Barry Devlin’s new book: Business unintelligence

One of the founding fathers of the data warehouse concept, Dr. Barry Devlin, is out with a new and quite provocative book. His first book, "Data Warehouse -- from Architecture to Implementation", (Addison-Wesley, 1997), is one of the classic textbooks about data warehouse architecture. His new book is titled ”Business unIntelligence – Insight and Innovation beyond Analytics and Big Data” aims to provoke new thinking and new, broader, even psychological approaches to support the process of business decision making.

”Business unIntelligence” may sound negative the first time you hear it, and the book certainly contains some criticism of the very data-centric and elaborately engineered (modeled) kind of business intelligence, which is today’s best practice. However, the book takes you into ”higher dimensions” such as shifting the gear towards information rather than data, coping with big data, the insides of people’s heads and much more. In that way “unintelligence” becomes a very positive thing – the way to go. As the subtitle states, we are talking insight and innovation. The current BI is really not that intelligent in many ways –more than I have room to describe here. However, here follow some of my “Likes”.

The foundation for Business unintelligence consists of information, process and people. Look, no Data! We have sort of tended to treat data and information as synonyms, which they are not, and we did not think about BI as a (set of) process(es). And people? We assumed they were the ideal rational agents. “Gut feeling” is certainly not a BI deliverable… Quote: “Only people can assign meaning and value to information. … Process exists to meet the needs of people” (note 1).

Our target is the “biz-tech ecosystem”, not the EDW and not BI. This means that us information-driven people should take the business people in our hands and move up to the enterprise level. Yes! Leading companies leverage IT to change and improve their operations. I have certainly been preaching business first and information-driven for some years now. Information-driven reengineering has indeed had some successes. Think retail, for instance. But it was not always called BI, maybe just supply chain or similar.

Early on Barry Devlin concludes that information is the precursor of data, not the other way around. Data is a subset of information, which is optimized for computers, not people. People consume information, not data. He distinguishes between soft information and hard information. Hard information is modeled and fenced in by metadata like in a relational database. Soft information is all of the rest, including documents, big data and so forth. He introduces the “Modern Meaning Model”, which I really like, and which you will have to get the book to learn more about that. What could “informal information” be, for example? By the way, which do you like best “context-setting information”? Or “metadata”? I am a full-hearted context-proponent.

As we move to include soft information combined with process-mediated data (hard information), we need to improve on several dimensions. Very importantly that of trust / reliance. Not even the worldwide web people have found the good solutions to this dilemma, yet.

If you are into BI and EDW, you believe in the great wall between operational and informational, right? Well, tear it down. Timelines and consistency both have a right to exist, and we must support both of them equally well. This leads to redundancy, SOA, federation etc.; in short the “spaghetti monster” architecture to some degree. This is challenging most BI-people’s belief about “the single version of the truth”. But we can, and we will have to live with controlled versions (contexts) of the truth… And we will have to get rid of “… other commonly-held assumptions (which, TF) have contributed to the relative stasis we’ve seen in the data warehousing world over the past two decades”.

Big data, typically “machine-generated” data, are here to stay. Quote: “In contrast to BI, big data immediately demands that operational and informational processes are combined”. Following a very good primer on big data technology Barry introduces a 3-pillar logical model of information: machine-generated, process-mediated (think ERP) and human-sourced (think documents). This leads to a host of different technologies beyond relational and MOLAP; HADOOP, NOSQL, semantics, SOA, event processing and more. Very useful is also his categorization of types of information in the flow: measures, events, messages and transactions.

Barry Devlin gives a very good overview of business development going from process-orientation to human psychology and much more (note 2). This leads to a sense-and-respond model for decision making, which is what unintelligence is in support of. I think it is very useful to think of BI (in the future) as such a sense-and-respond process, which is a key point in Barry’s book. Why did we not think about that before? The implications are quite large and combine operational, informational and collaboration. You might say that Barry in many ways wants all of the enterprise architecture, not just the data-side, which we used to think was sufficient for driving an enterprise. This makes SOA-style integration very important. I agree 100 %. We need to find a new division of labor between ETL, SOA and workflows – in the service of agility. In short ETL becomes part of “assimilation”, which /very) basically is load or link. And again, context rules everything.

Much of the soft information “resists the classification and categorization approach”, which is what we have previously called modeling. After a good recap of the history of BI and information management as well as privacy, Barry cross tabulates structure/content versus timeliness/consistency and arrives at an important conclusion that compound information plays a key role in the future. Have a look for yourself, in the book. After a brief and good overview of modeling (featuring business concept maps, fact models and ontologies etc.) he introduces the overall concept of the “Information Nugget” and describes the (considerable) consequences to what today is known as modeling. This leads to a good discussion about knowledge versus information. All data modelers should pay attention here (note 3)!

Quote: “… decisions are a human process, occurring in the mind.” And “… a little rational thought will convince that the reality of decision making if far from rational…” Barry explains why by way of a very good overview of cognitive psychology , the meaning of meaning and puts a lot of emphasis on colloboration. Our soft side is very powerful, and we use soft techniques (e.g. intuition) all the time! As a person working with concept structures and meaning (see, I am very enthusiastic about his “m3 Modern Meaning Model”, which is right to the point and very forward looking. The fact that we (BI people) have been bounded by the concept of rational decision making explains why BI and knowledge management has had far less impact on decision making then we like to think. He then describes his “adaptive decision cycle” and his “team decision making model”., which leads to a good description of the dimensions of the People layer in his overall conceptual architecture. I love it, there is plenty of room for improvement here for tool vendors!

The book concludes by summarizing his architectural proposals. They come in two levels: The conceptual level and the logical level. The conceptual level has “… the ability to outline the story that links business thinking with the process and information required for implementing it…”. The logical level is “… aimed squarely at IT and providing the functionality needed”. This is enterprise-wide, it spans the “great operational-informational divide”, so it is not just an upgrade of BI and data warehousing. Very important is also the human aspects, where we need to understand perception, cognition and collaboration in order to provide features such as a “Personal Innovation Platform”, “Information Orchestrators” and “Interaction Choreographers”. The recent history is full of examples of rational decision-making gone mad – take high frequency trading, for example. Dr. Barry Devlin sets out “… to find the common ground between rationality and intuition, intelligence and insight, stability and innovation, business and IT”. And he does this extremely well! A tour-de-force indeed!

Recommended for both business people interested in business architecture and efficient decision-making and for many kinds of IT architects, be they information, data, BI, data warehouse or enterprise architects. And people working with business development and business analytics, and big data people, and ….

The book may be preordered at Amazon

1) About people and meaning, see my discussion
2) Barry Devlin also mentions Design Thinking very positively, see a brief version
3) Concerning Concept Map vs. traditional modeling, se my discussion