Monday, April 5, 2010

How to Incrementally Read Anything

When I first started using Supermemo, it appeared that Incremental Reading was a valuable feature, but I never really used it.

Once I began experimenting with Incremental Reading (Successfully) and adopting the Incremental Reading mindset when I look at learning material, I am now fully convinced of its superiority for nearly all of my intellectual needs. It is light years ahead of its time, and like the core concepts that power Supermemo, eventually it will likely be utilized on almost every level of the educational system (However many years that takes). Until such a time comes, we can rely on the quirky program Supermemo to accomplish this task in a basic way. (I have my own vision of 'the perfect implementation of Superemo,' but I'll save that for some other time)

The only problem is that Incremental Reading in Supermemo works only with a pure electronic text that can be highlighted, copied and pasted. What if you have a pdf file, a physical book, magazine, etc.?

I have recently adopted a crude but (At least for me) workable solution for using the incremental reading mentality to process non-electronic books (Real books), pdfs, magazines, etc.

Simply create a new topic and use it as a bookmark for whatever you are reading.

For example: If I am reading a new Popular Science magazine (That I cannot yet get a digital copy of), I create a new topic within Supermemo and I title it: "Popular Science - April issue". I take the actual magazine and read it until I get bored, have to do something else, etc. If I find an interesting passage, I transcribe the sentence or paragraph into that topic and create an extract of that passage. When I stop reading I make a note of that in the topic; for example, If I stop on page 28 on the third paragraph, I change the topic to say "Popular Science - April issue; Resume on page 28, paragraph 3."

I then place the magazine in a box next to my computer labeled "Incremental Reading." When I see the topic "Popular Science - April issue," I resume reading. Create extracts only of interesting material, read until I stop, change the topic bookmark to reflect my progress, place in Incremental Reading box. Repeat until the magazine has been read.

I realize the proposed solution would not be ideal for all situations, but considering the limitations of the current technology, it will have to do. Also, your reading material (and a highlighter) can travel with you, and you can "save your progress" when you get back to your Supermemo pile.


  1. That's an ingenious way of dealing with a technological shortcoming! ;)

    What is the typical size of each chunk of text you incrementally read? I've already read the Supermemo site's articles on IR but, since I use a different SRS which doesn't have that feature, IR still puzzles me a bit. As far as I was able to understand, each piece of text will enter the reviewing cycle, right? So, how do you rate each reading? Do you always treat it as a "success"?

    Another thing that confuses me is that if one's reading something a bit more linear (like a novel, for example), the precedence of each block of text must be respected (right?), but I can't figure out how that can be harmonized with the reviewing cycle (if one fails at least one card, that card will be out of sync with the rest).

    Anyway, these are some of the issues that make me think about IR and how the guys behind Supermemo dealt with them. By the way, you're doing great stuff, you're going through untraveled paths and keeping us updated on how you're doing, which is great!

  2. Incremental reading is not graded. When incremental reading, you will see a block of text in the same way you might read a web page or a book. You read that block of text until you can no longer continue, then you click "Next Repetition," and you are shown another piece of text. Do the same until you have no more texts left to read. While reading, If you find something that can easily be converted into a flashcard (For example: "George washington was the first president of the US" becomes "Who was the first president of the US? George Washington), then you create the flashcard. Everything else that isn't a flashcard is deleted.

    Again, when reading something linear (Like a novel), you're not grading yourself, you are simply reading until you are tired. Also, when reading a novel, you only read one chapter at a time, so you don't have to worry about incrementally reading chapter 4 before you read chapter 2. If you find a passage that inspires you for whatever reason, you can highlight it (Extract it, in other words) and eventually make it into a flashcard (I've done that a few times).

    Incremental reading is not a simple process, and it is hard to articulate an easy-to-understand description as to why it is so useful and effective.

  3. Hey LittleFish, you really have a great blog. Because of your blog, I am on the verge of trying supermemo again after two failed attempts.

    My main problems on the first try which lasted about 6 months was that I tried to use big precompiled collections. As a consequence I added two much knowledge to fast at the beginning. The biggest problem was, though, that most of the knowledge in those collections was not relevant for me or to be more precise, since I did not compile the knowledge myself I couldn't value the information. After around 4 months I got frustrated and stopped adding items and two months later I stopped using it at all.

    During my second try, which lasted about 4 months, I concentrated on items i created mostly via incremental reading. My main problems were that I never established a good routine for dealing with repetitions and adding new knowledge. But the most frustrating issue was that even with my own items I could not estimate the value of that item when I created it. I was incrementally reading a text extracting some items that I found to be useful and after one month I was reading that text again because I had not extracted the right information. This is frustrating because creating items costs a lot of time.

    Summing up my two failed attempts, I have come the conclusions:
    First, if your really use supermemo, it takes a lot of time off your daily schedule (1 hour at least, mostly due to adding new items). Since I have never put that much effort into any other learning method I am wondering whether they would be more efficient, considering you could save the time that goes into
    software use and reformulating items.
    Second, the physiology of learning roughly values information according to frequency of usage and emotional attachment to the information itself or a situation it was used in. In incremental reading you have to value information upon first contact. If you decide that it is useful you put a lot of effort into it. The main problem is that real usefulness nevertheless comes from applying the knowledge in real world. So if you apply it often you remembere it without supermemo. I am not sure whether I made my point very clear but you could come to the conclusions that valuable information is stored anyway and therefore supermemo is useless.

    Any thoughts on this? It would also help me, if you could say a little bit more on what kind of information you have processed incrementally so far and how you estimate usefulness.

  4. Littlefish, thank you very much for your reply, I think now I understand much better how IR works. Those issues had been pestering me for quite some time, and had never been answered until now. Thanks! ;)

  5. Thanks Littlefish. Somehow I learned IR months before I even learned how to add flashcards individually, and I agree it is light years ahead of its time. Like you, I mainly use it for non-work-related things. It works better with lower stress levels and absence of deadlines, with more of a "browsing" mindset, like when surfing the web.

    Although I agree that it will enter into the education system, I feel I should actively help this process. I do not mean just by showing my friends, which I already have, but on a wider scale.

    The thing is, it's so alien to most people that it's hard to find an entry point into their lives. What do you suggest?

  6. It took me a while to understand how and why incremental reading is a good idea. I don't know how it could be introduced in an appealing way to society at large. Perhaps if Google made their own version of it and made it very easy to use?

  7. Hello,
    I am struggling to get a proper approach with incremental reading. I would like to study for a big exam which covers 300 chapters of different subjects. Every chapter has 6 or more sections. Could I make you some questions to know how to deal with this using incremental reading?
    If you prefer you can answer me with a private message as this could be a particular case with little interest for the others.
    Anyway, I am devouring your just discovered blog and squeezing every piece of valuable advice you write. Thank you for this.

  8. Sure, I would love to answer any questions you have about incremental reading. If you don't want me to post the questions that you ask, I simply won't publish the comment(s) you make. What were you wondering about?

  9. Thanks Littlefish. The point is how to organise these 300 chapters I have to do. So far what I do is I import the full chapter and then I extract each section also fully. This phase is without reading the information, just organising it. So this way I have a mother topic as chapter and child topics for sections. Then I dismiss for reading the chapter because it is covered by the extracted sections. Is it better to import the 300 chapters in one go, or in small lots?

    Then, I start reading each section with the order given by supermemo, and then I have to proceed to extratct the paragraphs. It seems to me with this way, I never cover the full chapter in a sequential way from beginning to end. Is this correct?

    Once I am reading sections, I proceed to extract paragraphs and afterwards proceed with clozing.

    I imagine, the best way will be to combine reading and memorizing. Am I right?

    Any other comment or help will be welcome.

    By the way, I am using 2006 version. Is it worth to upgrade?

    Thanks again, and good luck for this new year.

  10. I have delayed posting this comment because I can not yet respond to it. I have noticed there are areas of improvement I can make when it comes to using Incremental Reading, and I have been trying to improve in these areas. Once I have conclusive results, I will post them. Sorry for the delay.

  11. No problem. I appreciate.


  12. I've been using incremental reading in a different way than in the past. What has changed my viewpoint is this:

    The main point that stands out to me is this: Passive review is sufficient for stuff that isn't top-priority (Will cover in a test soon), and it is beneficial to passively review material in topic form for quite some time. You don't necessarily have to go straight from topic/reading material to flashcard. The response post will likely say this but in a more articulate and detailed way.

  13. Hello,
    I have been using sm and incremental reading but I have an important question about the algorithm or method. I have added hundreds of articles and my idea was to go through them with a first review and extract paragraphs which would be repeated till memorizing. But the program behaves very different than expected, and instead of getting articles I have not read yet, it repeats topics again and again. This way I advance very slowly with the new material that I would like to read for first time.
    I do not know if I should change some options in the program, or use Postpone, Priority or Forget tools.
    Or maybe it is the way it is.

    To have a rough idea of my needs, I expect having 20.000 topics to be memorized in 7-8 months. Initially I have 1.000 articles.

    Hope you can give me some help because this is stressing me.


  14. Hi

    The lack of PDF and LaTeX importing in SuperMemo sucks. For math, I have become super-skilled in HTML formating and Unicode. But some CSS is not handled well with the engine SM uses.

    The lack of PDF can be solved like this: PDF -> CHM -> HTML. There are converters for that. A CHM -> HTML converter that worked and is free is
    This software was able to handle a very large and complex PDF (a medical textbook).

    All PDF to Word converters I tested are bogus and ultimately fail at large, complex documents.

  15. Thanks anonymous for your hint on chm -> html.

    I imagine to convert pdf ->chm you use acrobat itself. Any other software?

    The best pdf to word converter for pdfs was abby finereader, but it is no perfect.

  16. Littlefish, thank you for your article!
    This is really very exciting to read that! The IR work isn't that easy to comprehend, but now it's much clearer.
    Thanks again! ;)