Umm... I think we need some more information about what's going on here (whether walking really is faster than jumping, what is meant by "horrible desync issues", etc.) before anyone can make a good judgement about this one.

whether walking really is faster than jumping

Turns out that jumping is faster. During each jump, the horizontal speed doubles for 4 frames in total. While this can't be caught by naked eye, it does save about 15% of the time during walking. Since the game uses many different addresses for the X and Y position of the player, I've decided to write a quick lua script that automatically tracks the addresses and shows them on screen: User movie #46843624007839304 With some tweaks, the same code could be used for other games that feature the same problem. Additionally, I want to point out a couple of other improvements, however there would save an extremely little amount of time compared to the one I just mentioned above. I want to point out a little trick I've found: during level 4-2, these is a place where you can skip jumping on a branch, if you use the correct timing and keeping the jumping button for the maximum height. (the submitted run does jump on the branch on the left first.) Also, I want to note that a very small amount of frames could be saved by jumping at the end of the slopes, since these make your speed a little slower at the end.

om, nom, nom... crunchy!

I hope Jigwally redoes this, it's so short and the improvements will be easily implemented.