[Feature Request] Difficult Words algorithm on the web same as App

At the moment, the difficult words algorithm on the web is merely an all-typing review for 20 words that were marked as difficult. If wrong, it does a multiple choice until you get it right.

On the (Android) App however, it goes through a 3-step-test cycle for just 4 words.


  • limit the number of difficult words session on the web to about 4-10.
  • make it a three or four step cycle, e.g.
  • 1st step: all-typing -> if right apparently not a difficult word, so cycle is finished
  • 2nd step: present the entry and immediately 3rd step: multiple choice test
  • 4th step: 2nd multiple choice test
  • 5th step: all-typing
1 Like

I’m fine with the way it currently is. If memrise changes it to that, then i’d like it to be optional. The sheer number of times I got something wrong due to typo would make the android system incredibly annoying for those words. Also, it becomes a problem when you’re reviewing hundreds, or even thousands of words, and could end up getting well over 100 words falling under difficult words. It would just make it even more time consuming.

I would agree - the Android algorithm for flagging words as ‘difficult’ is ways too aggressive, which means you get substantial numbers of words flagged as difficult and needing multiple revisions which really should not be there.
Also, it would be good if words did not get flagged as difficult whilst learning them? I mean, yes, I don’t know that word or how it is spelled exactly, that’s why I’m learning and repeating it? (duh)

Making something optional or configurable is always my favorite.

Though, if you want to have it faster, then the suggestion to have step 1 preceding the current 4 steps helps you, because it will exit when correct, or not really?

And if the answer in step 1 is wrong, well, then there is a good chance that it was flagged legitimately.

Yes, I think it is sometimes to aggressive, but sometimes too little for me. Especially when learning, indeed ‘duhhh…’

Why don’t one of you make a feature request for a less aggressive difficult word flagging, or to make it configurable/optional: e.g. normal, aggressive and relaxed :slight_smile: