Decay Rate Too High
Been clearing stale lessons in Irish regularly only to check the next day and see thirty new stale lessons. I don't care if the decay rate is based in empirical study of human memory; asking me to do over an hour of lessons a day is a good way to get me to do zero lessons a day. This wouldn't be a problem if Practice lessons were ordered by decay or even randomly, but instead they tediously start at the top of the tree so if lessons aren't cleared they get stuck in the wueue behind ones that get stale the next day. Very frustrating, to the point where I will likely stop doing the Irish lessons at all since they're crowding out all the time I'd be practicing ither languages.
They calm down if you leave your tree for a while and then go back and get all or most of the questions right. It takes the gap length into account as well as accuracy and whether you peek. I find there is little benefit to keeping a tree gilded once it is complete beyond week of two to reinforce the final skills. Better to leave it to settle and return for a full regild later on. Lots will stick but you can then focus on what didn't. But do keep it gilded until you finish it at least. Repeating every lesson the first time it degilds can help keep this more manageable as otherwise you will have it regild every day for however many skills it has. And don't knock repeating the basics. You need to have the patterns become entirely automatic to ever achieve fluency.
I completed the Irish tree and I had to go through two thirds of it until I even got to see how a non-gold skill looks like!
After that, I got one or two stales every couple of days. Compared to the amount I should review this was neglectable. After I completed the tree I did only my minimum of two reviews per day for a couple of weeks and by now there are now about 15 skills stale. Which means, I could make it golden again in about two hours.
Two hours after weeks of almost absense. That is IMO two little. Skills should decay much faster.
Maybe it is my learning style. When I was working on the tree, I usually started one new lesson every day, I reviewed one or two recent lesson and then made two or more "strengthen skills", resulting about an average of about 50 to 60 per day. I should also mention that I reviewed the vocabulary outside of Duolingo, so I guess my Duolingo success rate was much highter than expected if I used Duolingo exclusively.
As a matter of fact: At least for some time my low rate was at least partly because of an A/B test because suddenly eleven skills were not golden anymore for no apparent reason. Other people in the forum experienced the same and complained.
I just did a bit over an hour of reviewing and everything was golden again. After that I didn't notice big changes. Maybe they put the old values back into place? I don't know.
The end of that test affected me to. I had about a dozen skills degild in the blink of an eye, but then a few reviews later they all went back to gold pretty much simultaneously. Since then, though, things have definitely degilded faster than they did before.
In the Chrome DevTools window I can see something labeled sg_test_challenge_probability_experiment2 for which I have a value of .1 I wouldn't be surprised if those for whom completely unsustainable degilding rates are in a test group with a higher number.
AS El2theK has said, there is no requirement that you keep the tree gold.
Also, it sounds like you're trying to strengthen by using the general strengthen skills. I recommend, instead, starting at the end of the tree and working backwards through the degilded skills. That way, you review the more recently completed content and some earlier content simultaneously.