I've never heard of "wall, ceiling, floor". Or ever seen someone drag guides out like that. I mean, it makes sense, but seem like a lot of time dragging guides which aren't really necessary.
The methodology I've always used, which is basically the same thing without guides, is to merely place anchors at apogees and corners. After all, all those guides the "Dougster" is dragging out are merely there to mark the apogees and corners. You really don't need guides for corners, you can see them. Curve apogees don't have to be an exact science. As long as you are close to a curve's apogee you'll get a good path.
Overall though, I do use the same basic techniques, just without any guides . . . Holding Shift to pull smooth points at constrained angles.

What strikes me about the video is the smart guides will not snap
to a placed raster image like they are snapping in the video. In order
for the snapping to take place and use it to your advantage, that screened back
letterform must already be vector in nature. So essentially the
video is tracing a vector glyph to make another identical vector
glyph on top of it. I supposed it's okay to show the technique. I wonder how many
would struggle with the fact that guides aren't snapping like they do
in the video though? Without the snapping a user will have to eyeball where the apogees and corners are in order to place guides on them... so why not eyeball it while drawing rather than eyeballing it while dragging guides? I'm probably digressing too much :)