Recent threads made me understand that Skritter treats single-character words very differently from multi-character words. And apparently there is no difference at all between a single-character word added to a list and a single character that only happens to appear in a multi-character word.
I would say it's very confusing and counter-intuitive. How about simplifying it? In the real world you care only about words. You are going to write down a word, and either you remember it or you don't. It really doesn't matter if the word has one or more characters. It doesn't help if the word is a single character that you can write perfectly as a part of another word. You may still forget that it is the word you want.
Wouldn't it be easier if there was only one "forgot/so-so/got it/too easy" button, after writing the whole word? And if we treated single-character words in exactly same way as multi-character words?
I'm not sure if it is doable with the current architecture, but it would be definitely easier to use and understand, and it would make Skritter test exactly what is needed it the real world: writing actual words.