Hi Jim - admittedly, it is a tad obscure. There are two references there which I think are relevant:
"If for some reason the queen is lost in one hive the bees will migrate over to the other side to strength that nuc."
"The nucs in one box need to be given both a queen cell or a queen. It works best if you don't gave one side a queen cell and the other a queen. If you do, then most of the bees will migrate over to the queen side and the nuc with a queen cell will find it hard to except the queen cell."
I read that as indicating that bees from one half of a shared box will 'abscond' (or migrate) from one side to the other if they consider that their chances of survival are better there. This being in marked contrast to a setup with two separate nuc boxes in which the residents of one box are undoubtedly unaware of conditions within the other, and so absconding/migration doesn't then occur.
He goes on to say:
"At this point a solid inner cover is over the top, the bees can't get from one side to the other. Once the queens on both sides are laying, the inner cover is replaced with a queen excluder on top so the bees can mingle." And that's the bit I find extraordinary, that he actually makes it as easy as possible for the bees to abscond/migrate, should they ever detect that there's an advantage in doing so.
Must say I would be a tad wary of selective winter clustering with his set-up, with a risk of one queen effectively being abandoned - as I've been reading some early experiments conducted around winter clustering, and one finding which stands out was that bees tend to be attracted to cluster formation itself, such that a larger cluster is more attractive to unclustered bees than a smaller one. Again, that suggests a basic survival mechanism: that a large cluster has a better chance of survival than a smaller one. Size - it would appear - is 'everything' in the honeybee world.
'best,
LJ