We cannot understand the nature of knowledge without first appreciating that any definition, any consideration, any understanding, any knowledge of knowledge is inherently recursive.
By “recursive” I mean that knowledge, as an artifact, is fundamentally self-referential. It can only be understood in terms of itself—there is no external yardstick we can use to identify, discover, collect, qualify it, or measure it without referring to knowledge itself.
Many of the offerings of this blog will either build the case for this point of view or will reference it while attempting to unpick the behavior and characteristics of knowledge.
Until then, a clarification: how do we know that something is, in fact, “knowledge”?...
The word “knowledge” is ultimately just a word. It is a noun—a semantic label—that we apply to something or some collection of things in some medium [1]. As humans, we apply such labels to anything and everything we want to think about; we examine the thing and we assign it a name, a defining textual label. The label is intended to identify and connote certain properties we think the artifact or collection possesses and the label usually places the thing in a class of (perceived) similar artifacts or collections based upon (perceived) shared properties. These properties are both identified by and assumed by the application of the label. Additionally, we expect the collection of properties denoted by the label will enjoy some general agreement about what they are among the people who use it. So, when we are looking at something—whatever it is—and assign a label to it, we are applying some criteria against which we compare the something to determine that the something qualifies (or doesn’t qualify), for the label. In the case we are looking at—of "knowledge"—we analyze the something (to be better defined as we go along) with the intent of assigning, or not assigning, this label ”knowledge” to it.
Some things will pass this test and some things will not. A rock, by itself, would arguably not qualify as knowledge, while a list of the rock’s chemical components would. The list is not the rock, it is knowledge about the rock. Even if we were to arbitrarily state that anything we consider to be knowledge is, ipso facto, “knowledge” simply by labelling it so, we are still applying a test. In this case the “test” is the trivial condition that we have applied the label. In the case of knowledge about a rock’s composition, the qualification would be more complex, perhaps a list of known and accepted chemical identifiers such as element names and symbols, perhaps definitions of crystalline structures, names, and models, hardness and density measures, etc.
So this label-qualifying test, whatever it is, consists of two things: a process or mechanism for comparison of the something against a standard, a benchmark, or some other source of criteria. In the trivial case of “anything we call knowledge is knowledge” the process is essentially null (as in no comparison to anything) and the benchmark is also null. Perhaps more correctly, it is a universe of any and all possible benchmark values given by our definition as "knowledge" being whatever we so label. More commonly, we would apply some restrictive process (meaning this applies but that does not) against a bounded and predefined set of criteria. We might say this person has knowledge of (say) physics by scoring the results the person achieves in a college physics exam. To assert that someone has some particular level of the knowledge of physics we would restrict both the process (eg., using some standardized exam format of topic, question, location, test duration, etc.) and the benchmarking of the results (ie., scoring the exam paper against a body of knowledge in some standardized fashion).
In the physics test example, we can easily see the recursive and self-referential nature of knowledge: how do we know what the process is—how do we know how to conduct and score a standardized physics test? Secondly, against what do we score it? In this example, the knowledge-evaluation process is some level of matriculated exam format which the college has adopted—that is, a standardized process has been developed (read: the knowledge of how to set up and run a standardized test has been acquired [2]) and is being applied). Then to complete the knowledge identification and assessment, the test results must then be compared against another source of knowledge. This other source is usually the body of knowledge of the domain, in this case, physics. Perhaps the source is the knowledge retained by the physics professor and is validated by the process the professor went through to become accredited. Which meant the professor-to-be must have undergone some earlier standardized test against some other body of knowledge in order to attain this accreditation.
So, whenever we wish to identify something as “knowledge” we must call upon knowledge (that has already been obtained) in the identification process and the knowledge (that has already been obtained) contained in whatever we are comparing it to.
This means that to identify, acquire [3] , measure, and qualify knowledge, we must already have knowledge. Knowledge is—must be—always defined in terms of, well, knowledge. This is an intrinsic self-reference that leads to a logical recursion and is something I will pursue at length.
Spoiler Alert: there are things recursive and self-referential systems do that non-self-referential systems do not do. One is that they often, though not always, result in paradoxes [4] . And paradoxes do paradoxical things to logic, to proof, and to our ability to define and understand.
FOOTNOTES
[1] The requirement for a medium in which the knowledge is stored is inherent in the concept of knowledge. In the example of a rock, the characteristics of the rock that qualify as knowledge are not “stored” in the rock. The rock is a rock and the rock’s characteristics are the rock’s characteristics. It is in the identification and classification of these characteristics that knowledge is created or unveiled. It is then knowledge about the rock, not the rock itself. Clearly, for it to be retained, this rock knowledge must then be put somewhere (generally somewhere other than the rock itself). It is a “tree falls in the wood” argument to assert that, if knowledge is obtained, but not retained anywhere, it is still knowledge. Perhaps so, but since the knowledge dissipates, it is of no use.
[2] And in using the phrase “standardized process” we are telegraphing that somewhere there is another body of knowledge (the standard) which someone, at some time, must have acquired, validated (against yet another source of stored knowledge), deemed valid (through some learned process) and stored.
[3] It is pretty easy to see that to measure knowledge as in an examination to confer a degree we have to have some other body of knowledge against which it is compared. Later, I will attempt to show that to acquire knowledge we also need to have knowledge. And I will try to show that the knowledge-already-possessed and knowledge-to-be-acquired are closely related.
[4] St Paul, in Titus 1:12, ascribed the statement “Cretans are always liars” most likely to Epimenides. But if Epimenides (who was a Cretan) truly asserted that (all) Cretans are always liars this would be an untrue statement (since it is said by someone who lies every time) and therefore Cretans are not always liars and the statement would therefore be true. So this statement is both false and true at the same time. It is a paradox and is, in fact, an unsolvable proposition. We will revisit this kind of dilemma. Often.