"How we learn how you learn"
Almost a full year ago, Burr Settles wrote an article titled "How we learn how you learn", which talks about the science behind the skill strength meter on Duolingo.
From the article--"At Duolingo, our goal is to make language learning fun and effective. We think the best education should be full of play, so we're constantly developing new features that make learning new things — and practicing old things — feel like a game! At the same time, we're serious about taking a scientific, data-driven approach to all of our products, and about sharing what we learn with the world. In this post, we'll take a look at the science behind the Duolingo skill strength meter.
The article also talks about the strength meters, the student model, the experimental results, and more.
You can find the original article here: http://making.duolingo.com/how-we-learn-how-you-learn
Good luck! :)
The article probably does not talk about the endless bugs which seem to be in the system and "Words learned" user database, e.g that the "last practiced" column is quite often not updated or that single words from lessons are N O T added to the database.
Let me ask you:
How shall the "word decay" and "review select algorithm" code work reliable, if the underlying user database code seems to be not reliable?
An algorithm can not be better than it's database??
Why are so many words in my list due to practice 1 year, even I practice daily?
Why is the "select algorithm" half-written code, that it even even not select your really weak or difficult words?
Why is there no way to manually mark words as difficult and practice them?
Let's face it:
What we get in strengthen exercises seems to be just R A N D O M stuff.
Even the global "Strengten skills" is that f..... up (it can't mix content across skills, weak words, difficult words) that we NEED to rely on workarounds like https://www.duolingo.com/comment/6176795/Best-Way-to-Make-Your-Tree-Turn-Gold-and-Stay-Gold or 3rd party user scripts like "DuoLingo skill strength viewer" which just shows me for hundreds of DEAD (strength 0) words!!
The above in theory, the rest is real world....
You're raising some interesting concerns about Duo's algorithms which can be improved. I can personally attest to the inaccuracy of strength metrics on the Words section of Duolingo. There are so many words marked "overdue" while I practice them daily and I know that I know them!
I'm surprised they haven't given you a response.