How to paint the tree gold
What is the minimum daily number of points required to turn all lessons gold? Would meeting the daily "regular" goal be sufficient, or would the progress decay before everything was covered? Any idea what speed should do it?
Update: To answer my own question, relying on @pinkodoug's post (thanks @Moomingirl for the link to the link): Decay rate is specific to users and words, since it changes each time you translate them (getting one right slows its rate of decay, and getting it wrong increases the rate). Therefore, there is unfortunately no way to calculate the tipping point, because we don't have access to this data.
On my French tree, I have followed a practice of either doing two review and one new every day or else doing three review lessons, depending on how much trouble I'm having doing the reviews. If any skills are not gold, I review those first. Otherwise I do general strengthens.
In twelve months, I have only had to strengthen specific skills on three or four occasions. Even after a few weeks-long vacations. It was the incredible stability of my French tree that led me to think that Moomingirl might be able to reguild a whole tree simply by doing two strengthens per day.
At this point, it seems very clear that three a day will do it. Her tree got worse and worse for six or seven weeks, but since then it has steadily gotten stronger. As she says, it'll probably be all gold by the three-month date, five days from now.
It also seems clear that over nine months, two a day does not do it. that tree is almost entirely deguilded. There are a few signs that it's leveling off, so it's possible that it would reguild itself after another three months, but it's just as likely that it would look the same in nine more months.
At this point, I think the optimal strategy for achieving and maintaining a gold tree would be the following:
- Never do more than three lessons in a single day. Either do two review and one new or else do three review, depending on how hard you're struggling. If the tree is 100% complete and 100% gold, only do two strengthens per day.
- If there are any unguilded skills, do skill-specific reviews on them. If there is more than one, do the ones higher up the tree first. You may need to strengthen the same skill several times before it's gold again--each attempt counts as a lesson, whether it reguilds anything or not.
- If there are no unguilded skills, do general strengthens instead of skill-specific ones.
- During a lesson, do not use the mouse-over hints unless you want to signal Duolingo that you need more review on a word.
If you follow that system, I'm confident your tree will eventually be all gold, so you shouldn't worry how colorful it is in the meantime.
Yes, this strategy fits in with Duolingo's recommendation. Although it is a bit different, " Duolingo uses spaced repetition to calculate exactly when you should review vocabulary for maximum memory benefit.This means learners should first focus on finishing all the lessons in a skill, and later come back to review them at just the right times".
I don't think a general strengthening exercise will normally test anything from a skill that's only partially complete. I suppose it might if you took a long vacation in the middle of doing one.
Well, in fact I think it does because it strengthens words not skills. It will pick whatever words are weak regardless of where they come from. At least that's been my experience. I could be wrong though.
That's probably correct. However, the words in a skill you're still completing aren't likely to be weak yet, unless you are doing the skill very slowly. :-)
There is a user currently running a test doing two "strengthen skills" a day to see how fast the skills decay. The user's name escapes me, but I believe their preliminary results indicated that at least 3 practice sessions a day would be needed to keep the tree gold.
I'll see if I can dig up more info.
Edit: Nevermind, my lovely assistant did it for me.
Someone talking about me? :P
I am almost three months into the experiment. There are a lot of variables, so I don't think anyone can give you an exact number of points required. To summarise my experience so far, as best as I can remember:
Both my trees were completely golden before I went on holiday. By the time I got back after a week, a whole bunch of skills had degilded on my main tree (Italian from English), and a handful on my reverse tree (English from Italian). I'm sorry, I can't remember the exact numbers now.
I had completed my main tree about three months before, over the course of about three to four months. I tested about halfway into my reverse tree and tested out of all of the rest of the skills, completing the whole thing in one day, approximately a week before my holiday.
I agreed with Greg to do two 'general strengthens' per day on the main tree, and three per day on the reverse tree (because I had done it so quickly and recently, so we expected a catastrophic failure).
Skills initially degilded faster than I could regild them, but have gradually slowed down. As of today, I have 52 colourful skills on the main tree, and 7 on the reverse tree. At the rate I am going, I expect the reverse tree to be completely gold again within a week (so it will have taken three months at 3 strengthens per day). I dread to think how long it would take at this rate to regild the main tree. This is bearing in mind that it was completely gold before I took a week's holiday, and I have spent the last three months doing two strengthens per day.
As Alexis said, Pinkodoug has done a lot of investigation into this issue too, here is a link to one of his postings with a lot of information on how degild rates work: http://www.duolingo.com/comment/5487578
My next plan is to do three skill specific strengthens per day, starting at the top of the tree and then working my way down. Once everything is gold, I shall do a maximum of two general strengthens per day. That should make for a strong, stable gold tree. At some point, when even the subjunctive sentences seem too easy, I shall step away from the tree to spend more time reading and listening to Italian. At present, my daily strengthens don't take that long, so I use them as an Italian warm up. :)
A huge shout out to Greg, without whom I probably would have quit my tree in frustration by now. My Italian has come a long way over the last three months, and it's mostly thanks to his experiment. :D
Edit: I believe it is modern to put a summary for those of a very short attention span (although how they are learning languages, I don't know). So...
TLDR: Depending on the starting strength of your tree, how many questions you get right, and how often you peek at words, once your tree is golden, two general strengthens per day might be enough, three should be.
Interesting, so earning roughly 30 xp a day using skill practice would be enough. That coupled with one new lesson a day, will make one require an average of 40 points a day.
This is where the coach feature could really shine. Rather than forcing someone to simply earn XP, they could change it to require certain successful practices per day, e.g. 30 xp (2 strengthen sessions) + one new lesson a day.
I think that would go a long way into creating a good structured routine to effectively help one to learn a language.
Thirty points a day appears to be enough for me. There seem to be a lot of variables. Earlier on, I was sometimes only scraping through practices with one or zero hearts. At that point more skills were degilding faster. Apparently if you peek at words, the degild rate also increases. And, if you got through your tree a long time ago, the decay rates slow down too.
I love the idea of the coach being tweaked so that instead of just sticking to a number that you randomly chose, he could tell you how many practices you need to keep your tree gold for that day.
In the meantime I would go with the theory of skill specific strengthens only until the tree is gilded, then two general strengthens before each new lesson.
" minimum daily number of points "
There is no magic number. It varies by skill strength. Some skills need to be review 6 or 7 times, some only once, others in-between. If you only strengthen them until they turn gold, they will de-golden faster. So, I suggest reviewing them while golden, once or twice. It will keep them that way.
No magic, there are exact numbers. I think that when you review golden lessons, you just accumulate extra points to decay before the lessons are no longer golden.
So it's just a matter of the (average?) decay speed expressed in points per day. Any daily goal over that quantity would do the trick in time. Just like losing or gaining weight.
And further, if we knew the total number of points in a tree, we could also calculate just how much time it would take us to paint it gold at a given speed.
It depends on the level of decay, which varies. It's not "standard" so to speak. I don't what you mean by the extra points. Extra points after goldening it (.80 is the minimum level for gold, as I have heard from the below user I linked to) reinforce it, bringing as close as possible, or, to 1.
I'm not sure. Pinkodoug is much better at explaining it than I am, but even he will say that the system is complicated. But, he might have a specific answer for that.
There is no "total number". You keep going. I am now 57000k+ over the level "limit" for Spanish, and it hasn't stopped. The skills will always decay, there is no permanent stage.
You could certainly map points onto the strength of each lesson. This has little to do with the number of points you accumulate on your profile. The total number of points that equals the strength of 80% (if what you say is true) of all lessons is what I call a "total number of points in a tree".
Now, if we knew the decay rate of these, we would know how many points are "lost" in a day, and then gaining more per day than this quantity would eventually paint the tree gold.
You can have a look at a description here with some references to official statements by Duolingo staff.
Anyway, I do agree with you that there is a sweet spot, a point in which a particular practice routine will generate a successful golden tree.
It is early still early to tell, but once the coach feature gets heavily used, they may be able to extrapolate the recommended practice routine, or points/successful practices required to maintain the tree golden. They could also conceivably recommend a personalized practice routine for each user. It is also likely that they already have data about people who consistently keep their trees golden for long periods of time.
That is simply a mathematical/programming problem at this point.
Again, not sure. Pinkodoug may have some answers for you, I can't remember if he talked about specific decay rate statistics or not.
I have also spent three months getting my trees stable. I finished the three (Es, Fr and De) of them within days of eachother, but in a big rush so they all degilded very rapidly afterwards. I started regilding from the top up to 100xp a day on each, until they were gold again, then just did whichever ones had regilded plus one general. They finally got to the point where none or only one or two were degilding. I maintained them at this stage by doing 35xp ish on each. This does seem to be the required xp to maintain a tree.
I am continuing to keep the German tree gold like this but am now practicing Es and Fr in combinations on the cross trees. I may return to the other from English trees and regild them again later.