Post

Knowledge, Information, and Understanding: KL02 Analysis

A revised critical analysis of the KL02 subtitle's classification of knowledge.

Knowledge, Information, and Understanding: KL02 Analysis

Chinese version: /zh-hant/knowledge/knowledge-information-understanding-analysis/

Status: Revised on 2026-09-18 after direct review of the user-supplied subtitle file. This note separates the video’s pedagogical framework, its empirical illustrations, and its factual attributions.

Source: Source Record: KL02 Knowledge Versus Information Subtitle

What the video actually argues

The video is not only presenting the familiar four-part DIKW mnemonic. It makes three linked claims:

  1. Information versus knowledge. A contextual statement such as a historical event or one employee’s monthly sales is information. A stable connection behind multiple cases—such as a function relating sales data to pay—is knowledge, also called a rule or algorithm.
  2. Structure versus fragments. Memorised items are compared with bricks; the reusable relations among items are compared with the structural drawing of a building. Knowledge must therefore be organised rather than merely accumulated.
  3. From the world to a decision. It adds event, representation, and rule/form before data, information, knowledge, and wisdom. In the egg example, a real cooking event is measured under conventions, recorded as a table, explained with heat-transfer reasoning, and used to choose a cooking procedure. The video explicitly says that reasoning can also run from an existing model down to application.

This is a stronger and more precise account than treating the video as merely saying “facts are bad.”

What is worth keeping

1. Separate an observation from its representation

The video correctly directs attention to a basic distinction. An egg is put into water; 4:30, 95 C, and “soft yolk” are selected symbols or descriptions of aspects of that event. They are not the event itself. This helps expose omitted variables such as egg mass, initial temperature, water movement, probe position, and the operational definition of “soft-centred.”

2. A table does not explain itself

A table can reveal a pattern, but it does not by itself establish a causal mechanism, a law, or a procedure that will transfer. The video’s movement from an observation table to a relation that supports a changed-case prediction is a useful learning move.

3. Search for relations and boundary conditions

The useful question behind the building metaphor is: What remains related when a relevant condition changes, and where does that relation stop applying? That question discourages memorising a final answer without its conditions.

4. Knowledge needs organisation, but organisation needs a purpose

The IKEA analogy usefully highlights retrieval: unlabelled, unconnected material is hard to find and use. The needed correction is that organisation is not only storage. A good knowledge structure must support explanation, prediction, decision, or performance on a task.

5. The video admits both induction and application

At 00:08:01–00:08:07, it says knowledge formation can proceed bottom-up through induction and top-down through logical application. This is an important strength. A model should guide what is measured, and data should test and revise the model.

Conceptual limits

“Information” and “knowledge” are not natural opposites

The video chooses a narrow technical use: only reusable regularities, functions, or algorithms count as knowledge. That is useful for asking for a model, but it is not the only defensible use of the word. A supported fact, the meaning of a term, a procedure, and acquaintance with a particular case can also be knowledge. They can be ingredients in, rather than rivals of, a reusable model.

Therefore the corrected contrast is not:

information = worthless bricks; knowledge = the only valuable structure

It is:

facts, terms, examples, and procedures + relations among them = resources for a usable model

The seven items are not one homogeneous ladder

Event, representation, rule/form, and data concern an observation and measurement system. Information concerns an interpreted statement. Knowledge concerns support, explanatory structure, or competent use. Wisdom concerns a decision relative to goals and values. They are connected but not simply increasing quantities of one thing.

A measurement rule is also a possible kind of procedural knowledge. It can be placed before data only when “rule” is specifically limited to a convention that standardises measurement.

A relation is not yet understanding

Writing a formula, naming a rule, or repeating an algorithm shows possible recall. Understanding is shown when a learner can use the relation to explain, predict a changed case, transfer it appropriately, and inspect or revise its assumptions. Mastery requires reliable performance under defined conditions as well.

“Real-world contact” is necessary but not sufficient

The final explanation—that learners forget because they receive principles without enough real-world phenomena—is too simple. Contact with examples can be valuable, but durable learning also needs retrieval, feedback, comparison of cases, corrective practice, and time. Conversely, a person can have many experiences yet form a wrong model without these checks.

Errors and unsupported jumps in the subtitle

1. The historical DIKW attribution is wrong or seriously confused

At 00:04:47–00:05:22, the subtitle says that “the founder of operations research, Russell Ackoff” proposed DIKW in 1989; at 00:06:17–00:06:28, it reverses this, claiming DIKW was not first proposed by Ackoff but by David Williams in a paper titled Models, Metaphors and Symbols for Information and Knowledge Systems.

The second attribution cannot establish an earlier origin: the cited article was published in 2014 and itself discusses a longer, contested history of the DIKW model. Ackoff did publish From Data to Wisdom in 1989, but the origin of the hierarchy is disputed and cannot be settled by calling either person its single inventor. Calling Ackoff the singular “founder of operations research” is also an imprecise honorific rather than a careful historical description.

2. The seven-layer “complete version” is presented too confidently

The subtitle says the “complete version” adds event, representation, and rule/form. But no single, universally accepted complete DIKW taxonomy exists. Different authors use different definitions and sometimes reject the hierarchy itself. The seven-way schema can be a useful analytic tool; it should not be presented as settled taxonomy.

3. The first historical example is materially misleading

At 00:01:33–00:01:42, the subtitle treats “Homo sapiens left Africa 100,000 years ago and defeated Neanderthals in Europe” as a clean information example. The wording compresses a complex, multi-wave history into a false causal story. There were earlier dispersals from Africa, the major expansion associated with later populations occurred much later, and modern humans and Neanderthals overlapped and interbred rather than simply one group “defeating” the other. It may be a grammatical sentence, but it is not a reliable historical statement.

4. The “fictional story” example makes an invalid jump from explanation to recipe

At 00:02:12–00:02:31, the video moves from the claim that shared fictions can organise large groups to the advice that anyone who wants to mobilise many people can fabricate a story. Even if shared narratives sometimes help coordination, this does not make fabrication sufficient, necessary, ethical, or predictably effective. Institutions, incentives, trust, coercion, material resources, and social conditions also matter. The named entrepreneur examples are anecdotes, not a demonstration of a universal rule.

5. The egg procedure is internally inconsistent and physically under-specified

The example begins with an egg placed in rolling water and a temperature condition above 95 C (00:07:02–00:07:24). It then concludes with water “not exceeding 65 C” for four minutes thirty seconds (00:07:46–00:07:52). Those are different thermal conditions, and the transition is never explained.

More importantly, Fourier’s heat-conduction law does not by itself provide a recipe or protein thresholds. A cooking prediction needs boundary conditions and parameters: egg size and geometry, initial temperature, shell and contents, water temperature history, convection, and an operational texture target. Protein changes are not one sharp universal switch. The exact time-and-temperature instruction is therefore not supported by the stated reasoning; it should not be used as a general recipe.

6. The English-learning contrast is an overgeneralisation

At 00:09:03–00:09:24, the video contrasts years of classroom English with a child in the United States who “completely masters” English within a few years. The contrast identifies a real issue—language is learned through meaningful use—but “complete mastery,” amount of exposure, quality of interaction, literacy, correction, motivation, and learning conditions are all left undefined. It cannot show that application to real events is the sole or decisive explanation of learning success.

A corrected working model

Use the video’s distinctions, but remove the false hierarchy and add performance tests:

world event → representation and measurement conventions → data → contextual information ↔ conceptual model

conceptual model → explanation, prediction, transfer, and revision → demonstrated understanding → reliable, independent performance → mastery

goals + values + constraints + supported model → decision

This is a network with feedback loops, not a compulsory staircase. Facts and records are not discarded as “bricks”; they are tested, organised, and connected so that a learner can act and revise a model when reality disagrees.

This post is licensed under CC BY 4.0 by the author.