thank you very much ctbram, that is very informative.

You're right about them being in -1, because I often use -1 scale to mirror a duplicate object. Is this bad practice? I never thought of it as an issue before.

That's interesting about the STL watertight modelling, I didn't know that a floating hole would be so much of an issue.

I guess if I wanted to take the model into zBrush then I wouldn't be able to use a model using the method shown in the middle image because they need to be quads or tris. The right image is ideal.

When you say "This only applies if you only need a water tight model, do not need to smooth, do not intend to texture (you have uneven uv spacing so you would want even vertex distribution)" are you referring to the middle image only? Wouldn't the 3rd image smooth/texture? If not, what then is the best way to model it??