In the same vien as Seti, lets create a distributed simulation of randomly interconnted simulations communicating each other. The idea is to leverage the possiblity of self organization using feedback loops in complex systems to see what we get.
Each node simulates a series of seeming random and
meangingless calcuations ( which essentally they are ) which produces output which can be used to feed another node etc..
The random and meangingless calcuations can be done using a variant of genetic programming techniques which is seeded by some sort of realworld complex data set, say the human genome.
One node may just parse the genome into floats and feed it into the sin function which is it's output. Antoher takes a series of floats and depending upon it's past state and the data feed from the webcam, it will produce combine the data streams and feed it into one of seveal output buckets which it mangages. Etc...
Now the meaningless of this all, this project has no focus/purpose. It is just purely a manifisation of the unknown, in what will/can happen. The complexity/information/knowledge of the system derives itself from the real world and is thus hopefully unbounded.
It needs something akin to the peer-peer architcture to maintain its presistance, as there is always a chance the random simulations will crash(proably likely)
Hopefully after running for a few years it would evolve into something intresting maybe some sort of super calcuator.