Re, Ioncorpse's post. Thank God, I'm not alone in my opinion. I would much rather invent and calculate a solution than tweak & tweak & tweak, etc, until something falls into the right place and the solution completes. I will spend hours tweaking one node to get a good solution but with these multiple Dillo randomish levels it can get a bit tedious to tweak one node so one Dillo completes then tweak another node so the next Dillo completes only to put the first Dillo off course, etc.
I am very happy to have multiple Dillo's all with there own course to take, I know that is like having to do three or four levels in one solution, but to me that is better than having to seperate a group of Dillo's, which is chaos, and then guide them to their destination.
That aside, I have just figured out how to do this level without any random behaviour and it should be much cheaper than my first attempt.
EDIT. I agree Vankwish, it is controling this new state of chaos that is a new thing to all of us. I think I have got it in this solution as was the case in Waterfall. It could just be the begining of a new learning curve. Possibly it is just a matter of taking some care when designing new levels to be sure there is an organised way of solving it.