The spaced repetition accounting has a serious flaw
I like keeping my skill tree golden by making sure all of my words have not deteriorated. I gather that this is based on a spaced repetition idea, so that words you have just learnt deteriorate faster, prompting you to review them soon after you have first learnt them. After this first time revision, they deteriorate more slowly. After a second revision, they deteriorate even more slowly. It's a great way to get vocabulary into the long-term memory.
However, today when going over my 'deteriorated skills', I noticed that the words it says I have not practised in 3 months or more are words I translate everyday. For instance, in the present verbs section, it says I have not practised the verb faire (to do, make) in months. This is simply not the case. The verb faire is very common, even unavoidable. I know I practised it yesterday, because I took note of a construction that used it which I found tricky. This is only one of many, many examples.
So why is it that when we are using words in some skills, they don't contribute to the counts of the same words in other skills? Getting this right is the key to an accurate accounting of spaced repetition, and hence the key to ensuring that Duolingo is effective.
Our spaced repetition system tracks a mix of general words and their surface forms.
For example, both the idea of the verb "faire" and specific conjugations "(je) fais" or "(vous) faites" are accounted for, separately. Now, whenever you see a specific form the general form should also be strengthened, but there are still a few bugs that cause this not to happen, particularly for users who signed up many months ago like yourself.
We are in fact working to fix these bugs this week, so that whenever you see/practice a word, the general form is also updated and this is no longer a problem. :) Thanks for your patience.
Thank you! Good to know that it has been spotted. Duolingo is a great resource.
Thanks for the clarifications! Usually I just treat Duolingo as a black box and trust that the big green owl knows what she's doing, but it's nice to have some idea of the algorithms behind it -- especially when they are being constantly improved.