I agree with that. In the beginning words show up quickly. However, say you know a certain word quite well and have reviewed it, let’s say 10 times. How does Memrise know whether it will take you to forget that word in 2 weeks, 4 weeks, or 10 weeks? Of course, they will take an educated guess somewhere in the middle that is hopefully based on some scientific research and probably fine for 80% of the users.
But perhaps there are users who need a review sooner - otherwise you might need to spend extra time relearning the word, which could have been avoided by one extra review a few weeks earlier.
Memrise says that if you are a PRO member, Memrise adapts its alogrithm based on your personal learning characteristic. Still, how do they know what the future holds? They might know by now that I need a second review on 2 hours, a 3 third in a day and a 4th in a week. But how can they know my memory strength (or better: memory lapse) for over, let’s say, 10 weeks?
A feature where we can ask for a sooner review would certainly be great. Or could ask for all words from in a certain range, for an exam.
And yes, you can press the review button and just review all the words of the course again, if Memrise doesn’t recommend a review right now, But I have experienced that if you do a lot of reviews, then you can’t do a review on a course again. That is fine for a course I studied yesterday, but not for a set of words that I reviewed 2 weeks (or longer) ago for the last time.
To me the algorithm that determines whether you will strongly remember for long a word or not should be improved this simple way :
Words we make mistakes on, when reviewed, should always show up as a full typing challenge with full keyboard and not a shortened one with fewer letters that might more easily give a clue about the word.
Like, there are words for instance I know I won’t remember them after a few days/weeks even though they appear as watered.
And why do they appear as watered? Because sometimes, when not typing but given multiple choices, we rely on intuition rather than knowledge.
I think that when watered, words should have like a mandatory full typing challenge after a few days so that we don’t rely on intuition or luck that can be involved in multiple choices, and difficult words a weekly full typing show up until like for instance 7 times in a row (arbitrary) we don’t make any typing mistake.
It’s not that difficult to code and I’m sure it would greatly improve and strengthen our memory (although I’m aware that some people like the “no typing” courses but well, it’s their problem then ).
Memrise isn’t perfect, but they do use actual data to develop their watering schedule and determine when a multiple choice test will be more effective than a typing test for learning.
It’s not about being perfect, it is about improving.
And the topic is about words being considered as watered when we personally know they aren’t for us… So what I was proposing was a way to be ensured that words supposed to be watered really are and not be relying on intuition
).