I don't have much more to add than Nick and Steve.
There's definitely some forward motion on the project on display here in the number of situations that are being explored, but - as Nick pointed out - there's a lot of "level rules" you should be laying down. For example, it is FAR too easy to lose blocks and have them fly everywhere. The puzzles that work are the ones that are smaller and more self-contained.
There are some timing challenges and dexterity challenges emerging, which has the potential to produce the "Frankengame" effect where everyone pushes in different directions. Neither direction is bad, but make sure the team are in discussion and stay together with the direction of the project.
Finally ... just make the physics objects a whole lot heavier (or increase gravity). It'll reduce the likelihood of them bouncing around, and they'll feel more satisfying to play with.
And don't design puzzles around glitches or twitchy responses. If the player knows the solution (and your core challenge is mental, not physical) then it's just frustrating to have to sit and attempt it over and over until the physics lines up, or my stupid fingers can time things right.