tag:blogger.com,1999:blog-438942112364524044.post5207566991504672150..comments2024-03-29T02:15:50.063-07:00Comments on Biology of Distributed Information Systems: Five Challenges for Entreprise ArchitecturesYves Caseauhttp://www.blogger.com/profile/04812034190333969728noreply@blogger.comBlogger1125tag:blogger.com,1999:blog-438942112364524044.post-90108508763490048972007-02-22T13:20:00.000-08:002007-02-22T13:20:00.000-08:00Dear Yves,Maybe another direction for your researc...Dear Yves,<BR/><BR/>Maybe another direction for your research, if I may...<BR/><BR/>"How much autonomy an intelligence needs to be fed... ?"<BR/>-> what is the cost of adding extra autonomy, intelligence or whatever ?<BR/>I'm not talking about the monetary cost, but rather about an "entropic cost" : each step of additionnal complexity in the system requires more and more energy to maintain the system's coherence, performance, quality...<BR/><BR/>What about another way of solving the "big question" : simplifying the system, reducing it to "smaller" and far more understandable and manageable components, ultra-efficient to handle repetitive tasks, and relying onto other mechanisms (such as human being ?!) to handle the complex issues ?<BR/>A kind of Pareto's law applied to IT : 20% of the investment is probably able to handle 80% of the business processes ; are we sure that the 20% remaining really complex bit couldn't be handled more simply and cheaper with other means ?<BR/><BR/>Complexity leads directly to dependancy : only a few guys understand "how it works" (hmmm... do they ?!) and, worse, in case of failure, everything might stop (HLRV ?).<BR/>When you think about it, IT over-complexity leads to the exact opposite of its goal into supporting business processes : manageability turns into unmanageability.<BR/><BR/>See you !Anonymousnoreply@blogger.com