On Data, Information, and Knowledge

My employer, like many large organizations, has the hots for an idea called "knowledge management". Their idea is that if you can gather information from resources throughout the enterprise, collect them in one place and make them accessible, somehow an "enterprise knowledge base" will make people smarter and therefore make the enterprise run more effectively.

When I first heard this idea, I was immediately reminded of a quote by Computerworld commentator Frank Hayes that I saw in his September 25, 2000 column entitled "Conventional IT wisdom." The quote was:

"Data isn't information. Information isn't knowledge. Knowledge isn't manageable."

I remember being awestruck by such a simple yet well-conceived idea. At the time I was struggling with trying to extract some type on usable knowledge from hundreds of gigabytes of network traffic captures. A customer wanted to understand network traffic patterns in support of some server consolidation efforts. As with many such efforts, the customer only had the vaguest idea of what they really needed and could only express it in broad terms that eventually pointed to the idea that "I'll know it when I see it." I was on a fool's mission (they were holding out for the magic bullet that would make their problems go away) and didn't realize it until I read that Frank Hayes column while waiting for a meeting at another customer site.

Since then, I've spent much effort in my work trying to differentiate and distinguish between data, information, and knowledge. I've found some really great sites that explain the principals of Knowledge Management theory in excruciating detail, like this one. Instead, I opt for a simpler model.


I think of data as a basic atomic unit. In its purest form, data can't be subdivided, but what we think of as a single piece of data is often many pieces tied together. As an example, take the following piece of data:


Not very interesting, is it? In general, raw data is dull as it doesn't really tell us anything. Is it a temperature? A price? A rate? Without further clarification it's not very useful.


By combining our raw data with other data, we can attempt to move toward information. Look at this:


Well, that at least a little more informative. We at least know we're talking about money now. A little more data yields some real information:

Bob's net worth is $40.

While this is certainly informative about Bob, we still don't know enough for it be actionable. You can store it, manage it, and massage it, but you can't reliably take action upon it. And that's the problem with information: it is seldom anything you can use to choose a course of action.


The piece of information above isn't really useful because we know nothing about Bob. If Bob is a 4 year-old, a net worth of $40 isn't a real shocker. But what about the following:

Bob is 65 years old, has no job, no retirement, and his net worth is $40.

At this point, we probably know more about Bob than we really want to, but there is stil no real actionable path for us to take. This is still just information. We must add more information to complete enough of a picture to be actionable:

Bob, my father, is 65 years old, has no job, no retirement, and his net worth is $40.

This is still just information, but it leads to knowledge. It tells me that my family and I have some hard decisions to make regarding dear old Dad. Now we're getting into the area of knowledge. Note that I didn't say we had knowledge, because knowledge is something that exists in the individual human mind and cannot be clearly and succinctly stated in sentences.

So what is knowledge? Philosophers dating back at least to Socrates never really figured that one out. Knowledge is a sticky, gooey concept. We all have an innate knowledge of what it is, but are hard-pressed to put it into words. My best idea is that with knowledge, the total is greater than the sum of the parts. Data and information make up knowledge, but much of the data and information we know doesn't really contribute to the sum of our knowledge (like that pesky Post Raisin Bran song I learned as a kid that won't exit my memory).

Getting back to the "Bob" example, we can see how Bob as my father can integrate into my knowledge structure. Assuming that I love dear old Dad, I innately know that Dad may need help and see that my life may need to change to make sure that Dad lives well in his old age. On the other hand, I may integrate this with my knowledge of Dad to realize that he's a deadbeat and perhaps I should distance myself from him. How, and whether, this new information grows my personal knowledge base is controlled by what I already know and how the information I know is connected within my consciousness to create my "personal knowledge base."

I chose to talk about Bob as my father to make the information something that would be close enough to me to force it's way inside. But what about the previous statement "Bob is 65 years old, has no job, no retirement, and his net worth is $40."? This can also become knowledge, depending upon the makeup of my personality and personal knowledge base, or it could be just another disconnected factoid (or idle gossip) floating around in my head. Here are a few examples of how this information might be differently integrated into my personal knowledge base depending on my personal makeup:

As best as I can tell, one person can't directly impart knowledge to another person. You can write down all the information you know about a topic in sentences, but another person reading it will not obtain your knowledge of the topic by reading the sentences. They'll develop their own knowledge by integrating the information with their own system of information, knowledge, experiences, beliefs and prejudices, and it will doubtless be different from yours. This is why, as Mr. Hayes stated, "Knowledge isn't manageable."

So What?

I go back to my previous statements about many Knowledge Management efforts and customers looking for that magic bullet. They suffer the same fallacy. Whether you're trying to consolidate information to obtain the dream of "enterprise knowledge" or grasping for the Eureka! epiphany that will solve your problems, you're most likely trying to take a quick and easy shortcut to knowledge by simply gathering up some data and information.

The sad fact is that it isn't easy. If obtaining knowledge were as easy as gathering a whole boatload of facts and factoids, CNN and the like would make us all geniuses. That's why school was often so hard. We were given data and information and then tested on data and information, with a little test of knowledge thrown in once in a while (remember those nasty "word problems?"). It was up to us to develop knowledge and be able to use it. That's why one of my high school teachers was fond of saying "I can teach you, but I can't learn you." A teacher may impart information to you, but learning happens in your own mind.

There's no great lesson to all this, but I have developed a few ideas that help me navigate these problems:

Doesn't sound very promising, does it? Well, this is why human history is so rife with war and conflict. At the least, we can try to use our understanding of data, information, and knowledge to reduce the amount of conflict that comes with basic human interactions, and that's something.

Disclaimer. Any and all opinions expressed on this page are solely those of the author. The author accepts no responsibility for the accuracy of data provided on this web site, nor does the author accept any responsibility for the use and /or misuse of information provided on this web site. It is the full and sole responsibility of the user to determine whether and/or how best to use the information provided within their own circumstances, and it is the full and sole responsibility of the user to accept any and all benefits and/or consquences of their actions. Please see my Privacy Policy.

Copyright 2001-2009, Tim Sharpe. You are free to use this information for personal, non-commerical use without restriction. All rights reserved for commerical, organizational, or government use. Questions or comments to tim@beaststwo.org. Flames to /dev/nul.

[Python Powered]
All the Perl that's Practical to Extract and Report

12517 visits since July 14, 2008