Very nice stuff, I suggest you add an email address so that interested employers know how to get in touch with you 

@Charlieton: Thanks!    I guess I forgot to explain the premise of it.  In short, it was testing the output assembly language from the SDCC C compiler, targeting the zilog z80 (as a representative of other 70s 8 bit processors still widely used today), and testing certain handwritten optimizations (that can be implemented with a strict rule set) to see which ones were most effective, and which ones were most able to be implemented into compilers targeting the z80 (and similar processors.)

Overall I got 1st place in computer science, along with another student, so it looks like I'll be going to states!  I also got a 1st place corporate award (and money) from MIT, George Mason, the Central Intelligence Agency, the Air Force, in addition to a few honorable mentions and a grand prize from IEEE.  Ten times better than I expected, I was just hoping to place!

@Seiseki: Thanks!  Not sure game design will eventually turn into my profession (I'm aiming more towards a CS and an EE/CE double major), and my pixeling/art skills are still rather lacking, but it's definitely a forming hobby of mine  Just gotta get back to drawing and/or find some good artists to work with.

These are some vague questions, so the best I can offer is vague answers.  Not to mention they're going to have some of my own bias included. 

What would add replay value?

How many levels should there be?

How difficult should the game be?

What are the good things you've seen in shooting platformers?

What are the bad things you've seen in shooting platformers?

What would you like to see in the future from this genre?

A few things; huge worlds that are fun to navigate and explore have always caught my attention.  Having some sort of a Newgame+ thing that adds new missions, open new parts of the world up, etc. can assure at least one extra playthrough.  Many games have bonuses hidden all over the place, and let you replay the levels so you can get them all; this method generally annoys me unless the bonuses are realistically obtainable and unlock something well worth the effort.  Of course, the baseline should be that people want to play the game again because it's fun.

As many as needed.  Between quality and quantity, quality is the most important of the two; if you need to make less levels so you can make them better, that's definitely preferred.  If the levels are good enough, they'll make up for the lack of quantity.  Just look at portal.  Of course, on the contrary, having a lot of levels in a game feels good for the gamer, because then they can get very immersed in the game and feel like they're getting a lot of content.  I'd suggest getting a good base set of levels/missions done for your game alpha/beta phases, so you can get some idea of what your potential buyers/players think of whether or not your game is too long, too short, or filled with crappy levels.

Depends.  If you're going to implement different difficulty levels that the player can choose from, then make sure that the different difficulties are tailored to make sense, with rewards for those who play at the harder levels.  If not, start things slow and increase in difficulty throughout the game, with a learning curve.  Again, the alpha/beta stage is a good place to get input about this firsthand.

Some good things I've seen:

- Varied environments
- Intelligent enemies
- A degree of involved exploration
- Many weapons, but not so many that you don't have time to master them all
- Non-cliched storyline
- Leveling system based on collecting rare upgrades (Metroid at all?)

Bad things include the opposites of the above, along with:

- Points system (I always hate it when large games focus on collecting points; I'm a rabid hater of highscore mechanisms in non-arcade games)

What would I like to see in the future of this genre?  Personally, I'd like to see less sidescrollers with are basically Call of Duty in two dimensions.   

The palette thing is, no matter what I do I end up with several hundred to a thousand colors in an image I
thought I had 20. And I end up having to reduce it by force when I save an image and then go back over it,
fixing colors and ending up at a few hundred again.

This sounds more like a problem with the pixel-pushing program you're using, or the tools themselves.  Make sure you're not using dirty tools (tools that increase color count with little effect and/or push pixels in an unpredictable way), and instead stick to using the clean, precise ones.  For an example of both, just look at the tools in MS Paint: the pen/brush tools introduced in Win7's paint are considered dirty; the pencil tool is clean.  If you zoom up to look at what both produce, the pencil only uses a single color, and the brushes do a lot of anti-aliasing automatically, using hundreds of colors to do so.

Another possible problem could be that you're saving your images as JPEGs; doing so will produce artifacts, which basically ruin color counts and mess everything up.  PNG and GIF are good formats; PNG if you're using MS Paint, because paint isn't able to handle GIF well.

You could always look into using tools dedicated towards pixel-pushing cleanly; GrafX2, GraphicsGale, Cosmigo Promotion, etc. are often cited recommendations around here.

The art itself is quite fine, though I'd question the validity of the tutorial itself. The "Draw X, Do Y" style tutorials are rarely detailed enough to provide information. For example - how does someone reading this tutorial how to go from oval tool wheels to those fully detailed ones? How would they know how to correct their characters, or even what to make?
What experience level are you aiming your tutorial at?

Definitely agreed, and by extension, I would think that a pixel art tutorial would mostly focus on pixel-level techniques (which you didn't cover well), rather than general art skills.  The latter is definitely good, and the starting construction idea is an essential one, but since things like that hold true in just about all mediums, why not focus mostly on the refinement stage instead?