No Definition
As a corollary to the recursive nature of the artifact we call “knowledge”, it is interesting to note that we have no empirical definition of it. The Merriam-Webster dictionary quotes:
Definition of knowledge
1a(1): the fact or condition of knowing something with familiarity gained through experience or association
(2): acquaintance with or understanding of a science, art, or technique
1b(1): the fact or condition of being aware of something
(2): the range of one's information or understanding | answered to the best of my knowledge
2a: the sum of what is known: the body of truth, information, and principles acquired by humankind
Each of these definitions includes a word (underlined) that itself requires or implies extant knowledge. This is a circular definition, where “knowledge” is defined with respect other knowledge or to the activity of to knowing something. And then the activity of “knowing” is defined as having or acquiring knowledge. This noun-verb circular definition is not uncommon. In the definition of many labels, the associated noun/label is defined in terms of the verb/action required to identify and process the noun. For example: “reading” is defined in terms of words, and “words” in terms of the action of being read. This symmetry is not accidental.
Ultimately, people’s use of both the artifact and the action upon the artifact usually results in familiarity with the concept to the point of usability—while we may not be able to absolutely and empirically define knowledge, we do end up with a useful and usable sense of what it is and a good facility with it.
No Metric
The issue of assessing and quantifying knowledge we will address later. For the moment, we simply note that there is no such thing as a “knowledge unit”—there is no metric in which the quantity of knowledge can be definitively expressed. This is true wherever the knowledge is stored: in a human brain, in a book, in a graphical design, or in a computer program.
The latter has caused challenges in the field of software development in that the knowledge contained in the computer program is the real ultimate product of the development process. While steel-making, growing crops, manufacturing cars, and other human endeavors can quite easily measure some quantity of correct and usable physical output, in general this is not possible in the creation of software [1]. Software development does not really produce any physical product that completely, congruently, and consistently maps onto its knowledge content. Valiant efforts have been made in this area with counts of system data storage, data elements and data collections, system states and their associated events, input and output transitions, lines of functional code, even simple bit counts. As we will see later, these metrics all fail the test of measuring knowledge—they actually measure something else that the person doing the counting assumes (hopes?) will indicate the quantity of knowledge.
Any human construction activity that cannot be measured is, of course, seriously challenged if its construction progress and even its final output cannot be quantified.
FOOTNOTES
[1] Ah! but even these clearly physical artifact measurements invariably have a number of knowledge components: is the steel the right grade for purpose? Are the crops what is needed and of sufficient freshness? Do the cars work as intended--if not, what kinds of errors or deficiencies are in them?
Each of these questions highlights some attribute (read: knowledge) of the expected characteristics of the product and how much of that attribute is actually inherent in the finished product. They also hint at some process necessary to measure how much the expectation knowledge and the extant knowledge are congruent.