So… Did anything come of this? Did they address any of people’s issues?
1 Like
i still don’t get it
which is the conclusion of that famed meeting CEO Memrise - User base? where is the transcript?
users complain, memrise does … what
3 Likes
Hi @PollyWalnuts, I’m wondering which Goethe B1 course are you learning from?
I know of two possibilities:
One that’s described as “official” (and contains a ton of errors).
And another by Schwarzerberg that’s very accurate. (I’m a contributor to this one, so I may be somewhat biased).
I would encourage use of the Schwarzerberg course, to avoid learning errors.
1 Like
Well, that’s good! I’m using this one:
1 Like
Excellent choice 
If you spot any errors, please let me know either by starting a thread under community German courses or by just sending me a personal message through this forum. I’ll fix any issues.
2 Likes
I don’t know if @BenWhately has posted a transcript yet, he said he would, but I can summarize broadly what I think we learned:
Memrise was on the verge of running out of money and shutting down entirely about 7 years ago. They had already determined that their community course platform (which I will call “classic memrise”) was not going to grow enough to support the company, so they decided to save the company with a new “curated” language-learning version of their app, focusing on Mermrise-created courses for the most popular languages.
Yes, this was a big change from what they had built until that point. They intended to keep both of these going in parallel, the classic memrise platform and the new mobile-only language-learning thing. As many of you here know, they did a pretty poor job of this, both in the implementation and in communicating with existing users about what they were doing. The kept breaking stuff on the old platform, at unexpected times, often without warning, while at the same time making changes they presented as improvements (such as changes to the user interface) and telling us they were continuing to support and improve the platform even though their own courses were higher priority. The reality seems to be that they had already decided classic memrise was never going to work, but wanted to keep it alive alongside their newer stuff, and maybe they didn’t even realize how badly they were messing it up and breaking it in the process.
They did, however, attract a much larger number of new users for their new popular language mobile courses, and they did save the company. Which means they saved the (degraded) classic memrise along with it.
[ Along the way, they also say they found that making user-created courses visible in course search on the mobile app caused users’ average time on the app to decrease. It seems that Memrise equates average time spent on the app per session, with language learning effectiveness, though they did not justify this claim and I am very skeptical of it. However, if you accept that assertion, then it follows that they fond that making user-created courses visible, decreased language learning effectiveness. Based on this, they hid user-created courses from the app, and they say this led to users spending more time on the app. Therefore, they do not ever plan to un-hide them. ]
More recently, they made another big decision on a change of direction: Having one codebase supporting both classic memrise and their new popular language learning app is not working. It has been slowing down their development, and preventing them from creating the features they think they need for the users of the new memrise, and they think the company could still fail unless they start moving a lot faster. Apparently they’ve been struggling with it for the entire time, trying to make it work, and they’ve decided it just can’t. So, they need to separate the two completely, and that’s what they’re working on now.
What this means, according to them, is that they’re creating a brand new platform that’s not based on the same data structures and features as the one we’re using new, and they will move their mobile apps to that new platform, and with it, most of their users. However, the old servers can remain, running classic memrise, indefinitely, so they don’t have to make any big decisions about its future now. They say that the main problem they’ve had supporting classic memrise is that since it shares the same code base as their popular-language mobile courses, it makes it much much harder for them to work on. When the two are separate, keeping classic memrise’s servers running won’t be expensive for them, and they intend to just keep them running.
Also, because they do thing Memrise the company is not on stable footing right now and desperately needs new features quickly to get more users and get enough money to support them, they’re going to do this separation step as fast as they can. They say this is a reason why they need to disrupt some existing features, which they aim to bring back later. This part of the talk was pretty vague, and Ben kept it vague despite a few questions, so don’t ask me for details :).
Some other things this seems to mean:
-
After the separation, classic memrise may be much more stable than it has been since ~2015. A lot of the changes they made, and the bugs they introduced, were because of the work they needed to do for their new stuff. When the two are separate, they won’t need to change classic memrise to support development on their new system.
-
This also opens up the possibility of reversing some of the changes they made to classic memrise, if they’re simple. Reversing those changes won’t affect their new system, so they’ll be more open to doing it.
-
But, they’re also not going to want to spend developer time on classic memrise. At least in the near future. Or ever? This isn’t clear. So, maybe they’ll be open to making very easy changes, after the separation is done. Or maybe, only after they decide Memrise is on solid ground and financially stable. Or maybe never, who knows.
They also talked about “moving” courses from classic memrise to the new platform, but this isn’t straightforward. First of all, the new platform is for languages only, so non-language courses can’t be copied over. Secondly, in the new platform there will be one shared database for each language, not separate databases for each “course”. This is going back to the original “wiki” concept they had for classic memrise, that they moved away from around 2012.
On the new system, users will be able to make “word lists” which are similar to “courses” in classic, and share them. They were somewhat vague about what form this will make, in part because it sounds like they’re still designing it, but it did lead to some misunderstandings. A “word list” seems like it will be a collection of sets of items (like “levels” now) from the shared database, along with notes, videos, and other bits in between, similar to what we can sort of do with “multimedia levels” in classic memrise. So, migrating an existing course would seem to involve first copying all of the items from its course database into the shared database, merging whichever items match ones that are already there, vetting them all for correctness, and then creating a “word list” that matches the levels in the course.
They said they would do this for “active” courses, and for courses that someone asks to have migrated. Which means they will not do this automatically, or try to do it for all courses, but any course you care about that is a language course should be able to migrate. However, of course, the new system may not match all of the features we’re used to, so some courses may have content that cannot be migrated. We didn’t get a chance to talk about how they might handle things like synonyms, alternate answers that we create extra columns for (to have memrise say “you entered the ___ and we want the ___ instead”), alternatives, commas, etc. They do intend to include images and audio and video, but it may be handled differently and may not work for the courses we have now.
In response to some of my questions, Ben also indicated that:
- If another company wants to replicate what classic memrise does, they’d welcome it.
- If another company wants to just buy the classic memrise platform from them, they’d welcome that and try to help them. Apparently they had been in talks with someone (quizlet?) about that a few years ago but it didn’t work out.
- They think my request to develop a supported way to export full course content is a good one, that they probably should do it, but who knows when they will have time for that. So, maybe.
17 Likes
I would only say there have been a couple of instances where maybe a tiny bit of specificity would help to distinguish one word from another. I’ll keep my eye out for them. Thank you!
1 Like
Thank you @cos, for a detailed and informative summary.
It would be nice if @BenWhately liked or endorsed your summary.
I wonder if curators will be able to modify their migrated courses?
Without giving it much thought, I think I’d prefer all my curated courses to remain on the Classic (not official relaunched App) and I think this would be better for MemRise.
I would like the App to draw attention to our courses (as has been suggested after completing Level 7).
And I would love it if we returned to the planting theme and landscape.
3 Likes
Thanks Cos, and apologies for slowness in posting the transcript. I’ve been down with COVID. I’ll get the transcript shared asap.
One note on the above: right now Memrise is in a solid financial position - back in 2014 things were tight, but now we are fine. What I intended to say was that if we didn’t make a change now, then in a couple of years time we would be back in a difficult place, and that is not something I am prepared to risk.
4 Likes
If there are going to be two distinct sites then anyone who is on courses not language based would have then to work from the two sites if the language ones were migrated. That would be quite tedious especially if the two sites worked differently. Even if UGCs remained on the old version you might be using memrise’s new courses so it would still mean two sites.
I would worry that if the old site was sold the courses could be changed dramatically. I tried quizlet for example but didn’t get on with it very well so would be quite disappointed if that happened.
I wonder what the proportion of paid members is now. If memrise is in good nick is it because of them or the advertising. Is there going to be a difference between the old and new site in whether you have to pay or not.
2 Likes
Oh boy, that’s pretty much the worst transcript that I’ve seen in a long time - sadly, the sound quality isn’t very good, plus the text transcript is next to unreadable, persons were mixed up, etc.
So you’d really have to listen through the whole thing (1,5h).
Would anybody who participated care to summarize the call …?
1 Like
Isn’t that exactly what I did, just before they posted the transcript?
2 Likes
Here’s my brief summary of the call. Feel free to correct.
-
Memrise struggled financially before the official courses era has begun (before 2014-2015). Now there’s no such problem, but they have to grow and develop further to be stable. Introduction of the official courses was a difficult time for Memrise staff because users were against changes.
-
The classic Memrise website has been impeding the development for years. Now they are taking a risk and building a new platform. It will have all the fancy features described in the other thread. They have to remove mems from the old website to go forward. Nothing is decided about mems’ comeback.
-
After everything is ready on the Memworld website they’ll start to transfer community courses there. It’s planned to do it automatically, but not every course will migrate, like non-language courses. What to do with them Memrise team will decide later. Community courses are to stay on Memrise for a while, there’s no plans to remove them (*not clear if we can chose where our courses will stay). It’s not difficult to keep them in their current state.
-
Dictionary databases on Memworld should help users to build high-quality courses out of movies, podcasts, etc. (*not clear if it would be possible to add words which are not in the databases)
-
Memrise team decided to hide community courses because users were not staying long in the app after interaction with them. They are not planning to hide community courses on Memworld.
-
Instead of levels on Memworld will be scenarios. If you are a teacher you can write a scenario, a sequence of courses to learn.
One of the most interesting parts of the call
We’re in a much better place. And I mean, let’s creatively explore what we can do that like Steve says, We are more than happy to open source that code base like to work out how we can spin that off to someone let them work. We did talk with Quizlet about doing this. It turned out to be a bit complex to do but at the time and didn’t quite suit either of our businesses. But we were totally open to that. And yeah, no reason not do that.
5 Likes
The more I read about this, the more I ask myself why you, Memrise would risk breaking everything (including functionality in your official courses, yes that’s what has already happened!) and perform open heart surgery on your productive system instead of developing the new solution in parallel and switching things over when you’re ready.
And, sorry if I have to be a bit brusque again, but if finances aren’t the problem, how can it be that there are just so many countless errors in your official courses that haven’t been fixed in years, even though you have constantly been notified by your users about them.
Why is it that every update bears a very high risk that something no longer works, or a feature gets removed without prior notice (probably without you guys even realizing it!).
Why does it - best case - take extremely long until you fix a problem that you have actually confirmed? Actually, most of all problems never seem to get fixed …
I’d sure like a reply to that. Maybe you could shed some lights on this so we at least understand, @BenWhately?
4 Likes
In what ways?
What are the major problems? I’ve noticed @cos’ “repair the community courses” series. Are they mentioned there? If not, could you tell me what they are?
I’m sorry. Since I’m new here, I’d always like to learn more about the issues.
There are just too many to list them all here.
In Feb 2019 (almost three years ago!) I started listing the bugs/problems that I found in the Android app. See for yourself how many still exist:
And that’s the Android app alone. As for my list, I got tired of working on it (and it was quite a bit of work!) with so little happening, so I pretty much stopped reporting new bugs.
Yes, there has been a whole lot I didn’t even care to mention anymore, some are pretty severe.
Here’s just one (again: Android): the app intermittently (I get this like once or twice a day across my session that covers around 20 or so courses, very reliably) starts a session in a completely different course than the chose one, also very often (but not always) mixing up my carefully constructed order of courses. Or the number of times you see an error message telling you that a course could not be loaded. Or the fact that, when you start a review session with 1 or 2 words, the app will load eons and then start a session with more than those two words. Bunches of others, sigh. Don’t want to think too much about them or I’ll end up getting too upset once again …
3 Likes
just lol if that ends up happening. Nobody is going to make courses on Memworld.
- User created courses
- mems
- typing
Those are the three things you don’t touch, because that is where your most loyalty is and most hardcore supporters are. But you always end up messing with these three things. If you get rid of user created and or restrict my customization to them, I will end my subscription.
5 Likes