AgileAttitudes Article

 
 
 

Vol 02 Issue 08- The Essence of Agile - Empirical Process vs. Defined Process

 
 
Back to the list of articles Agile Attitudes Volume 2, Issue 08 July 25, 2005 A free bi-weekly email newsletter Brought to you by Agile Rules consulting www.agilerules.com ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Welcome to Agile Attitudes, a newsletter of ideas, insights and technical tips that help people find better ways to develop software. Feel free to share this with anyone - just be sure you send or print the whole thing, including the copyright notice. Directions for managing your subscription are below. O><O><O><O><O><O Agile Rules will be presenting a tutorial on TDD for Embedded Software development at the upcoming Agile 2005 Conference in Denver tomorrow on July 26. We presented a short version of this tutorial at the last Agile Bazaar meeting on July 14 at Tufts. Agile related meetings and other events in the Boston area are discussed on the Agile Bazaar email list. For info see: http://groups.yahoo.com/group/AgileBazaar/ O><O><O><O><O><O If you would like to receive an email as soon as we know of an event in the Boston area of interest to the agile software community, you can sign up for the announcements list at http://www.agilerules.com/mailman/listinfo/agileannounce O><O><O><O><O><O The Essence of Agile - Empirical Process vs. Defined Process by Nancy Van Schooenderwoert In software development there are two choices: either you plan your project thoroughly or you just "wing it". Right? I used to think so, but not any more. There is a third alternative. In the world of industrial process control there are two major approaches to controlling any process " the "defined" process control model and the "empirical" model. The defined model is what you are using when you attempt to thoroughly plan a software project. "Winging it" is just another word for chaos. So that leaves the empirical process control model " what is that? Empirical process control is the essence of all agile methodologies. It expects the unexpected. In defined process control, every piece of work must be completely understood. Given a well-defined set of inputs, the same outputs are generated every time. Given clear, stable requirements and well-understood technologies, a software project can be predicted. But when do we ever see those preconditions? Empirical process control is used in situations where the inputs are varying, and the process is too complex to produce repeatable outputs. Frequent inspection and adaptation is used to control such processes. In his book "Agile Software Development with Scrum", Ken Schwaber describes how he tried using defined software process methods in his software company with very unsatisfactory results. He met a group of leading process theory experts at DuPont, and described his problem to them. As Ken tells it: "They inspected the systems development processes I had brought them. I have rarely provided a group with so much laughter. They were amazed and appalled that my industry, systems development, was trying to do its work using a completely inappropriate process control model. They said systems development had so much complexity and unpredictability that it had to be managed by a process control model they called 'empirical'. They said this was nothing new, and all complex processes that weren't completely understood required the empirical model." This was a turning point for Ken. He realized what was so wrong in the software industry " "we were wasting our time trying to control our work by thinking we had an assembly line when the only proper control was frequent and first-hand inspection, followed by immediate adjustments." On first hearing about agile software practices, many people wrongly assume that it's just a form of "winging it" since it certainly is different from the defined process control method. Nothing could be further from the truth. Empirical control is a completely legitimate, disciplined way to control a complex process like software development. When you use a defined approach to control software development, you go too far on assumptions in an attempt to plan sufficiently. You get a false sense of security. Now you know what we meant last time when we said that your "plan" will build nothing but a house of cards. O><O><O><O><O><O More articles on Agile software topics at http://www.agilerules.com Within our company we have a sub-specialty in embedded systems. Our site has articles on embedded XP and we support a discussion list focused on the use of agile methods for building embedded software. The list signup info is at http://www.agilerules.com/mailinglists.phtml O><O><O><O><O><O To help you get started with in-depth research into Agile Attitudes topics, we have added a Library section to our web site at http://www.agilerules.com/library.phtml Order using our links and receive discounts up to 30%! O><O><O><O><O><O If you enjoyed this issue or found it useful, forward it to a friend! Help spread the word about better ways to build software. Invite your friends and colleagues to join our growing reader community at http://www.agilerules.com/mailinglists.phtml O><O><O><O><O><O Looking for a speaker for your next corporate or society meeting? We present dynamic, informative programs on topics of interest to managers and technical staff in their transition to more flexible, robust ways to create software. O><O><O><O><O><O Want to reprint this issue in your company or society newsletter? For permission to reprint any of the articles, contact us at info@agilerules.com. O><O><O><O><O><O Your feedback is welcome! Send feedback to info@agilerules.com To manage your subscription: http://www.agilerules.com/mailman/listinfo/agileattitudes O><O><O><O><O><O Brought to you by Agile Rules consulting 162 Marrett Road, Lexington MA 02421 Copyright (c) 2004, 2005 Agile Rules info@agilerules.com O><O><O><O><O><O Privacy notice: We will not release a subscriber's address to any third party for any reason. This is a strictly opt-in newsletter. No one is ever subscribed without their explicit request. </plaintext> </td id="bodytable_r1_c3_body"> </tr id="bodytable_r1"> </table id="bodytable">