(This turned into a bit of a wall of text (those who knew me on Memrise forums shouldn't find this to be too surprising). But I wanted to really explore these features in-depth, noting why they should be added, how they could be implemented, ramifications if they were to exist, etc. And so I hope this post helps should you ever choose to implement them.)
I've finally checked out MLWL since it seems like Memrise really will be deleting community courses soon. (Apparently, official old layout courses have disappeared as of today, according to a post on Reddit. That seems to be a sign that this is the beginning of the end…). From my experiences so far, I think MLWL's setup is comparable to Memrise's. Fantastic work so far, Neoncube!
That said, I'll make a few requests for features relevant to my own personal workflow. The first would be nice to have; the second I view as a necessity and without which, numerous old and poorly-made courses will break. (Which isn't necessarily a bad thing, honestly…). The first would be harder to implement; the second should be trivial.
Firstly, Memrise has the ability to provide hints for typing tests. Their implementation checks what you've typed in the field so far, going character by character until it reaches an incorrect character, then removes everything after the last correct character and appends a single additional correct character.
I personally view hints as a "second chance" (or third, etc. 😆) rather than each card being all or nothing. I find it lets me think about the word more deeply while keeping me from stalling for too long without making progress. (It's also extremely helpful for various advanced courses where the answer to be typed might be an entire sentence long). Basically my own workflow is as follows:
receive prompt -> think about it briefly and type a best guess ->if stumped or stuck for more than a few seconds, use a hint -> think again while using this new info and type a new best guess -> repeat until the word is completed
And with this workflow, I can usually clear even hard words within about ten seconds, give or take (possibly longer for sentences, etc.). And then I trust the SRS algorithm to determine whether I truly know it or not. And I find that, over time, even with using hints, I acquire the words every bit as well as if I didn't. I might acquire individual words slightly sooner overall (in terms of days/weeks, etc.) if I didn't use hints, but I make up for this by being able to study many more cards (and many more unique cards) within the same amount of time. Language learning is a balancing act between quantity and quality, and I think this achieves a nice equilibrium. It's a marathon, not a sprint, and repeated exposure will usually get you there in the end. (I average probably around 1000 reviews and 100 new cards per day over about a 4 hour span, though this can vary greatly). Actually, hints might even be my favorite feature that Memrise provides, and it's rare to see any other SRS include it.
---
That said, keep in mind that this request is also likely harder to implement than it sounds, as it would require reworking and fine-tuning the SRS algorithm, at least slightly. Hints should come with a penalty, as they do on Memrise. Since if you use even a single hint it's an indicator that either you don't know it as well as you could (and thus need to see it more frequently), or that the course is badly made and you need the hint to break up ambiguity (as is the case for numerous ancient Memrise courses). And using multiple hints should come with stronger penalties for each hint. The exact scale of what these penalties should be is extremely debatable and, to the best of my knowledge, no one has ever documented exactly how Memrise handles it either. Too punishing and you might as well just miss the word completely. Too lenient and you don't see the word frequently enough to acquire it in a reasonable amount of time.
---
My second suggestion would help somewhat with the ambiguity in poorly-made courses and should be quick and easy to implement:
Memrise auto-submits correct answers on typing tests. That is: if what is typed in a typing test box perfectly matches the correct answer (including capitalization, punctuation, accents, comma/semicolon-separated alternatives, etc.), then the answer is automatically submitted and the card is counted as correct. This feature is necessary for some poorly-made (usually old) courses containing ambiguous cards (cards with the same prompt but different answer). This feature should be a lot easier to add than hints and is required for some courses; please strongly consider adding this as a toggleable option even if you disregard hints in the meantime.
Hope these are helpful. If these were added I would find my learning workflow on here to be nearly identical to that of on Memrise (relatively-optimized over numerous years) and should feel comfortable switching over completely (likely to be a forced switch, soon enough). Not sure how widely utilized hints, etc. are on Memrise, or would be on here, but hopefully you can still find time to add them sometime in the future. All of your work so far is greatly appreciated! You're really picking up the torch from where Memrise is extinguishing it, so to speak!
(Also please add a donation option sometime. I'd love to throw $5 or so your way every once in a while to support this project.)
