[kantord]

I still think it makes more sense to implement languages that more people want to learn.

However, the long term goal is making it easy for the community to build courses, so once the project is mature, it should be possible to include a way larger number of languages.

I am also thinking of things like conlang enthusiasts being able to create courses for their own conlangs.

[mumblemumble]

I just wanted to chime in and say I think there's another good reason to focus on the more widely-spoken languages first: It's not just that there are more people who want to learn them, it's also that there are more people who can help contribute to the materials.

[kantord]

which makes it possible to create courses that have more content as opposed to being just brief introductions

[bluGill]

There are plenty of other places to learn the common languages though. If you have good content for less common languages there are not many choices and you become the default place to go.

[Vivtek]

I have some experience with Duolingo's course builder. I'd be interested in exploring better ways to do this.

[kantord]

I'd be interested in hearing about your experience with the Duolingo course builder

[Vivtek]

It was a couple of years ago, November of 2015 according to my notes. I was involved with the Hungarian language course for a while, but there were some in-group political problems and I ended up getting shut out rather abruptly.

The Duolingo course builder is a rather slick UI, but I found it brittle. Like many UIs, it does what it does, and then it stops. There's no direct access to the underlying database, meaning you can't do any kind of search on problem phrases for instance. You can't do any bulk updates of any kind. All you can do is navigate through the course structure to the sentence you want, and use a custom editor to make changes to it. Very clunky.

They have some kind of mechanism in place to flag sentences where learners have problems, but as I wasn't involved in Hungarian after it went live, so I don't know how they work.

There were systematic problems with the course material (it had been adapted from the English course for Hungarians), and I finally ended up writing a spider that walked the course and recorded the data in my own database just so I could do queries against it to highlight where certain issues needed to be fixed.

My #1 takeaway, and this would automatically be addressed in an open-source context, is that the database has to be exposed. I'd be happiest if the canonical course were actually defined in a document that had an independent existence from the live database entirely and could be version controlled. (Obviously multimedia resources would have to live outside that document, but you could have a descriptor for each one and use it within the course definition.)

By all means provide a builder UI to smooth the process - but at volume, you'll want some way to just work on things in text or you'll end up buried in technical debt. You might even want to model the builder on a Wiki, for instance - a set of documents that could be considered a single book made up of articles.

My time is limited (isn't everybody's) or I would promise you the moon in terms of cooperation on this project - I've wanted to see it for a long time. I hope I'll have the time at least to use the platform and provide some constructive criticism.