Query for Insight: Effect approach.....

[biglittlepictures]

Hi smart people,
I am working on an effect.  I'll try to describe it simply, so I'll lay it out in steps...


01- Imagine two teapots with an array of 9 sourceBoxes arranged in a grid, between them.
02- I'd like one teapot to emit 1 particle(s)
03- The particle(s) is a ball, in a smallBoxA, in a mediumBoxB, in a largeBoxC.
04- This particle(s) would seek out a random sourceBox to fly to.
05- Then when it arrives at the first random sourceBox target, largeBoxC deletes....
06- and that sourceBox sends it to a second random sourceBox (a rule of sending it to itself wouldn't be allowed ideally)
07- Then when they arrive at at the second target, mediumBoxB deletes...
08- that sourceBox sends it to a third random sourceBox (a rule of sending it to itself wouldn't be allowed ideally)
09- Then when they arrive at at the third target, the smallBoxA deletes...
10-  then this sourceBox sends it to the opposite teapot.
11- Teapot births a new particle(s) ...starting the process over.


This seems like it could be pretty optimized via a potential script vs node tree, but not sure how.  
Also am unsure how to tell step 11 to talk back to step 2.
Also unsure how it'd work if i scaled it up to have (for example) 50 teapots... in this case, how can I assure the "sender" and "receiver" teapots won't be the same?

Any thoughts or recco's appreciated!
-Joel