To understand Tinderbox guides, imagine that, as you’re working, a bunch of friendly old people are looking over your shoulder and offering helpful advice.
“That’s almost the standard spacing, but not quite: you should nudge it a little to the left.”
“Well, you know, those two notes are almost the same size.”
“It looks like you’re making a list!” (Just kidding.)
When several of these advisors, called “kibitzers”, apply to the same task, they yield in order of a (fairly arbitrary) priority. In the case of your video, I think that the “same size” kibitzer (which is looking at the blue note above) is taking precedence over the “bottom edge” kibitzer that you wanted. Or, possibly, it’s the “golden ratio” kibitzer.
Shift-drag will let you do what you want here. Kibitzers are really cool, but sometimes the old guys are just wrong.
(I had planned to let users adjust the priority of each kibitzer, but that’s just too complicated. As far as I know, no one knows which kibitzers ought to have priority, but if someone knows of some relevant literature — cf. constraint-based graphics and constraints in UI — I’m all ears.)
For perspective, kibitzers are chiefly intended to set up rough and ready geometries across large maps, rather than the sort of detailed alignments you’re doing in your example. For these, I sometimes use the Align menus for alignment, and sometimes use stamps or Quickstamp for setting the size and shape. On the whole, I think the large-scale geometry question is more interesting and perhaps more meaningful over the long run, but there’s very little research.