The maximal number of lessons per skill looks to be decreased, which is probably good (better separation). And 35 new skills (current course have 75) — quite impressive.
Thank you very much for your efforts, this is just amazing.
Sorry for a question (maybe I am missing some faq), but how is the experience within current tree treated? Is it reset or somehow merged?
Wow that is awesome! Can't wait to see the update!! This has given me a sudden rush and desire to quickly finish the current version of the tree. I'm only about to pass the third checkpoint though (41 skills to go) but hopefully I will have completed the majority of the tree when the new version is released. Not that it really matters anyway. It'll give me something to do on this 6 week break I am on though. Excited to be able to learn more of this awesome language!
We still have plenty of work to do, so I'd say at least a month until we consider it ready to launch.
Then comes the A/B test, which can go on for quite a while (the duration is not up to us). The new tree will only be released to all users IF it passes the test. If it doesn't pass, then it's back to the drawing board for us... and I might just cry a little.
The A/B test is a very different beast to the kind of alpha test the Hebrew contributors are looking to do.
Hebrew alpha test: They'll invite one or two suitable alpha testers and give them contributor privileges (this is the only way to allow them access to the incubator test bed).
Tree 2.0 A/B test: Users are randomly assigned to group A (tree 1.0) or group B (tree 2.0), and then the stats produced by the two groups are compared. I seem to remember that only new users are used as test participators. Generally you don't want people to know which group they're part of in an A/B test, as that can skew the result.
In theory, you should be able to create new accounts until you're placed in one that's in the 2.0 group, and thus be able to test the new tree. Keep in mind that if you do this, and then get bored with redoing the basics and quit after a few skills, you're going to skew the results in a negative way - making tree 2.0 less likely to succeed. So please, if you're going to do something like that, at least stick with it and give us a positive bias instead. ;)
I'm actually a little surprised they (because I'm assuming this is coming from Duo Central) (the Duopticon?) require an A/B test. Given how sociable users are, it will probably very quickly become apparent to them who's in one test cell and who's in another (and consequently a lot of new users may get annoyed that they're not in the 2.0 cell). I suppose one solution is to just be open about it, but that kind of messes with the data, as you said. Mightn't the better course be to just do a beta test of 2.0?
You'd be surprised at the number of users who aren't even aware that the fora exist, much less find their way to this post. You're part of the vocal and curious minority. :)
Personally, I don't think an A/B test is necessarily a bad idea, I only worry that the metrics that are tested aren't the ones that should be tested. Either way, I don't have the final say in it.
Any chance Tree 2.0 will include microphone/pronunciation exercises? I'm not tech savvy, so I have no idea how challenging it is to add those features, but it would be very helpful...especially since opportunities to practice speaking are hard to come by hear in Philly, PA, USA ;)