I like how the twitter one is titled "Portal's physics engine rebuilt in 50KB—on a graphing calculator" and then right below it: "Portal's physics engine rebuilt in 25KB—on a graphing calculator" and in reality it is under 16KB

One byte would be better in my experience. We might have >256 sprites, but we probably won't need more than 256 tiles. That initial x,y will be very useful, too. Are those bytes or words? (I am assuming they are the width and height)

Well, the trouble is that I have finals next week, and I go away to visit family from the 20th to the 30th. My inlaws live in a place that doesn't have internet and may or may not use electricity, my family lives where internet basically doesn't exist. This means I will need to wrap up my Apocalypse contest entry early, so I will have finals and the contest that I will be concentrating on for the near future.

However, I have difficulty concentrating on what i am supposed to do, so maybe I will accidentally write an engine

It looks 3D-ish, but it's actually just a tilemap in which the tiles are drawn from a birdview persfective instaed of top-down. The only thing that's handled differentely then on normal tilemaps is the collision. This makes that kind of games perfectely doable on calculators.

And I know how to handle the collision detection for the map! If I have time this weekend, I will try to make a scrolling map with a random AI thing that tries to chase you as a proof-of-concept.

I was originally thinking 32x32 or 24x24 for characters I think we should use 16x16 tiles, even though that means we won't have much to show on the screen at any given point. I made some rather nice tiles for an old game that were 16x16 that I might be able to share. I might also have to get around to writing a scrolling map engine >.>

I like the detail in 16x16 sprites, especially grayscale. It also makes the world seem so much bigger and you feel closer to the game

I got that as well. I should have tested if I could press [2nd][ON] >.>

You should post the link to the Wiki in the first post.

Yeah, I would suggest just running it from a shell, but that makes for a very interesting idea. Assembly programs are normally run from the OS by copying the original program to 9D95h. I think there should be no problem with archiving the original as the OS does not do any writebacks. Here is the hex code:
Asm(EFF14278B72003EFD84F
That must go at the beginning of your program! The reason is very simply that when the OS runs you program, it copies its name to OP1, that way when your program starts, it can reference the original program (for writeback and whatnot). However, there are a number of OS routines that will destroy OP1.

That should work, but I didn't test it.