Strengthening feels like weakening
I've completed two trees (Swedish and Welsh) and have been practising both extensively outside of the Duolingo website, so when I come on to Duolingo to try and sharpen up a bit, it feels sensible to use the 'strengthen skills' option.
The problem is that because I don't use Duolingo as my main source of learning (it had always accompanied classes, real-life interactions, audio courses etc.) when I come back to it, the skill tree levels have waned.
I'm then expected to answer basic questions in my target languages. This isn't completely lacking in usefulness, as even the very basic vocabulary can slip, if you don't say or hear certain words very often. But it puts me off using Duolingo.
It seems to me that the 'strengthen skills' function regresses to the earlier lessons which you've not done for a long time. If Duolingo was my only source of learning, that'd make sense. So wouldn't it be good if there was a way to practise without having to redo a lot of that stuff?
Perhaps if the 'strengthen skills' function pulled together questions from the whole range of the skills tree (where skills are completed), it would provide a better experience for tree veterans?
The general strengthen skills used to pull questions from the entire tree previously, but they changed it quite a while ago. I'm not sure of the reason for the change, but I doubt they'll change it back anytime soon.
After it changed, I stopped using it. Instead, I installed the skill strength user script and either use that to determine which skills to strengthen next or I focus on a category such as all verb skills. Going over very basic skills that I already know or have practiced outside of Duolingo on a regular basis is not necessarily a good use of my time. Some people strengthen backwards. I've done that in the past and it's far more interesting than the other way around if you have more than basic skills in a language.
You probably have a fairly good idea of which areas need more practice. You can choose to focus mostly on those. It's unfortunate that the progress quiz is no longer available as that was another way to identify areas that required more practice.
Duolingo's algorithms don't actually directly keep track of how strong a skill is. They keep track of the strength of individual words. Strengthen skills will put together a lesson made up of the words it thinks you most need to strengthen. Since you're doing a lot of work outside Duolingo, though, it won't know that you've revised words elsewhere.
There's some more detail about how Duolingo keeps track of our learning here: http://making.duolingo.com/how-we-learn-how-you-learn
Strengthen skills will put together a lesson made up of the words it thinks you most need to strengthen.
Yes, but 95%ish of the time it sure seems to think that the words you most need to strengthen all come from one skill, and that skill is whichever one is closest to the top of the tree and not currently gold.
Perhaps it's more cunning when the tree is all gold (although I have not observed this to be the case on the occasions I have had opportunity to check; it just seems to work from the top skill by skill), but obviously that wouldn't be applicable in the OP's situation.
Without an ignore button, like it is available on Memrise, we will never be able to focus on the really weak vocabulary, if the words and skills are picked from top to down of the tree.
Also there have been bugs in the old Python code what words are added to the list, "last practiced" update issues, selecting which bar strength words from the words list for a strengthen practice excercise (it picked quite often the wrong words, which were still stronger than other 1-2 bar strength words).
elizadeux is right, "Skill strength viewer" is one of the most useful scripts which may get it more right than wrong, but which update only works from source language English for the new portal as of now.
I always wonder about the high dead 0 strength word count.
Actually, I would love reading any offical announcement from the core backend developers, how old issues (posted so many times on the discussion forums) have been solved, what changes (improvements) they have made to the new Scala code during re-engineering.....
.....or if they just have been replicating 1:1 the old code (they also have been talking about refactoring) without fixing all those well-known issues before :(
@DuoLingo: Do you hear us?
Who is the DuoLingo core developer who has rewritten all the words list database add/update/select (based on strength, last practiced,...) functionality?
Please please write an additional blog posting about all re-engineering on making.duolingo.com and how the (core) code has been improved over the old Phyton code and those open, well-known issues.
If the tree is gold, it's OK to do general strengthening, but if you've been away for a long time, it will drive you crazy to start from the beginning. Instead choose lessons from further down in the tree, and you will find that you can strengthen more than one skill at the same time, and the easier lessons will probably become gold at the same time without your having to do them.
Given that such behavior is so linked to the word strength algorithms, which are perhaps one of Duolingo's most active areas of tweaking, A/B tests, etc., I think the answer is that, yes, it's pretty likely a user will see changes in the details of this behavior over time. A lot of people seem to have gotten switched to different test groups a couple months ago and then switched again a month later, just judging by a flood of comments complaining and then several saying that their complaints seem to have been heard.
Best plan will ever, I think, be to be cognizant of what concepts in the tree one has and hasn't mastered. The later need more focus whatever their skill color might be. Personally, I've probably spent the bulk of my Duolingo review time on skills that were already gold because I was trying to actually get something to stick, in at very least my short-term memory. [I personally was in a test group with artificially slow strength decay rates for much of my time here.]