Dismembering complex topics before or after learning the structure of knowledge

From SuperMemopedia
Revision as of 22:15, 2 March 2020 by SuperMemoHelp (talk | contribs) (→‎User Followup)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

From: {{{From}}}
Country: {{{Country}}}
Sent: Feb 29, 2020
Subject: Dismembering complex topics before or after learning the structure of knowledge


In the example item on the immune system it is advised that the item be revised incrementally as comprehension grows. However, this seems like a very costly process to do item-by-item, when it could be done on a topic before forming cloze deletions. Intuitively, I would have reformulated the passage into the following four extracts on the first repetition, and then upon being presented the individual extracts, I would make clozes on phrases bolded here:

1. (Immunology) Inflammation is produced by eicosanoids and cytokines.

2. (Inflammation) Eicosanoids include prostaglandins and leukotrienes.

3. (Inflammation) Prostaglandins cause fever and vasodilation.

4. (Inflammation) Leukotrienes attract certain white blood cells.

I would then add context as needed to individual items if they prove lacking upon future repetitions.

Is this approach sufficient to allow the structure of relationships to emerge in mental maps over time, or does it risk creating a fragmented knowledge structure?


You present two alternatives:

  • neat paragraphs to produce neat clozes (high formulation cost, low re-editing cost)
  • bulky paragraphs to produce bulky clozes (low creation cost, high re-editing cost)

Which method works better for you, you will see by using a few examples of both approaches. Your neat approach will bring better results tomorrow because of the effort needed to re-organize knowledge. The long-term outcome is less certain and may depend on the type knowledge and on you. The messy approach is safer in the long run because you will remove context only when you feel it is well established.

Only you will know what overload you experience, and what delays. Each time you depart from the perfect spaced repetition schedule due to delays you will need more context. At the same time, the more you delay the less the cost of re-editing.

The balance of preference, overtime, may switch from neat approach to cheap approach.

See also: Knowledge darwinism to understand the extra value of "messy" items.

User Followup

Thank you for the quick response!

To clarify, are you saying that high retention of atomic clozes is a good indication that one understands the overall structure? If not, how can I test my structural understanding when practical application through projects isn't an option?

I am concerned that despite good retention, I may be invisibly losing sight of the big picture and interrelationships, and thus losing applicability in the process. If retention is a good indicator of structural understanding, reformulating individual items seems disproportionately expensive compared to the recall advantages I would enjoy from doing so, as I already enjoy better retention than predicted (~97% retention with a 10% FI).


  • high retention of good clozes will be a good assurance only if it provides the full skeleton of necessary connections. This structure is hard to predict or reason out. This is why experimenting, and letting cloze evolve is very helpful
  • standard repetitions are a reasonable test on the quality of knowledge. When the structure starts crumbling, you will start forgetting. You will see things seem to be going wrong and you need supplementary knowledge
  • if you do no get a practical test on your knowledge, you will also test it in further reading/learning (if this is not the case, perhaps the knowledge is not as useful as it seems?)
  • 97% retention is great news, however, you need to see if you can keep it up like that for longer