Although I'm still using 126.96.36.199, I'd like to support Cornan's request that developers take another look at this issue.
I notice that when I switch from Normal to Expanded View, most of the active thought's parents/children/jumps are clustered very closely to it, but one or two are considerably further away. There are three issues:
1) Mouse wheel resizing has no effect on the main cluster, as Cornan says.
2) Mouse wheel resizing works only on the thoughts that are already distant and pushes them offscreen.
3) Close clustering hides some children behind others, making them impossible to find and originally led me to think that some child thoughts got left behind when I switched from Normal to Expanded View. This is the fault of the algorithm which, it seems, still cannot prevent thought hiding from happening. I guess that spacing everything apart a bit using the mouse wheel would have helped me find the missing thought, but for reason (1) this can't be applied, and as a result I have to drag each thought out manually in order to find the missing ones. This is unacceptable.
I seem to recall that this was a big problem in v5.0, but I thought the layout algorithm had been fixed.
PS I'm running on Windows XP with a Logitech M400 laser mouse using Logitech SetPoint Control Center software v4.60.122.