Hi Mira:

a very interesting paper. I find it useful in several respects.

What is the difference between your collaboration interface
and our participant graphs. "Each algorithm needs its ideal 
data structure" was the motto in APPCs.
Is that still the case in your new model except that the bindings
can now be more flexible, i.e. object-level and constructed on demand?

Why do you use the "override" keyword in Fig. 14? It seems that
you enhance, keeping what you already have.

-- Karl