Point: http://www.scotthyoung.com/blog/2012/08/05/forgetting-is-good/ "Why Forgetting Can Be Good", by Scott H. Young:
> People often ask me how they can guarantee they won’t forget anything they’ve > learned. But I think forgetting isn’t such a bad thing and that trying to > avoid it completely is a loser’s strategy. > > ...I’m skeptical of the value of an SRS for most domains of knowledge. The > problem is simply that just because you’re reminding yourself of an idea > doesn’t make it useful. Useful and important ideas recur frequently, so > spaced repetition is naturally built into the process of learning > aggressively. > > People have asked me if I have a mechanism to review material from courses I > completed earlier in the MIT Challenge. Initially, I had considered creating > one, but going through the classes has shown me that it isn’t necessary. For > any given class, some ideas will be very useful and important, others less > so. If an idea is useful, it shows up in more than just one class. Huffman > coding has probably shown up in 4-5 classes I’ve done, so I’m getting tons of > repetition even though I never use a formal system to remind myself. Even > languages, the favorite child of SRS seem amenable to this approach. If you > spend most your time actually communicating, the words and phrases you > memorize are precisely the ones that come up most frequently. Perhaps just > going out and speaking a lot is the best kind of spaced repetition. The same > is true in non-academic learning. When I read a book, I try to deeply > understand it, but I don’t make any system to guarantee that knowledge is > perfectly preserved. The reason is that I know if the ideas are useful and > important, they will show up as themes in other books. > > ...When Should You Look Back? > > Often the paths and trails into unexplored territory circle back to familiar > ground. This presents an opportunity to relearn those old ideas or to view > them in a new way and get a better connected understanding of them. At the > beginning of the MIT Challenge I took a class which introduced the Laplace > Transform and, honestly, I didn’t get it. I could manipulate them well enough > to pass a test, but I couldn’t see what they were or why they were useful. > However, because they were an important idea, they showed up again and again. > Each time I circled back I got another chance to learn them more deeply. Had > they not been important, I might never have seen them again and they would > have faded from memory, but if they aren’t too important, why would that > matter? Counterpoint: http://blog.learnstream.org/2012/08/spaced-repetition-in-natural-and-artificial-learning/ "Spaced repetition in natural and artificial learning", by Ryan Muller: > He [Scott] argues that aggressively pressing forward in learning new and more > advanced material will naturally re-expose him to material from before, > making a spaced repetition system unnecessary. > > Such a result has been found for an elementary math curriculum. In _Why > Students Don’t Like School_, Daniel Willingham (another of my favorite > learning bloggers) summarizes the results of a longitudinal study: “A student > who gets a C in his first algebra course but goes on to take several more > math courses will remember his algebra, whereas a student who gets an A in > his algebra course but doesn’t take more math will forget it. That’s because > taking more math courses guarantees that you will continue to think about and > practice basic algebra.” > > ...A good *curriculum* teaches you an artificial subject in a similar manner. > It’s like a game that’s carefully set up to advance in difficulty as you use > the skills and equipment you’ve gathered on the way. Subjects like math and > science may be artificial, but curricula have undergone thousands of years of > refinement to be somewhat learnable. This is what Scott Young is relying on > when he presses forward with the MIT challenge. > > But Khatzumoto’s method is setting up an artificial world where one doesn’t > exist, where not even a good curriculum exists. He’s building his curriculum > in place. Topics like foreign languages have some options for curricula, but > why not make them contemporary and interesting by using real media? There’s > no curriculum for being up to date with the latest trends in business or > software engineering, but it’s important: you need to be able to converse > with others in that vocabulary, and you may pick up some wisdom along the way. > > Spaced repetition specifically replicates some of the advantages of a natural > environment. Memory works like this: when we encounter important things a > number of times in different contexts, we begin to learn them in the > abstract. Otherwise, we’d be totally overwhelmed by the number of abstract > concepts we could apply in any circumstance. By artificially spacing > repetition, we allow the context to vary via the passing of time. Not only do > our physical surroundings change, but the knowledge we have that can be > related to what we learn changes. > > So I don’t agree with the extent to which Scott thinks aggressive learning > makes spaced repetition unnecessary. Moreover, he’s overlooking the “spaced” > part of the concept: that *not* repeating, and allowing the context to vary, > is an equally important part of the equation. But there is an art to keeping > that environment spruce: Khatzumoto says, “I choose decks in order of > fun/priority and I delete extensively. If I’m avoiding a deck, then I go on a > deletion spree, and I keep deleting until the deck feels good again.” My own observation is that an optimally constructed curriculum *could* effectively implement spaced repetition, but even if it did (most don't), unless it is computerized it will not adapt to the user. -- gwern http://www.gwern.net -- You received this message because you are subscribed to the Google Groups "mnemosyne-proj-users" group. To post to this group, send email to [email protected]. To unsubscribe from this group, send email to [email protected]. For more options, visit https://groups.google.com/groups/opt_out.
