Link: http://www.etc-architect.com/?p=73
Once an architecture manager told me that he could get a good business case for all types of architect except of data architects as they would only be used to create data standards that were the death to all business. This is up to day a common view and relies in the early history of our profession. So the first data architects started to build data models of ever and all data including screen layouts. For us this may sound way over top, but since their was relatively few data on IT systems this was really not that much. The problem was just the governance of their data models, as they argued that all changes would need their approval as otherwise there would be utter chaos and business would stop.
Since this was clearly not manageable to do, without freezing the business, most people just ignored them and found out that while there was a little of chaos in terms of information clarity the business was actually doing much better without data dictionaries and standards. So a new generation of IT Heads and other manager were all told this as a warning. That in terms lead to situations that data got into a really chaotic state with the business starting to suffer. Once this was noticed the real stupidity started in way that typical for short sighted managers. It was argued that if IT departments would stop to develop their own code and instead just buy COTS package they could move the responsibility for data management and standards to the COTS vendors. The COTS vendors smelling some good business readily agreed on the assumptions that their software would be implemented without any customisations.
Now the situation got worse as the COTS packages got customised and additionally the COTS vendors also off laid data architects and modellers as they sought that their own data model was no longer changing. While it is true that a data model keeps very stable for many years, constant small change adds up in changing it. So after it got clear that just offloading the responsibility to COTS vendors did not work either a rather strange (or stupid) solutions was found that is still in place today. The solution was that data was given in the ownership of a business owner who had no knowledge on data management. These owners usually handle their responsibility by first ignoring it or use it for their advantage. Only if the dat starts to get out of control and hurt the business consultants are called in that will try to rectify the situation with expensive clean ups and restructuring. Additional IT will complicate the process by creating large master data management solutions that will try to repeat data management in almost the same way as the early data architects and modeller in locking it down with a strong governance and start the whole madness all over. Thankfully nearly all of these master data management initiatives fail early before they start to hurt the business.
The answer for breaking this stupidity can usually be found by a data architect leading a data standardisation through thought leadership. Since this data architect will however first need to understand all processes and all changes to processes as data follows processes there will never be too many of his kind, so the circle of data stupidity will continue ad nauseum.
