Thanks for the feedback so far; that's definitely what I was looking for, and helps.
I think there's one really good piece of feedback implied here that no one's stated outright - I really should have gotten this out there sooner. I had the game in a very playable state a month or two before the end of the contest. I could very easily have made a post about it and gotten a lot of this feedback earlier - pretty much everyone on this forum is really open and helpful about stuff like that. Doing that could have really helped shape the game in a positive way.
I did solicit feedback from some friends in the real world, but they gave mostly positive feedback - they don't know the internals of the console like the people on this forum do; they don't know that the movement feels slightly off because I'm not considering acceleration/etc. It's not their passion.
Now, onto the real feedback. I'm not going to quote everything as that'd make this post unbearably long, but I did want to comment on a few things.
I figured out the warping easily, but there seemed to be no logic to it. Warping from the same place could go to icy or heated, apparently according to what was needed next, but it still felt weird that there was no rhyme to what came next.
I had actually never thought about having more logic behind this at all. It's bloody obvious now that you've said it - I really don't know why it never occurred to me.
I think the only real problem I had with it is that there's no subsequent "reevaluate your assumptions" puzzles later on, in the style of the brick wall "learn the puzzle mechanic" at the very beginning. (by which I mean, either the first puzzle mechanic should have been better hinted, or there should have been more Real Puzzles later on rather than just platforming)
One of the friends I showed this to right after the contest had this feedback too - I think it's really valuable feedback, and one of the worst misses I had. I didn't spend nearly enough time on level design - focusing instead on rather inconsequential pieces of the engine.
* It'd be nice if there had been an extra ice block here so that you couldn't jump over it and get trapped on the left:
Ugh, crap... I was really hoping no one would find that. It's a limitation of the engine - I can't actually put blocks in that row. In retrospect what I should have done was moved the entire structure down a tile, and put an extra block on top. Too late now, I suppose.
The collision detection is highly flawed. I would hit item bricks from the side most times. Jumping on an enemy's head without being killed was way too difficult. In my own designs, I prefer to make collision boxes large for good things (killing enemies, getting items) and small for bad things (being hurt).
The item bricks bug I knew about but couldn't find a good solution for in time. You should have seen the bugs that didn't make it into the final release... this is the first platformer I've written for the NES, or anything really, and it really shows. The enemy stuff is definitely valid.
At any rate, I had never considered that sizing idea before - it makes a lot of sense and I may very well adopt it in the future. Thanks for the tip.
This last one is a minor issue, but you're setting palettes outside of vblank, which makes ugly color streaks flash on screen right before the level loads (only some emulators show this, but it's visible on hardware). It's an easy fix, wait for the start of a frame before updating the palettes. Even better, fade the palette in to ease the transition when a level starts, you're already doing that for warps.
Good eye! I thought I had caught all instances of that; I didn't realize it was causing problems until pretty close to the end of the competition.
To address the recurring feedback, lesson learned on the movement/physics. I thought it would feel natural enough without proper acceleration/deceleration in movement, and I was definitely wrong about that. I just got used to it because I spent so long staring at the stupid thing.
Last thing, to anyone who felt this game was a complete waste of their time, I apologize. Based on the detailed scores, I can tell I got a bunch of 1s - that tells me a lot. I can't give you that time back, but I'll try to make sure if I enter again, it's something worth playing. It was eye opening to see this game scored second-to-last in polish/completeness - far behind a game that could not be beaten. That really clued me in I did something (well, many somethings) really, really wrong. Perhaps I mean I thought about the entire contest wrong. (Note: I actually really liked Wo xiang niào niào, but not being able to finish the game at all seemed unfortunate. This soumds more negative about that game than I want it to, but I'm struggling with words here, sorry.) So, I'm sincerely sorry for wasting your time.
Thanks again for the feedback.