The experience of using supermemo for 2 years
- There will be plenty of trial and error in learning supermemo's interface, as it is unintuitive and inefficient. It's not difficult to use; however, as an example, know that there is a hierarchical structure for organizing questions, named the knowledge tree, and if you import items from a text file, it won't be obvious where it will place the items. It is predictable however; it will drop them on whatever category your Tools toolbar selects, but you could just as easily think that it could be dropped in the same category of your currently displayed item, or in the category selected in the knowledge tree. Regardless, if you don't mind having to learn about supermemo this way, it's not too much of a problem.
- Supermemo can corrupt your data, but the authors know this, so they provide a function to check and repair for errors at File -> Repair collection, which also allows you to backup. In fact, in the version I'm using, Supermemo 15, usually there will be a miscount of items after I add/delete a few items in the knowledge tree. Additionally, about 3 times, I got a few words in a few questions dissappearing, and I don't know what triggered that, however it was just for a few questions that were ordered next to each other, and reverting to an earlier backup that was just a few hours old fixed it. It's possible the lexicon rebuild (which is part of the rebuild function) caused the issue, but even if that was the case, repair and backup often, as it only takes a few minutes and it protects your collection.
- Spaced repetition works, even if it doesn't feels that way the day when you do the repetition. Even for difficult material, you will notice you remember more and more with each repetition.
- The one thing at which spaced repetition is the best at, is long term memory. For every other aspect of learning, there is something better: priming (recalling material several times in one session, this also provides a modest cognitive improvement with repeated use) is better for placing things in the short term memory, cognitive performance is best improved by exercising, behavioral therapy, diet, rest, and meditation, plus lists are best handled with mnemonics. However, the way supermemo is built, it tends to push you towards improving these other areas.
- It will take months, if not more than a year, to be efficient with spaced repetition in the short term (few weeks) vs massing (single massive study sessions, and review before the test) with supermemo. Formulating questions (most freely available questions have poor formulation), typing and editing them, adding them to supermemo, doing the first review, and fixing errors in questions takes time, while with massing you just read and recall (if you want, otherwise it's even faster at the cost of lesser learning) is more than 20 times faster for me. This is the experience of a medical student, who can tackle 4 pages daily with supermemo (each page is roughly 40 items), but could read almost 100 pages in a single day; note that I don't use incremental reading (for reasons that I won't discuss here to conserve time, but newcomers should definitely use incremental reading) which alleviates the problem, and that I'm guaranteed to know over 80% of the pages entered in supermemo over a few weeks, while I will only have a superficial knowledge even in the short term (maybe even less than 20%) of the pages I massed, and it's not sustainable to do high-volume massing for more than a few days, but for a test in a few days, massing works better because long term recall is not what is important (it shouldn't be that way, but it is).
- Only add something to supermemo if you understand it, but it's fine to add it if you haven't memorized it. If something is poorly formulated, edit it on the spot even during reviews if it doesn't takes more than 5 minutes.
- You won't remember unordered lists of more than 3 pieces of information, unless your brain can somehow group the all information in 3 or less chunks. It's not the same being asked to remember 6 numbers grouped in 3 pairs, than remembering 6 unrelated numbers. If you have an unordered list, give them an order; this is not a tip, or an improvement, it is a *necesity*. I have tried to do that from every angle I can think of, and no approach works for more than a few repetitions, on unordered lists of more than 6 items that cannot be grouped. It's better if the order makes sense, and even more if it's related or useful, but the vital thing is that it has an order.
- For ordered lists, it is much better to use mnemonics than cloze with chunking. If you do not know what cloze is, don't worry; the point is that you should use mnemonics. Specifically, look into the method of loci paired with visual linking; when you research what these are, the concept might seem strange, but it works. Note that it took me 3 days to learn how to use the method of loci correctly, and longer to refine it, and it takes about 3 times (and even more at the beginning) more work than learning a simple item, but it's by far the best way of remembering lists.
- The final drill is important for newer and poorly memorized. It is also the hardest part of using supermemo, as you're always tackling your hardest items.
- The best way I have found of handling interference (recalling something different from what you want) is to make a new question that asks for the difference between the concepts in conflict, even if it would be a weird thing to compare. The point is to force yourself to recall both concepts AND differentiate them. Multiple choice can achieve this but there are methods to get them right that avoid recalling all the concepts or differentiating them, so the usual questions are best.
There is much more, but I have spent too much time already.
You might expand on the point why you do not use incremental reading. It seems like an ideal took for an overloaded medical student, and those who take a pass usually harbor some misconception that prevents them from making a good use of this sensational tool.
See: Criticism of incremental reading to find some food for thought.
Original poster's reply to comment
These are the reasons why I don't use incremental reading:
- Much of the material I use comes from non-electronic sources (oral transmission or books)
- Most of the electronic material is not in HTML; I could convert it to HTML format, and use incremental reading, but when I started using supermemo the tools available (tools not related to supermemo) produced poorly formatted HTML
- I use a specific program to edit text (Vim); with it, and importing questions from text files, and I'm very efficient at creating text-only questions
- I've honed my question-formulation skills over time, and I prefer not to use questions generated by software. However, I don't know the current state of question generation, so things might have improved since I last tried it, or I might not be using the question-generating tools as well as I should
I have no doubt that as incremental reading support in supermemo improves, and electronic text technology and content increases, incremental reading will eventually become the obvious choice for newcomers to supermemo. I suspect that in 1 or 2 decades, text books might be distributed as wikis, which are the perfect format for incremental reading.
With all your reasons above, you should still love incremental reading: if you can type/edit questions and answers in text fast, you will love editing short texts even more if a short text can generate two or more questions. With each cloze you save on typing and formulation. Details you can fix incrementally. If you are proficient at formulating good Q&As, you will find adapting cloze deletions even easier and faster. Your productivity at item-generating stage might double. Your learning efficiency might increase too due to cloze redundancy and reduced forgetting.
(Not Coz) Is there any report of someone mastering a subject using Incremental Reading? (like an aspiring doctor/a medical student). How is the adherence to IR by SuperMemo users today (November, 2014)?
There are some reports and links here: http://super-memory.com/help/il_full.htm We do not have any statistics on the use of incremental reading, but the flow of questions by mail seems to indicate that this proportion is increasing. In 2000, IR was a subject of ridicule as an unnecessary ballast in SuperMemo. Now it might be one of the reasons people choose SuperMemo rather than other apps (incl. IR-less SuperMemos).