Data, Information, Knowledge, and Wisdom: Our Pragmatic Ignorance

A common way to model the relationships of data, information, knowledge, and wisdom is the DIKW pyramid. Starting with its base, each concept builds on top of the former. The gist is that we require data to form information → we use information to gain knowledge → and we use knowledge toward developing wisdom. This model is one doorway to understanding the relationship between these concepts.

For workaday pragmatic purposes, common understandings of those terms and their relationships may be worthwhile. But for our acceptance of a pragmatic ignorance, this model will break down when we examine the meaning of any of those terms in depth.

This post isn’t in any sense an exhaustive look at DIKW (there is a lot out there about that). I’m just expressing that I think it’s a little funny that we need to fein understanding to use information.

DIKW and a Common Understanding

Data, in the common conceptualization of the DIKW model, are something used to create information and as such, data themselves lack context. Context comes on the scene once we arrive at information, which is essentially data that we imbue with meaning. Generally, that meaning is arrived at through various organizational techniques. Rubin in his Foundations of Library Science[1] describes knowledge and wisdom as much the same. They’re both a “cohesive body” of information applied to some human end, the difference being that the end to which wisdom is applied gets a sort of value judgment, it must be beneficial to the world. Similar definitions are not hard to find. For example, Rowley (in “Where is the Wisdom that We have Lost in Knowledge?”) looks into wisdom and defines it through the application of knowledge toward some end, wedded to a value judgement that it must be the “…what does the most good…”[2].

These types of explanations deal with two things: data or knowledge. We both relate information to data, and wisdom to knowledge by adding some other concept to the mix. In the former case it is meaning and in the latter it is a value judgement (virtuous application). The addition of the concepts upon one another fits the image of the DIKW pyramid.

Breaking Down the DIKW Relationships

A different definition of data is a string of basic letters or numbers, which don’t necessarily have to be meaningful but do have to be the “value” of some attribute[3]. This view, explained in Meadow and Boyce’s book on informationr retrieval systems, requires that the symbols also be organized before they can be considered information. They also require information to be “…data that changes the state of a system that perceives it…”[4], which means the addition of a person’s brain or a computer, to perceive it.

Buckland distinguishes between being informed and information-as-thing (1991), which is a useful way of setting workable boundaries for information relationships. By information-as-thing, he means an object from which a person could receive information. These objects include documents (electronic or not), which could be composed of text, images, sounds, etc. He includes data in this type of information, characterizing it mainly as a computer’s means for storing records.

In both Buckland and Meadow’s cases we can talk about the data and information relationship even though neither offers a foundational definition for data. I’d like to say that we can do this because the definitions they provide are lodged in symbols that we already accept (at least in our common cultural backgrounds) or recognizable “documents.”

Buckland and Meadow diverge when it comes to the relationship of a person with information. In Meadow’s case information requires a person (or at least a computer, eventually used by a person) whereas Buckland looks at a state of becoming informed. Though they may draw boundaries in different places, they both recognize some sort of human intervention. For Buckland, knowledge can be considered another type of information, namely information that someone has perceived.

Diverging from DIKW

Frické accepts the usage of “data” in a similarly pragmatic way as Meadow or Buckland explained it but considers information a subset of data. This is reasonable if we consider information as a selection of data, which satisfies some sort of query or is otherwise functional for a purpose we set.

In this sense, knowledge becomes a set of instructions that a person can transform from data. Frické identifies some problems with these relationships. For example, he shows a great deal of inference is required to answer “why” questions, which the relationships in the DIKW model cannot support. Eventually he argues that all data is information but not all information is data, while knowledge and information have the same meaning. For wisdom he does not veer much from the other notions, in which it is knowledge put to use, and done so with a positive value.

Although these common understandings of data and information are useful toward common purposes, further questioning causes the terms’ relationships to break down.

Popular examples of data tend to be things like numbers or letters, which are not organized in words or in the context of meaningful statements. However, such examples don’t work as the basic element or source from which we build with information, knowledge, or wisdom. We can do a thought experiment, which shows that these examples of data come already bearing a meaningful context.

Here it is: imagine that you were raised on a remote island with no exposure to languages that use the letter “a” much less, alphabets. A Canadian statistician lands on the island and after striking up a friendship with you, starts to explain (in your own language) his work. What happens when he tries to explain his studies on the number of people that selected “a” in a multiple choice survey? He’s given you “a” as a datum but “a” does not exist for you. It would be neither perceivable as a datum nor useful as a datum. You can’t even talk about “a” as a datum without some sort of context for what it is. This can be extended to a computer’s on/off functioning to argue that the bits used to represent an “a” also require some sort of contextual recognition that those ons and offs represent that “a”.

The thought experiment shows that we cannot truly offer examples like “a” to define data (as we commonly think of it) because they still result in something requiring a meaningful context. The pragmatic definition that information is data imbued with context, gets us no closer to finding the source of what we can consider data. Within our commonly accepted, pragmatic usage of these terms, we’re prevented from describing a true relationship of data to information.

Without a definitive basis to define data, information, knowledge, and wisdom, we end up in muddy waters trying to relate the terms. Taking a pragmatic approach and focusing narrowly on their common usage or even a technical definition, which allows for the pre-acceptance of symbols, is what lets us live with models like DIKW. However, living with these models and definitions is willful ignorance—an odd basis for wisdom.


  1. Rubin, R. (2010) Foundations of Library and Information Science (3rd ed.). New York, NY: Neal–Schuman Publishers.
  2. Rowley, J. (January 01, 2006). Where is the wisdom that we have lost in knowledge?. Journal of Documentation, 62, 2, 251-270 p. 257
  3. Meadow, C. T., Boyce, B. R., & Kraft, D. H. (2000). Text Information Retrieval Systems. San Diego: Academic Press.
  4. Meadow and Boyce. Text Information Retrieval Systems. p. 39

Leave a Reply

Your email address will not be published. Required fields are marked *