neoncube
I usually choose a different mode based on the practice type. If I'm practicing Portuguese to English (so I see the Portuguese word and need to pick or write which English word matches), I will do multiple choice. Then I do typing for the other direction. This is to save time, and make because it gives me a somewhat easier test to start building association and familiarizing myself with words without forcing my brain to remember without any context or clues.
What I was thinking is a system which does pick words randomly, but weights this random selection towards words which are most often wrongly selected. I think this will mostly bias towards similarly spelled words, but it would also get false friends and words with similar meaning (bigger and greater, longer and taller, etc). I think this would be useful because for a lot of words the main difficulty is not really to learn and remember what they roughly mean, or how to spell them, but the difference between them and similar words.
But I do see how multiple choice can be a crutch/excuse for poor discipline and lack of motivation/enthusiasm, I am absolutely guilty of using it this way to some extent, so I understand not wanting to use limited developer resources on it.
