I try to avoid giving general statements about game design but it's almost accepted as a universal truth - Learning is fun. Which is what I was trying to do with with my LD48 entry The Mute King, but it fell short on many fronts. From that list of design errors and careless mistakes, I've shortlisted the following which completely ruin the experience, holding it back and not allowing players to appreciate the game where it succeeds.
A More Compact Possibility Space
The spell system in the game requires players to cast using mouse gestures on the screen. The game recognizes simple directional swipes and some shapes but none of this is ever communicated to the player. Players no longer want to try and demystify the system infront of them because it's possibility space has been inflated to a size which makes it less inviting and more intimidating. Learning by hit and trial can be enjoyable but not with an infinite number of combinations to try out, that's torture not experimentation. To encourage experimentation, a more limited and compact possibility space would be required, examples of games like Aquaria and Magika come to mind which do this really well. A combination of buttons or items on screen allow players to realize and understand the complexity of the system, a system that the players now know can be tackled - making depth more accessible.
A symbol in the center of the screen appears after a spell is cast successfully but the player is not always given any feedback on how exactly the spell works. Some spells are tied to a certain type of entity, for example the clone can only be spawned using plants but not trees or bushes. I was expecting players would experiment with the spells but I wasn't particularly doing a good job in encouraging them, since casting a spell without the intended result still drains the player's mana. So if the cloning spell is cast where there are no plants in the near vicinity, the spell drains your mana but doesn't reward you with any results, just a cryptic symbol that congratulates you for discovering a recognized mouse gesture. There are many ways to indicate what an action is supposed to do, one of which is an intricate tutorial screen with lots of information and text, again something I'd never do with my game.
The game could use a brief explanation of each spell but that might still not be enough to close the feedback loop. The spell system simplified into a couple of buttons with permutations leading to different spells would contain the open ended-ness of the gesture system the game currently employs.
This game was a great learning experience, I had these ideas rattling around in my head and the LD48 was a good excuse to try them out. There were other things I wanted to accomplish in the game but I could do only so much in the forty eight hours that were allowed. Players will eventually extinguish the game after they've learnt all there is to learn about it by trying every spell with every game object and their combinations which might result in new and interesting dynamics and behaviors. Its wishful thinking but all we need is a system that can procedurally create meaningful constructs that demand and facilitate learning, a game that doesn't only generate content procedurally, but gameplay and mechanics as well.