I am fairly new to Maya, I have been spending close to 10 hours a day practicing with the free tutorials, reading the Maya Help Documentation, and reading the few forums related that I can to learn as much as possible. After not knowing what I want to do as a profession for the last 4 years (since I graduated highschool) I finally realized I want to do 3D modeling, game graphics specifically. I have come up with quite a few questions in the past few days I haven't really been able to answer though.

1) What is the difference between 3 and 4 sided polygons? It seems some models (such as the dog tutorial) use 4 sided polys and other models (such as the World of Warcraft characters) use 3 sided polys. I am just trying to figure out why one is chosen over the other.

2) When using the Hypershader I create a lambert for the imageplane to use as a reference. I do...
Create Render Node > File > Load Image and sometimes it loads onto the lambert, and sometimes it doesn't. It usually works if I go through the process again, but it rarely works the first time I do it. Is this normal or am I doing something wrong?

3) Is there a way to keep the imageplane with the reference image from becoming semi transparent when using X-ray? Basically some of my reference images aren't too dark to begin with, and when I use x-ray I cant see the lines on the reference image because the primitive objects are darker than the plane.

4) What is the reasoning behind using the Delete By Type History command? On some tutorials they say when to use it, and some say not to use it. I am trying to figure out when you are supposed to use it, and why.

5) Is there a way to completely set everything in Maya back to default without reinstalling?

6) Nurbs vs Polys vs Subdivs? Im trying to figure out how one decides what to start with and what to finish with. Thus far it seems starting with Nurbs, converting to Polygons, then using Subdivs to smooth it out is the most common method. As far as I can guess Nurbs are the easiest to work with for characters atleast, then polygons are more efficient but sharp, and subdivs are used to make the polygons look better?

7) How can you check to see how many polys your model has?