Making sense of: KMWorld – Best Practices in BPM 2008
I just read a January 2008 KMWorld collection of White Papers and here is my take on it. I particularly like Editoral Director Andy Moore’s musings on the BPM market. Moore first asks why business intelligence vendors (BI) are disappearing and if therefore BPM is new BI. I have voiced my belief that BI is creating knowledge illusions before and think BI goes the route of customer relationship management (CRM). Everyone does embedded CRM now. So will BPM do embedded BI? The vendors in this piece obviously think so. I agree that business process needs access to live data for decision making but no historic data. I don’t see the benefits of the predictive modeling part of BI either. Predictions are never properly weighted for probability. A prediction works on a SINGLE mathematical hypothesis that is built from inaccurate past correlations on inaccurate past data. The more historic data you use the less likely a prediction gets as the data correlation hypothesis can not model the variance of real causation. The world is not only changing now, but it was changing back then.
Laura Mooney of Metastorm believes that these problems can be solved by centers of process excellence where everyone in an organization collaborates on process strategy, analysis and execution. Well, here we go. What is a strategy? It is a high level plan that is now compared to an outdated correlation hypothesis. Excuse me, but a strategy silo will contain nothing but guesses about the past and future that are now compared to each other – no matter how you name them. It creates confusion and not understanding. Understanding is based on intuition related to real world observation. How likely is it that a group of very diverse people from operations, staff and IT will agree on anything when they collaborate on strategy? Get out from your silo and get real by speaking to employees and customers. When anything needs a center of excellence then it is too complex, too expensive and too distant from reality. Reality is NOW.
Brett Stineman of EMC at least moves closer to what I have been saying for 20 years. There is no process without content. I say the content is the process! The summary state of inbound and outbound communications content defines the process progression. Process flows are an oversimplified abstraction that cannot deal with unexpected events and complex process interdependencies and most of all – they resist change. The huge amount of intersecting flowchart grids are like the armouring of concrete. Andy Moore says that we need real-time responses to environmental stimuli rather than automated processes. I agree. But how would a strategy silo that verifies my analysis that drives my process execution react to stimuli in real-time? Beats me.
Rashid M. Khan of Ultimus says that “change is the mother of obsolence.” Right, Saddam Hussein has sneaked into our daily language more than he could imagine, but I agree here. I also agree that roles, rules, steps, forms, services and data are the core building blocks of processes. Kahn, in difference to EMC, completely misses the need for content. I do however not agree that XML is the right solution for a data model. XML is way too limited and hugely inefficient to move the data model around with the data in one of the most inefficient data structures there is. We at ISIS Papyrus dumped XML internally because a tag as such means nothing. Even when we use SOAP and WSDL there is a lot more logic needed within the XML writing and reading code to put the right meaning into the data. Therefore process management needs a central life-cycle repository to manage ALL above elements related to process and content.
Vasu Rangathan of ArborSys sees the need for content and for a repository but just wants to store his XML there. He speaks about XML offering WYSIWYG and I wonder when that happened. Is he talking abut XSL or DOCX or any other X-thing? XML is huge, slow, inefficient and hinges on the hope that all people of this world will agree on the same tag names and structures. Nope, will not happen. Not even within a business. 80% of Java code does XML parsing, XSLT transformations, and data validation without central meta-data control. It is a software management nightmare. The Semantic Web is a pipe dream. Rangathan further sees content as rigid documents, when the business reality is that each document is a dynamic mix of business data, text, dynamic graphics and images. Adobe InDesign doesn’t cut that!
Microgen plugs the need for processes also needing transaction like performance and fully integrated with business rules. I agree with that. A properly designed workflow system would allow parallel transaction processing across distributed servers – which most BPM systems don’t do. Well, there go all XML, Java, BPMN and BPEL based workflow systems out the window. No high speed transaction processing there.
Robert Shields of Telelogic mongers the usual amount of FUD (fear, uncertainty and doubt) before he flogs process flows and Visio modelling as the solution to bring multiple BPM, CRM and ECM products together with ERP. Good luck!
Let me conclude here: Yes, process management executes in real-time. It is therefore closer to business reality than strategy, process analysis and business intelligence. But process flows do not adapt in real-time. The truth is, state-of-the-art BPM is just another way to develop applications but admittedly more flexible than coding. The problem is the illusionary abstraction of process analysis by flowcharts. Work does not flow, but business content changes over time as people and applications interact with it. We need to cut out the strategy and analysis middlemen and put the people who do the work in charge. As they work with the content, the process management software needs to recognize the complex events behind their actions. These complex events represent the process progress. Change over time is no longer an abstract analysis effort but happens by learning from the users. Then we are finally real-time not only in execution, but also in adaptation. The Papyrus User-Trained Agent does just that. Hmm, that means management has no control over business process? Not so. Management has no need to audit the process analysis, but needs to audit what really happened. In a business process – according to Michael Hammer and every other BPM guru – the customer service outcome is relevant, and not how the process was performed. Suddenly we don’t monitor process execution against an abstract strategy and analysis, but care about business performance. And suddenly we don’t need BI guesswork.
By the way, can we drop that stupid word AGILITY? Its a dog sport, for Pete’s sake!