This is my response to a Forrester publication on Enterprise Architecture Evolution:
Quote: “The recently published Forrester Wave™ on enterprise architecture (EA) tools demonstrates that the real benefit of these tools is not in the modeling itself, but in the modeling that supports real EA objectives — such as knowledge-sharing, collaboration on standards, and providing aggregated dashboards that help decision-makers choose from different scenarios.”
I happen to not fully agree because while all the above is important the EA tools proposed are outside the actual application and purely in the development environment. Evolution is however an internal function of the execution itself. Here my response:
Henry, interesting subject indeed!!! But does the term EVOLUTION not imply a seamless functionality within the application system? Nature seems to evolve without someone performing simulation and analysis and then makes an assumption as to which changes might bring an improvement? Well, we could have a religious discussion here on ‘Intelligent Design’ but let’s stay away from that. Evolution works by taking fitness information straight from the execution and feeding it right back into the next cycle of execution. Darwin was even apparently wrong that fitness happens by chance because he was not aware of the functionality of GENE EXPRESSION. Genes do not have to change to improve the design, they just have to be expressed in a more fitting way.
We need to do the same with software! Analysis, simulation and monitoring are NOT AGILE, Henry! They are just little less rigid. In any case I propose that no-one wants or needs rigid processes no matter how well analyzed or simulated! Simulation is guess work and yes, using the Monte Carlo method the guess work can be optimized but it still is a ‘gamble’ (sic). Las Vegas calling. Business needs an environment in which processes – and applications representing those – tune themselves with fitness information coming straight from user interaction in real-time into the next execution cycle.
This is what the Papyrus Platform does with its User-Trained Agent. Let the users do their work and feed real-time performance data right back into the execution environment to improve the processes by discovering decision patterns based from user interaction. Analysis and simulation are – like process I might add – outdated concepts from manufacturing. Running a business well is not manufacturing, it is about empowering the business user to service the customer and not by putting BOTH into a process straight jacket.