The Floor may not be Jelly - Thoughts on breaking your own game's rules

I just recently played The Floor is Jelly by Ian Snyder. It's great and you should totally play it. I recommend watching the trailer. Not only will it hopefully convince you to buy the game, but it also pretty much explains what the game is all about (so I won't have to bother doing that myself).

The Floor is Beautiful!

Minor spoilers ahead, I guess... (next two paragraphs)

I enjoyed most parts of the game really, but I found the last world to be especially interesting. In this world, the "jelly" of which the floor and walls are made, changes properties every few levels. In one level it's hyper-bouncy, in another one it doesn't bounce at all. I'm not gonna spoil any more, but it gets really crazy at some point.
I loved these sections because they're a great example of how much variety you can get out of an existing mechanic (in this case: jumping around on jelly) without having to add new secondary mechanics. In fact, the last world doesn't use any of the secondary mechanics that have been introduced in previous worlds (e.g. water), and instead focuses solely on the jelly itself and possible variations of its attributes. Granted, some of the sections are less fun than others, some even barely playable, and I personally would have cut out a couple of those parts. Still, in my opinion it manages to be one of the more entertaining and gameplay-wisely diverse worlds of the game.

Also, I very much liked the theme of the world which is somewhat glitchy and pseudo-broken. It fits the gameplay perfectly and I'm guessing this might be a hint as to how the concept for this world came to be. As someone who makes games, you know those moments when you tweaked some variables and suddenly your game behaves in some weird, unintended way. Most of the time, you'll have a quick laugh, maybe even take a screenshot to document this glorious moment, then fix your mistake and get on with it. But sometimes, you might find something really amazing by accident.
Bugs turning into features isn't an uncommon occurrence (e.g. Tribes: Ascend). But actively looking for interesting new gameplay within a working system can be even more effective (and more fun). It doesn't even have to be complicated either. I'm guessing in The Floor is Jelly it was just a matter of changing some constants to make the floor behave in a different way, and yet it allowed for such diverse gameplay.

I would love to see this approach to game design realized in more games: Build a solid working system that can stand by itself. Then, when you think you exhausted everything you could do with it, instead of adding new features, play with the ones you already have. Go crazy with it. Think about possibly interesting "What if...?" scenarios. Then try them out. In case it doesn't turn out to be as fun as expected, you can still always go back to status quo. Either way, you'll probably have learned something from it.

This is actually how the concept for Snake Pit was born, except I took an already existing system and tried to change the gameplay to be something completely different. What if the game didn't end when you bite your tail?
Also, the idea for some yet-to-be-released bonus levels in DYO came to be in a similar way...

What if the minotaurs weren't of the same size?