What to Do After Creating Extracts In Incremental Reading?

Photo by Victor Garcia on Unsplash

Incremental Extract Processing. During Incremental Reading, after extracting, just leave the extracts (mini-articles) be. Continue reading or moving onto the next element. Review the extract(s) only when SuperMemo shows it to you. Only immediately process an extract if you’re absolutely sure how to do so.


This article is about Incremental Reading’s (IR) sub-component: the way and when to review extracts. If you don’t know or understand IR, read My workflow of Incremental Reading and The Significance of Incremental Reading Part I and The Significance of Incremental Reading Part II first. Then actually use IR. This article will make zero sense if you don’t use Incremental Reading.

If I’ve convinced you that IR is the best learning/reading tool (which I wholeheartedly believe), then how to best use it, how to maximize its effectiveness and efficiency are your next priorities. In other words, optimization for the best tool you have. I’m still learning as I go so take everything with a grain of salt.

PS: Mini-article(s) = Extract(s)

What Is “Incremental Extract Processing”?

One major principle of IR is, as the name suggests, Incremental. Gradual. Cumulative. Everything is broken down into steps and parts that span across a (long) period of time. During IR, you are actively reading and extracting. One critical question is,

What do you do after creating extracts? Do you continue reading, i.e., ignoring the extract(s), or immediately process them?

Here are the official answers (with minor edits) from Incremental Reading:


After you finish reading a portion of one article, choose Learn or Next repetition to proceed with reading other articles. [Me: so you do not immediately process the extracts.]

At extract time, you are already forming passive trace memory engrams of the extracted sentence. The optimum strategy then is not to proceed with generating cloze deletions, but to move on to other elements in the queue or to other extracts in the same article (if the high priority of the article justifies it)

Return to reading the topic [Me: in this context, topic = mini-article = extract] next time it comes for review.

Use Alt+Left arrow to edit the newly created cloze (e.g. to add context clues, shorten the text, improve the wording, etc.). However, if possible, you should do such mini-jobs incrementally, i.e. on the next encounter with the clozed item.

Note that the process of descending from the source article to individual clozes may take years. The whole process is incremental and is paced by the declining traces of memory.

Incrementalism should be used not only while reading, but also in the follow-up processing and formulation of knowledge.

Your work on extracting fragments, producing cloze deletions and editing them should be incremental. In each review, do only as much work on the learning material as is necessary!

The above selections are my answers to “What to Do After Creating Extracts In IR?”

Q: “When do you deal with the extracts?

A: Not immediately after extracting. Only further process the extracts, i.e., paraphras­ing, summarizing or clozing, when SuperMemo shows them to you.

I call this “Incremental Extract Processing”. During IR, in most cases, you don’t have to and shouldn’t immediately process the extracts. Simply, do not further process, such as rephrase, summarize, turn extracts into clozed/Q&A items immediately after extracting. Leaving the extracts alone, only process them when SuperMemo shows them to you, is a better strategy. As suggested from the official answers, what I’m suggesting is inherently the “standard procedure” of IR; I’m just writing to emphasize another aspect of Incrementalism: not only for the reading materials but also for extracts.

This video Demonstration of Incremental Reading and Spaced Repetition with Dendro is a good introduction. I consider the software Dendro one implementaion of SuperMemo’s IR.

The Benefits of Incremental Extract Processing

  1. Improved Metamemory Judgments by the Power of Spaced Repetition

  2. Minimizing the Novelty Bias And Priority Bias

At its root, Incremental Extract Processing is trying to balance between knowledge redundancy, time cost and opportunity cost.

I’ve separated this portion for another article; otherwise it would be too long. If you’re curious about the rationales behind Incremental Extract Processing, please stay tuned!

Closing Remarks

The following sentence summarizes what this article is about:

Incrementalism should be used not only while reading, but also in the follow-up processing and formulation of knowledge.

I hope this article has thrown light on this subtle component of IR: yes, reading your collection of learning material should be incremental, but so does processing the extracts produced.