The problem with absent auto-accept, is that it leaves a user without a way of dealing with inherent question ambiguities.
There are languages with optional spaces, optional pronouns and optional contractions, word order might be flexible in the case of phrases, and some grammatical forms might not have a clear 1 to 1 correspondence between question and answer languages.
Without auto-accept, one has to additionally memorize all these specific details for every item being learned. This has nothing to do with learning a language itself and thus makes the whole process much less effective. At that point, one also has to review each course separately, as different courses might teach different meanings of the same words, and there is no way to distinguish between them during the review session.
There are systems that solve this issue even better than auto-accept – like fuzzy matching system, as in Duolingo, or user’s ability to correct automatic grades, as in Anki – which do not require brute-forcing your way through all possible alternatives. But Memrise has neither.
Why? Just hit enter, and it will accept your answer just as it would’ve if it had auto-accepted at that same point. The difference is you have to decide to hit enter, that’s all. The very same answer would still be accepted. This should make no difference whatsoever in whether that answer is accepted or not, as long as you hit enter at the same point as the auto-accept would’ve triggered. I don’t understand what you’re trying to convey here at all.
For years now, Memrise has very solidly built up a reputation of being truly awful at communicating with users. Silence, mixed messages, contradictions, confident claims that get walked back, more silence, and so on and on and on. You’ve made a very large number of people bitter and angry, people who actively recommend against Memrise on social media and to their friends. When you had the community forum recently, one of the things I thought I got from that was that Memrise had learned that you need to to better about this, and make honest and clear communication more of a priority instead of jerking users around perpetually. Are you saying that no, Memrise has not learned this, and it is still a very low priority for the company?
@BenWhately asked what the question about “respect” meant, and I gave him what my answer to that would be. Here is another example. You have here a post that people read as meaning that you plan to delete all mems at some point. People keep commenting about that and you keep replying that it’s not true, actually. The post itself says you will keep updating it, but you haven’t updated it, so people keep reading it and asking and then pointing out the contradiction. You continue to not update the post to clarify. And eventually, after people have pointed out a few times that you should update it, you respond that it’s just not a priority, you’re too busy.
This is fundamentally a show of deep disrespect for the user community.
The difference is that often enough it is not clear, at which exact point the auto-accept would have been triggered.
What I’m saying is that often a question has several plausible answers. With auto-accept, you don’t have to guess which one is expected in any particular case, because you can try several to see which one is accepted by the system.
Of course, auto-accept is not a real solution to the problem of ambiguity, it’s rather just a crutch, but now we don’t even have that.
But that’s exactly the reason why auto-accept is bad. If you know a valid answer, that answer remains valid without auto-accept, you just have to know that it is a valid answer. You do not need to know all the valid answers, any more than you did with auto-accept. You just have to know any one of the valid answers. With auto-accept, though, you didn’t actually need to know any valid answer - you could just stumble on it without realizing it’s valid. That’s harmful, IMO.
What I’m saying is that often a question has several plausible answers. With auto-accept, you don’t have to guess which one is expected in any particular case, because you can try several to see which one is accepted by the system.
That’s an example of what I’m talking about. Auto-accept encourages a form of trial-and-error where you never actually know an answer, and fool yourself into thinking you’ve learned it when you haven’t. You can get marked right, the spaced repeat gets longer, and you do the same thing again next time, never even remembering that you neither know it know nor knew it the last time. It promotes you failing to learn the item at all, ever, and not even realizing it because you keep getting it “right” by experimentation and guessing each time. Yuck.
This totally undermines the point of spaced repeat memory. It fools the learner, who thinks they’ve completed the items and doesn’t realize they haven’t actually learned them.
No, that’s not the case. The system expects you to know the exact single answer that is considered correct. “any one” will not do.
Like here, where the only “error” was just incorrectly guessing, where to put spaces (keep in mind that they are completely optional in Japanese):
or here, where a word, which was expected in kana, was written in kanji:
or here, where the different (but just as valid) word was used for “but”:
(and there are many more examples like that)
I know, that having auto-accept comes with its own drawbacks as well. I never said it was a perfect solution to the problem above, but at least it was the one we had. Giving tools to the people that are using the platform for learning, IMO, is more important, than trying to stop the ones that want to play the system against their own interests.
You just removed one of the important features for many users, and it’s not a priority to inform them you are going to bring it back after some time? Welp, conspiracy or not, it does not sound as if we are going to see mems on Memrise again.
To be clear, I am not guaranteeing we will be able to bring it back. It is something we’d like to do and I personally am keen to do, but I am not going to update an official update with a possibility because that would bot be fair to people. Right now it is competing with all the possible things we could build. As such I am deliberately not making a commitment, especially not by any date or with precise details on what the feature will look like.
What about updating the part regarding the complete removal of mems from the current Memrise platform?
You’ve already stated multiple times in this thread, that such plans no longer exist. If this is indeed the case, why not place it into the official message?
I don’t know how to read Japanese, but what you’re describing sounds like a bug - something optional is considered required. Now I see what you mean about using trial and error with auto-accept to try to find the correct answer, but that doesn’t sound like it actually addresses the bug. It’s just marking things wrong that are actually right.
Memrise didn’t have auto-accept for many years, to begin with. What did people do then? Was this bug where optional spaces are required, introduced after auto-accept?
I beg you, do not remove mems from memrise!
They give emotional support to learners and therefore not only rise learning effectiveness, but also make people happier during and after learning session.
You’ve done a good deed by creating memrise for which I am very greatful, but you can contribute a bit more to human happiness.
Totally agree. I just want the old and good difficult words. I want to repeat the words as many times as I feel its necessary to remember them. They ruined the whole point of difficult words.
A bug is when a function works improperly, but Memrise doesn’t have any functions that handle such things in the first place.
At the very least it was providing functionality that helped deal with the problem. I doubt we will ever get anything better than that, and at the moment we have nothing at all. A good fuzzy matching system is hard to build (even Duolingo one does not work all of the time), and manual marking corrections are unlikely to be even considered by Memrise (although they were proposed multiple times by users even in this thread) because they only care about beginners.
I have no idea. I briefly used the site around 2014-2015 just to study English, so naturally, I wouldn’t notice any issues with spaces, even if they were there. My current account was created three years ago, when I started learning Japanese, and, as far as I remember, the auto-correct was already in place at that point.
Yeah, I mentioned this elsewhere, but synonyms are reason enough for auto-accept. There are so many German verbs I am studying with the same two-word definition on Memrise and no identifier as to which one it wants. I have no way (in general, but especially in Classic Review mode) to know which one they are asking me for (or even for which course), so I just have to run through them until one hits. I try to tell myself hey, at least you’re drilling synonyms. Shrug.
No, synonyms are an example of the corrosive effect of auto-accept. A course like you describe is just a bad course, no different from a course that has wrong answers or misspelled words or false definitions in it. Covering up bad courses like that with auto-accept allowed them to proliferate, and got more people to take those courses rather that either get them fixed or take different courses.
A course should always accept any synonym that’s valid. When a course fails to accept a valid synonym, that needs to be reported and fixed. But because of auto-accept, people just accept it as is and don’t get it fixed and it drives courses to be worse over time. This is one of the main reasons auto-accept is so bad for memrise.
I see the point you’re making, but I also find it a little unfair and excessive to call a user-generated course (the only ones I can use because I already went through all the official German that Memrise has to offer ages ago) a bad course because the course creator failed to add every single synonym in the target language for every word. I mean, this is just me, but I try to avoid looking a gift horse in the mouth.
But yeah, sure, it would be nice if Memrise had a way around that. Reality is they don’t. I personally have other things that infuriate me way more about this website, but everybody has their thing I guess. Obviously the solution would be making it an option to turn auto-accept on or off, not making a blanket decision for every type of learner as this site is already wont to do. If they did not fix the synonym thing and got rid of the auto-correct, this site would cease to be usable for me. Stats would be thrown way off, I would constantly be having to review words I already know, etc.
I think you got part of my point but not all of it. For years, Memrise didn’t have auto-accept. These missing synonym issues are similar to other bugs in courses. When people ran into them, they’d report them, and get the course owner to fix them. For abandoned courses, they’d ask to be added as maintainers and fix it themselves. People still do that for various other errors in courses, but now thanks to auto-accept people just let missing synonyms slide and don’t try to get them fixed. It makes things worse for everyone. It’s one of the reasons auto-accept is damaging to memrise. And in the meantime, courses still can have other errors, so auto-accept doesn’t actually solve the problem - it just lets you gloss over this one particular kind of error (and avoid getting it fixed).
No, I see what you mean, definitely. What I mean is that on, for example, a 5,000 word frequency vocabulary course, I would be sending an email for pretty much every word (hours and hours and hours of time on corrections) because man, there are a lot of ways to say a thing. Even just to say “thing,” I mean THING Synonyms: 112 Synonyms & Antonyms for THING | Thesaurus.com, you know? It sounds like an awful lot of work.
