Summary: To maximize the speed and the benefits of learning with newer versions of SuperMemo, you will need to learn more about knowledge management and the flow of knowledge in the learning process (rudimentary knowledge of SuperMemo is needed to understand this text)

Most important points and skills:

• creating a To Do category based on rich text components
• examples of content-related knowledge processing that goes in parallel with the learning process (gray inserts: Step 1, Step 2, and Step 3)
• constructing occlusion tests
• understanding the multidimensional flow of knowledge
• knowledge processing options

In this article I would like to discuss the optimum strategy that would maximize the inflow of quality knowledge into the learning process at minimum time. It should be noted that long-term memory is still the bottleneck of the learning process. Consequently, your chief cost in learning will be paid to repetitions (i.e. the effort of keeping knowledge in memory). Yet a noticeable proportion of your time is spent on importing knowledge to SuperMemo and formulating it for optimum recall. Actually, this fraction of your learning effort will often be most painful and inhibitory. Many would-be-users complain that they do not employ SuperMemo because it takes too long to type in the learning material to the program. This article is an attempt to suggest a set of solutions (other than the obvious one: get ready-made learning material)

New reading options and a wider application of rich text and HTML components in SuperMemo 99 through 2002 are likely to influence the way in which you acquire new knowledge with SuperMemo. In early versions of SuperMemo for DOS, all knowledge was represented as questions and answers in limited-length text fields. This required lots of discipline in formulating items and there was no intermediate stage between the outside sources of knowledge, and the knowledge stored in SuperMemo

Today, SuperMemo makes it possible to import entire articles in electronic form, extract most important fragments, gradually process these fragments at speeds that reflect the importance of particular pieces of knowledge

A collection in SuperMemo becomes a conglomerate of pieces of knowledge of various priority, status and degree of processing. Consequently, SuperMemo 99/2002 adds the following new qualities to learning:

1. Faster learning: You can quickly introduce rough notes into SuperMemo and later gradually convert them into well-structured knowledge. This is prioritizing via formulation. Less important material may remain unstructured and, as such, less well remembered. You will see how passive notes gradually fade in your memory and how their individual components will need to be reinforced by formulating specific well-structured items. You will make such reinforcement decisions on the one by one basis depending on the importance of the fading material and the degree of recall problems (see examples below)

    

2. Better understanding: You can gradually build understanding of complex phenomena. You will start from importing a set of unrelated or even chaotic articles describing a studied complex problem. You can then gradually proceed through this material and gradually build the understanding from basic or simple facts towards details or more complex components of knowledge

    

3. Creative thinking: You can gradually resolve contradictions in a body of knowledge. If you study a new branch of science with a number of contradictory findings and models, you will find SuperMemo quite useful in grappling with contradictions. By introducing the chaos of new research into SuperMemo, you will gradually locate contradictions and strive at building better and more consistent models in your memory. You will build understanding, resolve contradictions and ultimately creatively discover new truths about the learned material. You will quickly lean towards theories that are better supported by research findings. Those supported poorly will be less firm and will often cause recall problems. You can reformulate or delete these. As you process the material, the creative build-up will be gradual and you will intelligently manage the flow of knowledge in your collection
   

SuperMemo 2002 encourages you to (1) collect rough notes from various sources, and (2) introduce these notes into the learning process (even before these notes assume the ultimate well-structured shape). Such rough notes will often not comply with the two pivotal principles of SuperMemo:

• active recall principle: passive presentation of knowledge is ineffective. Instead of reading that Theodore Roosevelt won the Nobel peace prize in 1906 for mediating in the Russo-Japanese war we should be asked Who won the Nobel peace prize in 1906 for mediating in the Russo-Japanese war? or Which year did Theodore Roosevelt win the Nobel peace prize?, etc.
• minimum information principle: the act of recall in a single repetition should be accompanied by minimum added effort related to reading comments, wordy questions, etc. Each recall, i.e. each repetition, should refer to a single fact or rule: a single basic memory unit

In this article I would like to stress that the approach in which you store rough notes and text fragments in SuperMemo does not have to contradict the principles of effective learning. Indeed, the new tools provided in SuperMemo 2000 encourage this approach and assist you in efficient knowledge management. SuperMemo 2000 helps you build better understanding of complex material and creatively resolve contradictions in knowledge you learn

Here are some typical ways knowledge enters your SuperMemo collection:

1. Articles: If you find an interesting article on the web, CD-ROM encyclopedia, or any other article available in electronic form, you can copy it to the clipboard, and paste it to SuperMemo with Edit : Add a new article (Ctrl+Alt+N). Note that you need to enter the professional level to make this possible (File : Level : Professional)
2. Items: If you want to add a single important piece of information belonging to a well-defined category, you can select this category in the category combo-box, choose Add new and type in the contents of the item. Usually, you will formulate a question and an answer. Occasionally, you will import a file to illustrate your item (e.g. if your category template includes an image component, you can import a jpg or gif file with Import file). In earlier versions of SuperMemo, adding items was the primary way of expanding your study material
3. Notes: If you want to quickly introduce some rough notes taken while watching TV, reading press, during a lecture, board meeting, or meeting with a friend, you can define a separate To Do category that will allow of the fastest possible transfer of rough notes into the learning process (see yellow inset below to find out how to set up a To Do category). Your To Do category might use a question and answer template based on Rich Text components (RTF format). Using Rich Text will help you process knowledge gradually depending on the priority and availability of time. If you frequently add sounds or images to your items, your To Do template might include a sound component and one or more image components. In SuperMemo 2002, you can add a new note with Edit : Add to category : Note (Alt+N)

Depending on the way you introduce knowledge to SuperMemo, its further processing will look quite different:

1. Articles: If you have pasted an entire article from an electronic text, you will use options such as Remember extract, Task extract, or Queue extract to generated smaller and more manageable pieces of information. For a detailed discussion see: Devouring knowledge, Reading the Internet, and Incremental reading
2. Items: If you have added a well-structured element to its target category, it will most likely need little further processing except for minor corrections, editing, etc. A subset of these elements will naturally be rescheduled, removed from the learning process, dismissed or even deleted; however, the scale of these actions will be incomparably less than in the case of the next category, i.e. rough notes
3. Notes: If you have added a rough note to your To Do category, it will require further processing that will include:
   • splitting into smaller fragments
   • rewording
   • generating cloze deletions
   • converting to a simple question-and-answer format
   • moving to appropriate categories (and perhaps merging with the category template at the same time)
   • introducing the item to or removing it from the learning process, etc.

The gray insert presents a couple of examples of real-life situations, which could result in adding rough notes to your To Do category. Your choices will naturally depend on your interests and needs. You may opt to use SuperMemo to learn professional knowledge only, or languages, or proverbs, jokes, family names, history, or knowledge that you have long forgotten since your high school years (see: User Comments to better understand the knowledge selection issue)

Your notes will probably be added to the pending queue; however, once your pending queue grows long enough, you will want to immediately memorize those notes that are particularly important 

Once you memorize your notes, they will come back to you for review in the learning process. Depending on the importance of a given note you will most likely choose one of the following:

1. If the note contains very important material, you will immediately convert the note to a set of well-structured items (see further examples in gray insets)
2. On a less important note, you can postpone the review by choosing a passing grade (as in standard repetitions). Naturally, each time you postpone processing your note, the interval before the next review will grow larger at each repetition. Soon you will notice, that delicate fabric of memory structure that keeps the note in your memory falls apart as a result of passive review (only active recall can ensure good retention). If you postpone processing the note 4-5 times, the interval might be long enough for you to come to the conclusion that you should either delete the note (you seem not to have time to process it) or process it only partly (e.g. using cloze deletion options on most important keywords)
3. Occasionally, you can manually decide when to schedule the next review by choosing Jump interval (Ctrl+J)
4. You can shift the note back to the pending queue with Ctrl+R. This way, it will be processed only after you pass all the remaining items in the pending queue
5. You can dismiss the note with Ctrl+D. This way you will come back to the note only if you locate it with search (e.g. Ctrl+F) or stumble against it accidentally
6. You can delete the note if you believe it is not important enough or if you think that you are not likely to ever convert it to well-structured items. Keeping notes indefinitely in the learning process is a poor strategy that is likely to pollute your repetitions and reduce your focus (you may find yourself reviewing less relevant old notes that leave little traces in your memory while you could better use this time for retaining quality knowledge)

Processing notes is not an all or nothing process. If you notice that upon third or fourth review, important components of the note seem to fall apart in your memory, you can, for example, pick a single keyword and use it with Reading : Remember cloze to memorize one cloze deletion that will effectively restore this particular semantic relationship in your memory. You could also use Reading : Remember extract to extract only the most important sentence that will later be subject to the same review process

Please have a look at examples below to see how rough notes listed above get reworked in the first step of processing (or later):

Reasons for processing a note	Results of processing
You consider your Kiribati note a facultative contribution to your general knowledge and keep it rather for passive review. However, you notice that upon fourth review, you do not seem to have kept an important association between Kiribati and the international date line. You decide to use Reading : Remember extract to memorize part of your note (the original note will display the extract in different font as shown on the right) The extracted note will require further processing in the future. It does not form a well-structured item. Most of all, it does not ask any question. It does not meat active recall principle	KIRIBATI. The Republic of Kiribati consists of 33 coral atolls and islands in the Central Pacific Ocean that straddle the equator and the international date line. The soil is poor, but the islands are well covered by coconut palms that provide copra, the main export. Population 75,000. Most of the people on the other islands live in villages of thatched huts. The official language is English, but it is rarely spoken away from Tarawa Atoll. Most families fish and grow bananas, breadfruit, papaws, and taro
You consider broadband important enough to process your definition with Reading : Remember cloze. Your goal: remember the word broadband and remember the three most popular broadband technologies. As a result you generate four cloze deletion items (the original note will display deleted keywords in different font as on the right)	Broadband - new technologies of fast access to the Internet such as DSL, cable access, or satellite broadcast (starting at 0.3 Mbps)
Upon the fifth review you noticed that you are not likely to recall Mike's son's name. You decide to use cloze deletion again (Reading : Remember cloze) and generate the item shown on the right	Question: Mike married Monica in 1999. They have a son ...(name) born later the same year. He quit his video rental business and now works for a friend in his well-known restaurant Astoria Answer: Robert
You decide to give up the definition of blancmange and reconcile yourself to just being aware that blancmange is a kind of dessert. You manually process your items to the form shown on the right	Question: blancmange Answer: kind of dessert (jelly-like)
Remembering the map of Kosovo seems important to you and you decide to convert your note to a series of questions (illustrated with a map as Answer on the component pop-up menu)  (alternatively, you could use an occlusion test to recognize areas obscured in the picture, e.g. with a circular shape component)	Questions: What is the location of Macedonia relative to Kosovo? What Yugoslav region separates Bosnia and Herzegovina from Albania? What does Macedonia border with in the east? Answer:

It is easy to notice that the process of formulating items above has not been completed. As soon as individual notes come for review you may decide to generate more extracts, cloze deletions, occlusion tests (see the insert below for a occlusion test recipe), etc. You will also convert many items to their final form and move them to their target categories. In a well-managed learning process, all notes and extracts will ultimately be used to generate well-structured items and dismissed (for future reference) or deleted (if no longer needed). As all new items are built from clozes and extracts as children of the parent extract or note, the actual deleting of a rough note should take place only after all children have been moved to their target categories

With passing time, depending on your priorities, more and more changes will be noticeable in your maturing learning material. In our exemplary set of notes, after another month or two, the following changes might occur:

Reasons for processing a note	Results of processing
You decide to simplify your Kiribati extract and generate two cloze deletions that will help you remember that (1) Kiribati is made of a group of atolls and islands and (2) Kiribati spreads over the international date line	Kiribati consists of 33 coral atolls and islands in the Pacific Ocean that straddle the equator and the international date line
You move your broadband definition to your Internet category, reformulate the item and apply the default category template (here: blue text on the white background)	Question: Internet: What is the collective name of new technologies of fast access to the Internet (e.g. DSL, cable access, or satellite broadcast)? Answer: broadband
You simplify the question about Mike's son, move it to your Private category, and apply the default category template	Question: Mike married Monica. They have a son ...(name) born later in 1999 Answer: Robert
Your blancmange item seems complete. You only move it to your General English category and apply the category template	Question: blancmange Answer: kind of dessert (jelly-like)
You notice that your Kosovo map contains some interesting visual clues about facts you do not know. You decide to add questions about these facts and delete the original Kosovo note as fully processed	Questions: What sea does Albania access? What province is Pristina a capital of? Does Slovenia border with Albania? Answer:

Multidimensional flow of knowledge

The flow of knowledge in SuperMemo proceeds along the four major pathways that make up a complex multidimensional flow of knowledge: 

1. Quality flow - flow of knowledge from articles and notes, through extracts to cloze deletions and well-structured items
2. Pool flow - flow of knowledge from reading lists and pending notes towards the pool of pending elements and ultimately memorized elements. Occasionally, some elements will move to the dismissed pool for future reference. Processed articles and extracts will often be deleted for the sake of conserving space and avoiding excessively rich searches in your collection
3. Category flow - flow of elements from your To Do category to other categories such as: biology, geography, history, economics, private, Internet, etc. 
4. Time flow - flow of elements in time from shorter to longer intervals. Because of the fact that a large proportion of processing is done at repetitions, flow of knowledge towards longer intervals slows down the three remaining flow pathways 

Quality flow proceeds from articles and from rough notes, through stray images and text fragments towards cloze deletions and well-structures items based on active recall and minimum information:

1. articles enter SuperMemo via reading lists. Once you locate interesting texts in electronic form, you can import them via the clipboard using Edit : Add a new article (Ctrl+Alt+R). While reading an article you extract its fragments (e.g. with Reading : Remember extract) that later enter the review process. For a detailed discussion of reading lists see: Reading the Internet
2. extracts are best added to your To Do category based on rich text templates. Using rich text makes it easier to process extracts. Further extractions and cloze deletions leave a permanent colored-font mark in the extract so that you do not have to keep wondering which parts of the extract have already been processed. Here are the three typical ways extract enter SuperMemo:
   1. as extractions from larger articles with Reading : Remember extract or Reading : Queue extract
   2. as pasted via clipboard directly from longer electronic articles (e.g. collected on the net, e-mail, CD-ROM encyclopedia, etc.). Reading lists are best suited for meaty articles full of contents that requires meticulous processing. However, you will often read longer articles that will only sparsely contribute to your knowledge. Very often, a whole article will have zero contribution or will contribute a single paragraph. In such cases, importing it to your reading list would be a waste of your time or a waste of space in your collection (reading lists make your collections swell quickly into dozens of megabytes!). For sparse articles, select the paragraph of interest in your web browser, choose Ctrl+C (copy to clipboard), switch to SuperMemo, select To Do category, choose Ctrl+A to add a new note and paste the content of the copied paragraph (e.g. with Shift+Ins or Ctrl+V)
   3. as typed in directly from other sources (e.g. press, TV, books, phone talk, etc.). Very often you will need to type a note quickly without much time to consider the appropriate formulation for learning. Using techniques described in this article, you will later convert such notes to well-formulated learning material
3. sentences, pictures, sounds, etc. are best added to the To Do category. Once you encounter an important sentence or phrase, you will quickly paste it in to a new item or type it in (in case it was not available electronically). The same happens with pictures that illustrate an important idea. Once you locate a picture on the net, you can save it on your disk, add a new item, use the To Do Picture template on that item, and use Import file to import the picture to this new item (do not paste images via clipboard as this will store them in space-consuming BMP format)
4. cloze deletions will most likely be created from extracts and sentences still located in the To Do category. Optimally, To Do category should be rich-text based. This will make it easy to see which parts of a given extract have already been processed with cloze deletions. On the other hand, template in target categories would better be plain-text based. Text components use much less space and have quite a number of other advantages 
5. questions and answers will either be added directly to the target category in the ultimate shape or will be produced from cloze deletion items after careful analysis of the wording and formulation. Well-processed cloze deletions from To Do branch can be moved to the target category by choosing the category in the Category combo box in the Element parameters dialog box (Ctrl+Shift+P)

Pool flow

Pool flow occurs between the four essential pools of knowledge in your collection (plus the fifth pool of deleted items):

1. tasks - tasks will usually have a form of articles and article extracts placed on your reading lists. Tasks do not take part in repetitions and are scheduled for review in prioritized lists called tasklists. You pick tasks for reading with Learn : Reading list (Ctrl+F4) and remove them from the list after reading with Dismiss (Ctrl+D). Fragments of articles may enter other pools of knowledge by means of reading functions
2. pending elements - rough notes and items that await the entry into the learning process. Pending elements do not take part in repetitions and are scheduled for their first repetition in the pending queue. Pending elements enter the learning process one by one when you select Learn : Selected stages : New material (Ctrl+F2) or when you answer with Yes to Do you want to learn new material? in the last stage of daily repetitions
3. memorized elements - elements that take part in repetitions. Those elements are guaranteed to stay in your memory with probability determined by the forgetting index (e.g. for the standard forgetting index of 10% this probability is around 95%)
4. dismissed elements - elements that you consciously eliminate from the learning process and keep in your collection only for reference or archival purposes
5. deleted elements - elements that ultimately lose their appeal and are no longer needed

Category flow

Usually, the category flow is a one-step process for each item. Once an item assumes its final shape, you can choose Element parameters (Ctrl+Shift+P), select the category (Category field), and click OK

Time flow

Time flow is the most obvious flow component for every seasoned user of SuperMemo. The figure below roughly illustrates the flow of knowledge in time depending on its difficulty (understanding this graph is optional for further reading):

Fig 2. Time flow of knowledge in the learning process

The horizontal axis corresponds with the repetition number and the vertical axis represents intervals (logarithmic scale). Despite popular belief, the semi-log scale does not produce a linear graph here. Clearly the increase in the length of intervals slows down with successive repetitions. Moreover, the graph corresponding with zero lapses (black curve), results from the superposition of items with lower and faster increase in intervals (determined by difficulty). The bell-shaped curve is determined by all contributing items (below repetition number 10) and then only by difficult items or items with low forgetting index for which the increase in the length of intervals is significantly slower (above repetition 10). To see the above graph in your own collection, use Tools : Repetitions graph in the browser

Flow dynamics

The quality flow is unidirectional, i.e. you gradually move from unprocessed articles and rough notes towards quality knowledge that takes part in the learning process. Category flow is also unidirectional: from unprocessed reading branches to well-formulated category branches.

The pool flow is more complex and often multidirectional. New elements usually start off as tasks in the reading lists, rough notes added to the pending queue or important notes added directly to the learning process (memorized pool). Articles and extracts stored in the reading list (task pool) are dismembered into shorter extracts that flow into memorized or pending pool, while articles themselves are tossed into the dismissed pool upon review (usually they are not deleted until children elements formed during reading are completely processed and moved into target categories; the process that may take months). Pending elements gradually move to the memorized pool (usually at the speed determined by the availability of time for learning). Memorized elements may flow back to the pending queue (if they are considered of lower priority) or into the dismissed pool (if they are considered worth future reference but not worth remembering). Some memorized elements will be deleted if they become irrelevant or are considered no longer needed 

Processing knowledge in the learning process

You should remember that all items introduced into your learning process require endless attention in reference to their applicability, formulation, importance, logic, etc. In a well-planned learning process, it should not be necessary to review items in the periods between individual repetitions. However, when an item comes up in a repetition, you should make a quick and nearly instinctive assessment of the following:

1. Do I really need this item?
2. Do I really need to know it now? Or can I remove it from my memory for some time?
3. Is this item difficult to remember? If so, why?
4. Is it factually correct?
5. Is it as simple and clear as it could be?

Here are some typical actions you will take depending on the answer to the above questions:

1. editing the item. In case of questions and answer, you will use keys such as Q, A, or E to enter a desired text field and edit it. In more complex items you will use Ctrl+T to circle between components, Alt+click to switch a component between editing and dragging modes, or Ctrl+E to enter the editing mode
2. forgetting the item. If you think the item is too difficult or not important enough, you can postpone learning it. For this purpose, press Ctrl+R to remove it from the learning process. This will put the item at the end of the pending queue (i.e. the queue of elements that await the entry into the learning process)
3. rescheduling the item. If you know the item well or for some other reason want to manually increase (or decrease) the length of the inter-repetition interval, press Ctrl+J to select the date of the next repetition
4. dismissing the item. If you are sure you are not likely to need the item in the future, but you would like to keep it in your collection for reference or archival purposes, press Ctrl+D. Dismissed items are removed from the learning process and from the pending queue. If you create a reading list item and process it gradually (e.g. with Reading : Remember extract), you will want to leave it in your collection until all its children are processed (e.g. moved to appropriate categories). SuperMemo will automatically prompt you once the last child of such an item is moved or deleted. You will then be able to choose if you still want to keep the parent item in the collection
5. deleting the item. The key Del is very useful in cleaning your collection from garbage that results from your desire to know more than your memory can hold. In editing mode or in spelling items (i.e. at times when Del plays text editing functions), you may need to use Ctrl+Shift+Del instead of Del. Please note that deleting an element in SuperMemo will delete all its children!

Further reading:

1. Incremental reading
2. FAQ: Incremental reading
3. Reading the Internet

User comments

(Tomasz Szynalski, Poland, March 21, 2000)
Question:
My view is that this article advocates an unrealistic approach to learning! I do not see many people being able to find time to memorize names of their friends' children!
Answer:
This article does not predicate on the quality or quantity of knowledge that should be added to your collection. Those two must be determined by the user himself or herself. As for the quality, the selection of material features prominently in Six Steps to Excellent Memory (Step 2, Identify what you really need!). The examples in gray inserts should not be considered a recommendation as to the sort of knowledge you will use in your particular case

(Michal Ryszard Wojcik, Poland, March 21, 2000)
Question:
My collections are solely language-oriented. I passionately learn English. Memorizing proper names, dates, numbers and other dry facts sounds like a pretty good way to waste my time. I am not convinced
Answer:
The only rational approach to deciding which material should enter your collection is via the cost-benefit analysis. With a dose of practice, this analysis becomes a semi-automatic process, and should painlessly blend with your life. The cost-benefit analysis should serve as the sole platform for deciding what is worth adding to SuperMemo, and what is not (neglecting the fact that your learning may also be a form of enjoyable pastime). The cost-benefit criterion is: if costs of not knowing a piece of knowledge is greater than the cost of repetitions in a given period of time, add this piece of knowledge to SuperMemo (otherwise do not add it)

Cost

Cost of a single well-structured item in lifetime ranges roughly from 40 seconds to 3 minutes. At any given moment it may be approximated for a 30-year bracket, by multiplying Future repetitions (in element data) by Avg time (in learning statistics). In SuperMemo 2000, this value is already displayed for you in element data.  If your item is ill-structured (i.e. difficult to remember), this cost may bloat!!! To eliminate ill-structured items use leech analysis

Benefit

Each piece of knowledge in your collections should be associated with a tangible benefit. Only you can accurately guess the value. For example, the value of memorizing the opening hours of your gym in a given time might be approximated by multiplying (1) the probability you will choose wrong hours by (2) the time-cost of missing the gym. For example, if you believe that the probability is 25% and the cost of choosing wrong hours is 40 minutes, the cost of knowing the opening hours is around 10 minutes. In such a case, memorizing the opening hours will be cheaper than missing the gym. However, if the probability is sufficiently low or the time loss sufficiently small, you should not add opening hours to SuperMemo. For example, if the probability is 10% and time loss is 3 minutes, you are not likely to recover your learning investment! There are naturally less clear-cut cases in-between. Use your best knowledge to figure out the outcome