1. Forum
  2. >
  3. Topic: Duolingo
  4. >
  5. The rate of learning tree dec…


The rate of learning tree decay is too severe

@DuoLingo Team - it would be really appreciated if you could comment on the following:

I love DuoLingo, and use it daily, and have now completed my French tree.

I like the idea of the tree decaying (so that I have to revisit and strengthen each subject) but I am finding that the rate of tree decay is much faster than my typcial daily XP target. I have set my daily target to be 20 XP so this is the minimum I do - I sometimes do more when time allows. But the tree decay can be 5+ subjects per day, so this means I am not even "treading water" - I am going backwards. This is VERY demoralising and frustrating. I took the decision to ignore the decay and just pick subjects at random, but this does not help me learn because I lose track of which subjects I have practiced recently and which need re-visiting.

There is a very simple solution to this - DuoLingo should set the rate of decay in accordance with your daily XP target.

Is anyone else finding the rate of decay to be too severe?


November 26, 2017



Duolingo very rarely takes up such suggestions, no matter how good they may be.

While you're waiting, try this: do a non-gold skill over and over again (or maybe only once, who knows?) until you have answered all the questions (in every lesson) w/o making a single mistake and w/o peeking at the mouse help. Then see how long the skill stays golden. Or if that's too onerous a task (and it isn't easy, that's for sure), then just try this with the non-timed practice, and see if that helps.


I believe that the most efficient way is not to do it over and over again but do it once a day and come back the next day. That's what I understand spaced repetition to be about.


The question was about keeping a Duolingo tree golden, not about supposed efficiency in learning a language.

Be that as it may, this method is quite efficient once you get moving with it, and review can be done as often as desired or needed, of course. Use the "strengthen" exercises available after a skill has been completed, for instance.

If you're simply repeating what you've read elsewhere buit haven't tried this, why not give it a try to see how it works?


i agree! the one thing that always makes me leave dl for a 'short' hiatus - usually a couple months at a time is because i get too overwhelmed by all the decayed skills. i have been doing probably 75 xp a day on average over the past couple weeks, and i have made pretty much no progress in making my tree gold. each time i log back in to DL, a massive amount of skills are decayed.


You might want to ignore it (or how many times do you want to re (and over-)-repeat skills after they lose their strength, again and again?) and use the user script "DuoLingo skill strength viewer" on the web portal to show concrete word strength and skill total percentages.

Just try to randomly strengthen skills, how you like it....

However, there is a bug in their "Words" /vocabulary/overstream stream API, which does NOT display 2-3 grammar skills in my strength viewer list: https://www.duolingo.com/comment/24416411/URL-vocabulary-overview-backend-stream-is-missing-4-verb-skills-words-list-incomplete


Thanks Thomas - how do I get to the "DuoLingo Skill Strength Viewer" ? I cannot see it?


Thanks Thomas I will check it out....


Also please note Bilko.FR, DuoLingo's strength and word decay system is developed under the assumtion, that you only use their software / portal.

For vocabulary and daily reviews of the backlog with a good spaced repetition (SR) algorithm - but including support to IGNORE some words, review out of cycle, support for pausing long-term review index card boxes for completed courses, etc. -
you should heavily make use of 3rd party external flashcard software like Memrise, AnkiSRS, FlashCardsDeluxe, SuperMemo, www.languagecourse.net vocabulary trainer VT (Leitner index card boxes), self-made (paper) flashcards including 2-3+ or more example sentences, etc.

And if you do:
DuoLingo does NOT know anything that you review specific words externally in a SR.
IMHO it would be better they would write a plugin for Memrise Co. to sync external words and their review state.

Currently I am trying to concentrate my DuoLingo reviews on their "grammar skills" (and ignoring a bit the rest, else I would probably have to make 100-150XP to 250XP / day).

Sadly to say, there is zero customization ability per user....but I have the same (half-part) problems on Memrise.

NO, I did N O T find the full-blown solution (German: "Eierlegendewollmilchsau") to this :(


OK, "eierlegende Wollmilchsau" is a fantastic word! Thanks for that.


Thanks all to those who have replied so far...

I really would like more feedback on this topic.

My gut feeling is that those who do not find the decay severe are able to devote more time to DuoLingo per day, than those who are struggling.

DuoLingo is a great learning tool - but it should not become a test of endurance. By allowing users to set their own rate of a decay to a level they feel comfortable with will encourage them to keep using the tool, rather than getting demoralised and giving up on it.


The thing is, a normal user with zero knowledge about "forgetting curves" and mathematical algorithm tuning should not be able to adjust the algorithm all by him-/herself.

But I agree that it would make sense to support more options to get away from "the standard" curves and having only a rather fixed system.
I can't remember the thread where I wrote it into, but even the "Memrise standard intervals" (see wiki) do NOT fit me well all the time.

e.g one of the Memrise recent code updates killed a very good user script from Cooljingle, where you could "kick back" certain words, which you reviewed after pausing "out of the standard Memrise intervals" but still remembered / recalled them well.

For other words and doing "over-watering reviews" my experience is, that I often do not know some (not all) words even I review them earlier than the standard Memrise interval; AFAIK the algorithm for those (e.g 66 days) over-watered words will N O T be reset to 4-12h (it stays to the value set before for the normal blue watering / review period).


I really don't see what the problem is. You don't have to keep everything gold and it's really useful to know which areas need practice.


Hi Paul - many thanks for the reply.

Please don't mis-understand me... I probably should have said in my opening comment that I am not bothered at all whether my tree is gold or not, or how many points I have, XP's, Lingots, Streaks, Levels, Fluency or whatever next. I am just attempting to learn for myself the basics of a new language. It would help me greatly if the rate of decay was at a level I can keep up with on a daily basis, so that DuoLingo can guide me methodically and chronologically over the subjects I need to focus on.



You need 40 XP per language per day to make progress.


The decay is really helping me learn, when you perfect a lesson it seem to not decay. So if things are decaying fast do it without tips or hints and no mistakes.

  • 2122

Hi Bilko - I agree that you can go through frustrating phases of treading water (I do 20xp a day as well) but eventually things calm down and you make progress - I am now at about break even - all staying gold and doing a new lesson about 2 times a week - I am about 16 topics from the end of the tree - but the continual repetition of lessons done 2 weeks or 2 months ago is really helpful to concrete in the ideas. So for me the rate (after a year or 2!!) is just right for keeping topped up (&compliments actual lessons I do in a school in the evenings). Good luck - keep going!


Me personally, I don't find it too severe. Because we learn more than 1,500+ words at the end, I would expect the decay to be pretty severe. If we review just a tiny fraction of the 1,500+ words a day, we wouldn't be covering much of the words and would eventually start forgetting many of them as time goes on.

I would just say to not pay attention to the lessons that have decayed and just work on the ones you need a review on.


It's works well as is. No need to change anything. Handy to refresh knowledge, if you're not a user who logs on everyday.


I think you need to uncouple your daily target from your concern about skill decay. I set my target at 10 XP, but that doesn't mean I stop with 10. I do what I have time for. I am finding that the longer I study and the more times I go through the tree the better I do on getting the answers right without peeking. It's becoming easier and easier with time.

So give it time. There will come a day when you are looking for a skill that needs refreshing! That's a promise!


Thanks Julia for taking time to reply.... I will persevere with this but I am interested in your situation.....

Are you saying that you are now in a position where you can keep the decay "manageable" with 10XP and a little extra occasionally ? I have not been able to achieve this without spending 50+ XP per day minimum.


No, not just 10 XP, but that's all I expect of myself. I have the whole tree golden except for subjunctive and past. Today, the conditional showed up as needing practice--which it does. I'm still making mistakes when I strengthen that skill. But on the earlier skills I can breeze through one strengthening exercise, make no mistakes, and it stays golden for quite a while.

Here's my situation: I've been doing Duo for about 5 years; before that I worked with PImsleur (which I intend to get back to), and I have audited the second and third Italian courses at a local college (in fact, we just covered the conditional and I need to work on it for the exam). I've finished the tree, but the skills at the bottom are difficult for me, so I've worked on keeping earlier skills gold and working harder on what we're covering in class.

So, after all this time, doing some little bit of Italian almost every day, I'm beginning to get it. I can make up my own sentences, understand (slow) spoken Italian, etc.

Which is to say that the longer you stick with it, even if you can't spend lots of time every day, the easier it will become.

Have a lingot!


@Bilko: Finally the solution idea hit me in my face:

You need to join an Android club!!

If you do, you will compete with 14 other club members....and trust me...you will not stop at 20XP/day, if the rest does not fall asleep all the time.
. .
I have two guys this week in my Portuguese club, who do ~2300XP and >1500XP; one of them joined the last days....they race each other....the rest is about ~350XP and much lower...

So yeah, being in a club - and not only looking at the web dashboards for following users and their weekly XP progress - might help you to push the reviewing to a completely new level :-) :-)


You might want to forget about keeping all skills golden for a while, and focus on keeping them all at a particular level - 2/5 for example, or 3/5. With my French tree, I kept everything at 4/5 while I was still progressing in the tree. Right now, now I have finished the tree, I focus on keeping everything golden from the bottom up, bút also still take into consideration the skills that are dropping below 4/5. It can be a bit of a hassle, but you have to think in small steps; keeping everything golden all at once won't be doable within a few days, it actually takes a lot of time. And rightly so, as DL's algorithm tells you where you weak points lie which you have to improve first. Adjusting the rate of decay won't help in my view, as you will get a false sense of proficiency in your French - or any language.

Apart from this, try to avoid as much as possible looking at hints, as they influence the algorithm. Also try to avoid making too much mistakes; this sounds obvious, but what I mean is that you check your answers before you continue. In this way, the algorithm will recognise that you make less mistakes and that you actually understand the material; if you make a lot of mistakes, the algorithm won't know whether it's because you don't know the answer or you're were simply rushing a bit. And also in this case, the algorithm is working as it should: it actually helps to sometimes take it slowly and think about the answer you have given, instead of just rushing through it to "get it over with" as quickly as possible.


Thanks for your reply..... and I like your suggestion about keeping everything at a particular level eg, 4/5 I am going to try this...

As mentioned previously, I am not bothered about whether my tree is golden or not - that is not my motive here. I am just wanting DuoLingo to "guide me" methodically on a daily basis (based on the limited amount of time I can devote). Everything "decaying" very quickly is not helping me keep track of which subjects to focus my limited amount of time on.


Fantastique! And I think you have just the right mindset: not caring about the gold itself, but thinking about ways how you might benefit didactically from the system. I'm sure that you will get much more satisfaction out of your studies by going for 4/5 first and I wish you all the luck with that!


Hi Bilko.FR! I really don't know, but I hope that Duolingo takes up your suggestion. Even though many comments like this are rarely accepted, maybe this one will be. I hope it is! :)



This is my favourite method
Best Way to Make Your Tree Turn Gold and Stay Gold

I only have to do 2 - 8 strengthenings a day to keep four trees gold (English from Dutch, German from English and their reverse trees)


Thank You Pentaan - I will take a look...


Yeah, Duolingo is always working to figure out a decay rate that best achieves their overall objectives on average, but unless you happen to be precisely at that average, then what they set isn't going to be optimal for you. And, even more significantly perhaps, Duolingo's objectives aren't the same as yours, even if they can coincide. Duolingo in part wants you to do more lessons. That means they can show you more ads; it also means you'll learn more and faster.

Anyway, so yes, you want the decay rate to be slower. Other users want it to be higher. Any which way, a really useful thing to do I think is very much along the lines of what slogger suggested: work on a skill until you're pretty good with it. I wouldn't say perfect; I would say maybe 85%, like 17+ right in a timed practice with no hints. When you achieve that mark in a skill, move on, and don't worry about what color it comes to have later (although you'll probably find it stays gold for quite a while).


Thank you Piguy3 - Great response. I will work on the suggested strategy.

I think what I am hearing from all of the feedback is that: (a) some people want the decay to be slower (b) some people want the decay to be faster
(c) some people want the decay to remain as it is now

So DuoLingo - let's make the decay rate user-definable therefore keeping everyone happy (AND keeping everyone returning to your site)


I totally agree ;)

Learn a language in just 5 minutes a day. For free.