadata is the real easy task (compared to business metadata), so some companies obfuscate and claim that the usual technical metadata is all that is needed.
So, what seems to be happening is that faced with the problems you outline as well as the people issues, the favorite response of business and government is a version of “run for the hills.”
The only bright side I have seen is that some vendors of standardized data models have created real metadata for each and every entity and attribute. They then use this as a selling point. This is far from a "cure-all," but at least it is a small step in the right direction.