You're absolutely right about that. It would make more sense for the players to actually use the knowledge first-hand. ... I'm definitely rethinking the classroom idea now. ... For example, say I want to teach the player how to play the guitar. In a minigame or puzzle game format, he or she could play a live show on stage at night. ... I think that might be more fun than going to a guitar class in a classroom for sure. ...
I think I have a real game design now. I'm removing the classroom for all skills that don't have a practical use there. Instead, I'll create individual minigames for various locations.

Which is what I said in the first post.

Learning through application of knowledge in fun/exciting situations that most people never get to enjoy (like being in a rock band, or being in some fantasy world of dragons, or whatever it is) is both more engaging, and cements the knowledge better through use.

I'm removing the classroom for all skills that don't have a practical use there.

With a little creative thinking, you will soon find that there are no skills that only have a practical use in a classroom. If you can model the skill and evaluate it at all (that is, if it has any use in a game), it can be better modeled in a practical situation through a puzzle and skill-based resolutions in narrative gameplay.

As I said before, that doesn't mean adding 2&2 to kill zombies. That's a bad example. True application makes the knowledge and skill relevant, and still models the real elements of the skill or knowledge closely enough to be educational.

After finishing a song, the player can be scored based on accuracy or creativity (I'll have to think about scoring methods)

Accuracy: If you're just following sheet music, that's fine.

If you're letting them go off script or make their own music (which seems to be what you're suggesting), you will find scoring to be a monumental task which the brightest minds in music and computer science have yet to fully solve (although we're getting closer to computer programs which can evaluate music, I don't think we're quite there yet).

If you keep to the scripted format, you'll be safe for evaluation, but it will also remove a lot of the reason for players to work together or learn the fundamental concepts of music, since all they have to do is give the right inputs at the right time. This can hamper motivation outside the rockband demographic of a party game.

Working the music and concepts into the narrative as a way to advance the story or overcome other obstacles can help rectify this: You need to construct and outside reward and additional feedback for successful mastery. It doesn't matter if that's casting spells, or constructing some kind of dating sim. This reward will help define your ultimate demographic.

If you're letting them go off script or make their own music (which seems to be what you're suggesting), you will find scoring to be a monumental task which the brightest minds in music and computer science have yet to fully solve (although we're getting closer to computer programs which can evaluate music, I don't think we're quite there yet).

+1 just for this.

If you are a single person team, I think the more important thing is to be consistent and keeping it simple. (at least initially) instead of coming up with great ideas that will probably need a team to implement.

So my advice would be to keep the design simple. I think RPG models will best suit educational purposes as RPG itself is about character development. The user will more easily get attached to his/her online character in that type of mechanism. So, I think the big idea in this educational games is this "Create a REALLY GOOD! game and integrate the educational model within" so that the "students" will have to study hard in order to achieve good results or ranking in the game. The overall experience will be rewarding because the game itself is rewarding.

Ok, think about WoW. Why do people spend too much time to achieve a top-notch gear score, or like do quests for legendary items like crazy? Because, they want a higher dps (damage per second - a score to measure player's overall ability / or healing per second etc.). This is a way to show off that you are successful in playing WoW. As they succeed they improve their gear and get better results. This cycle goes on and on. The WoW mechanic is this simple. (Note here: many games successful today are directly showing an overall measure of your success so that the player gets the sense of improvement as that score increases - example: in League of Angels(called battle rating) and many more)

So what I propose is, students would need to study hard in order to increase their "ability" in the game. Like, in order to get a legendary item you would need to achieve a near perfect score in a math mini game. Of course, there has to be a glamorous game behind this to work. Let's give another example. There are dailies in WoW that you need to complete every day to get good gear for your character. Make that daily quest to study, i don't know, chemistry. That mini game or even a like dull, coursera like format will be enough. I bet the student will have more willpower to complete that daily and get his reward. Once he/she is done, he can play the game as liked with better gear.

So to sum up. I think, playing the game should be an end reward. But, in order to enjoy playing the student would have to study. As he studies, he will be more successful in the game and he will WANT to study more.

Here is a game that uses this mechanic to motivate people to be productive: https://habitrpg.com

Yeah, I think overall it is well implemented. But, the game itself is not that rich to stimulate that much motivation. Now, think that instead of pixel graphics there are these fancy graphics and a really addictive game mechanic like wow.

If I was head of Blizzard I would definitely try some stuff like this. There is too much game-play and motivation going into waste.

But meh, as a single-person team I better not to attempt. I can barely complete a mediocre browser rpg