Three Layers of Evidence

System

Three Layers of Evidence

The Zettelkasten note-taking method has made book writing and writing scientific papers easy for hundreds of years already.

Read the full story here

Will

Great deep dive into creative knowledge work. This is the reason I stick around this forum, to be exposed to nuggets like this. I love the actionable flavor of the post. Thank you @Sascha. You and @ctietze are an inspiration to up my game and improve my thinking practices. I'm certain I'm not alone in my appreciation for these gifts.

The Zettelkasten Method is the concrete manifestation of the abstract principles of good thinking practices and knowledge creation.

Eurobubba

Hypothesis: drowning in swimming pools and electricity consumption (for air conditioning) are both driven by extended warm weather.

Darryl

Great post @Sascha. Much appreciated. The clear explanation of the three layers, how they relate to one another, and the practical examples were very instructive. Being able to see how the synthesis layer varies depending on the data and the interpretation of it (which both remain constant), was a big help. (It seems like you're really dialing in your interpretation and application of the ZK method. Can't wait for the course!)

Sascha

Thanks for your kind words.

@Eurobubba To many factors. Heightened wealth and prosperity could contribute as well.

ptohver

Thanks for a very informative post. I'm wondering, which method would be better to use: having phenomena in a separate note from the interpretation, or having them in the same note. On the one hand, it seems more flexible to have them in separate notes - I could theoretically have several interpretations of the same data and several trains of thought based on the same source data. On the other hand just writing notes that state facts I've read without any interpretation seems a bit dry. Then again, the Kant example resonates with me: there may be source data I don't even know how to interpret yet, thus having a purely phenomenological note makes sense. How do you approach this? Is it plausible to have a combination of both: sometimes having total separation of the three layers, sometimes integrating them into one note?

Sascha

If you> @ptohver said:

Thanks for a very informative post. I'm wondering, which method would be better to use: having phenomena in a separate note from the interpretation, or having them in the same note. On the one hand, it seems more flexible to have them in separate notes - I could theoretically have several interpretations of the same data and several trains of thought based on the same source data. On the other hand just writing notes that state facts I've read without any interpretation seems a bit dry. Then again, the Kant example resonates with me: there may be source data I don't even know how to interpret yet, thus having a purely phenomenological note makes sense. How do you approach this? Is it plausible to have a combination of both: sometimes having total separation of the three layers, sometimes integrating them into one note?

When I have a source, I first try to be very acurate with describing it. The value is the filtering and the compression. You should need less words to make the same statement better.

I seperate those layers not necessary by putting them in seperate Zettel. But when I have more than one interpretation I make this step to keep my Zettelkasten clean.

ptohver

Thank you!

Sascha

@ptohver said:
Thank you!

You are very welcome.

brunohanai

Hi!

I often refer back to this post. I really liked it, and it helps me to remember some important concepts.

Context:
I read a book about communication and was introduced to a concept named "clarity over comfort". It refers to the idea that it is important to communicate precisely, even if it can be uncomfortable.

My doubts:
1. Is this a "layer 1" idea? Like "Concept: Clarity over comfort (source: author X opinion)"
2. Or this is a "layer 2" idea? Like "(In a specific work situation) it's important to be precise even if it may bother someone."
3. Or is this both "layer 1" and "layer 2"? A note for the concept and another note with the interpretation about when and how to use that concept?
4. OR, (in your opinion) should I use this not-backed idea as a start point to research the topic and create a note only after that?

It would be nice to receive some ideas from you.

alkhan

Then, is it appropriate to say that the relationships between the first, second, and third layers are "increasing generalizations"?

For example, I have two "pattern" notes:

"Partial context before learning resulted in less comprehension and recall than full context"

• People who were shown a contextual image before hearing a passage once were able to comprehend and recall significantly more than people who heard the same passage twice. [@bransford1972Contextual]

"Context before learning a passage improved comprehension and recall better than more exposure"

• Subjects who had partial context did not show significantly greater recall and comprehension than those who had no context of the passage. [@bransford1972Contextual]

So my interpretation was:

"Prior knowledge must serve as semantic context to aid future learning"

• Having prior knowledge does not guarantee better comprehension and recall. That's because prior knowledge aids comprehension only when it can serve as a conceptual base for new knowledge. Specifically, it must first "become an activated semantic context" where new knowledge can "hang onto." [@bransford1972Contextual]

Other examples

Description: Having partial context did not significantly improve comprehension and recall than having no context
Interpretation: Coherence is required to form semantic contexts from prior knowledge

But, how do I synthesize them? Say that I'm going to use all the interpretation notes above:
"Prior knowledge must serve as semantic context to aid future learning" and "Coherence is required to form semantic contexts from prior knowledge"

Would the note below be a good enough way to synthesize the interpretations, assuming that this note is going to be actively developed in the future?

---

20210617184509 Σ Textbook Reading Workflow

Choosing your materials

• Treat secondary materials as entry points for primary sources [[20210617185135]]

Aim for coherence when reading

• Prior knowledge must serve as semantic context to aid future learning [[20210617184626]]
• Coherence is required to form semantic contexts from prior knowledge [[20210617184719]]

---

Obviously, that was a mock example, but would that be roughly how you'd implement it? I really don't know if I'm implementing it right, despite re-reading and trying to implement this for a few weeks.

A concrete example (in Zettels, rather than in principle) of implementing the three layers would help a lot.

alkhan

(Apologies for the double post)

@Sascha Here you said: (my emphasis for context)

The next layer of evidence is the interpretation of data patterns. The phenomenological layer meant you describe what you saw. Here, you interpret why you saw it.
[...]
Now the concept of causality can come into play. You can explain the correlation above by saying: Drowning people in swimming pools generates amplifiers for the nuclear power plants in the U.S, and that’s why when more people drown in swimming pools, more energy is generated in nuclear power plants.

Going down to the examples:

Phenomenon: Serotonin is part of the physiological substance that constitutes status. Cooperations and alliances are integral part of status.
Interpretation: Serotonin is a regulator of negative emotion and emotion in general can be optimized via dominance and cooperation.

[...]

Phenomenon: One study showed that some training can increase the relation of fat to carbohydrate you use for up to 24 hours.[^jamurtas2004] Another study showed that in overweight people, fasted training in the morning results in a ad libitum reduction of calories reduction of carbohydrates.
Interpretation: There is an acute effect on the metabolic flexibility that is distinct from the general metabolic flexibility one has and that is used as a model in current science.

[...]

Phenomenon: Feature Requests are truly that. Users normally express their wishes in the form of a specific app behavior. E.g.: The Archive should allow for multiple tabs. The Archive should allow for multiple windows.
Interpretation: Feature Requests depend on the individual representation of the problem. They are the concrete manifestation of more abstract problem. User say “I want multiple tabs and multiple windows!” rather than “I have difficulty to handle complex projects with a lot of notes and layers. I think that multiple tabs and windows would help me.”

I'm confused about "interpreting why you saw the pattern" and "the concept of causality" in these examples. Do they simply mean that the interpretation layer is the one that explains the pattern, or is it a generalization of the phenomenon?

Overall, I figured that this may be a better question to sum up everything:

What are some good personal prompts we can use so we can go from a lower to a higher layer?

Would it be appropriate to ask:

1. Phenomenon → Interpretation: "Why did this happen?"
2. Synthesis: "So what's the practical implication of these things?"

Sascha

@brunohanai said:
Hi!

I often refer back to this post. I really liked it, and it helps me to remember some important concepts.

Context:
I read a book about communication and was introduced to a concept named "clarity over comfort". It refers to the idea that it is important to communicate precisely, even if it can be uncomfortable.

My doubts:
1. Is this a "layer 1" idea? Like "Concept: Clarity over comfort (source: author X opinion)"
2. Or this is a "layer 2" idea? Like "(In a specific work situation) it's important to be precise even if it may bother someone."
3. Or is this both "layer 1" and "layer 2"? A note for the concept and another note with the interpretation about when and how to use that concept?
4. OR, (in your opinion) should I use this not-backed idea as a start point to research the topic and create a note only after that?

It would be nice to receive some ideas from you.

Four. Neither an observed pattern (regularity) nor an interpretation (at this point). But I suppose that the author gives empirical reasoning behind his advice (e.g. "I observed many people and the best people did this and that. So, do this and that."). The mock sentence of mine has both an observed pattern and an interpretation (here, in the form of advice).

brunohanai

@Sascha said:
Four. Neither an observed pattern (regularity) nor an interpretation (at this point). But I suppose that the author gives empirical reasoning behind his advice (e.g. "I observed many people and the best people did this and that. So, do this and that."). The mock sentence of mine has both an observed pattern and an interpretation (here, in the form of advice).

The thing is: I really like that book, but, yes, you are right about the author's empirical reasoning. I think that idea was a piece of good advice, but I agree that I can go deeper.

Many thanks!

Annabella

@Sascha Forgot to comment on this post, but I guess it's never too late. So, thank you so much for this post. Really wonderful advice. It clearly separates what happens in a Zettelkasten in a very satisfying way.

Also, thank you for your video on processing empirical studies and the layers. Your explanation cleared up doubts that I had.

However, two doubts remain so far. Mind helping me to clear them up too?

Firstly, what do I do if I want to interpret something based on prior interpretations? E.g.: Imagine the following items:

• Item 1

  • Pattern. How to deal with a backlog.
  • Interpretation. The approach works because of X, Y, and Z.

• Item 2

  • Pattern. Work constantly on problem sets.
  • Interpretation. The advice treats problem sets as backlog. That's why it's so efficient.

In item 2, I want to interpret the pattern by using my interpretation in item 2. I guess I could use the synthesis layer. But then, it seems kind of... redundant? E.g.:

• Interpretation. The advice treats problem sets as backlog. That's why it's so efficient.
• Synthesis. Indeed, it works because it treats them as a backlog. [Visualize connection to interpretation note from item 1]

And then the other doubt. To keep the related notes together, links are useful. But, what do you put in the context for the link? I see only two situations: Links to a layer above or below. See the "diagram" below for reference:

Pattern <-> Interpretation <-> Synthesis

For links to the layer below, I've been using the abstract, sometimes paraphrased.

Then, for links to layers above, it depends on the context. E.g.: If I'm explaining why something works, I may make an argument that more or less says:

"This works for multiple reasons. Firstly, X.[[Link to interpretation note]] And secondly, Y.[[Link to another interpretation note]]

But I'm curious about how you tackle this.

Sascha

Mh. It really depends on the actual patterns and interpretations at hand.

But the pattern in item 1 looks more like an product of synthesised interpretation. Any "How To" is already a synthesis of an obversation of a pattern, an interpretation (e.g. hypothesis on cause and effect) and the merging of the interpretation and a goal.

Abstractly:

1. X and Y correlate. (perhaps you looked at some studies)
2. X is cause of Y. (some more studies, proposition of a causal mechanism etc)
3. Y is good.
4. Do X because it causes Y and Y is good.

Keep in mind that Y cannot be scientifically derived but only religiously or ideologically. (Having a dogma)

I don't think you need to be so overly formal about this: You need a place in which so observe and describe the patterns, a place in which you interprete these interpretations and a place in which you synthesis interpretations.

It does not matter if it is all on one note, multiple notes or else. At least, not in general. Probably, it matters for the specific use case.