Questions before designing a dispersal protocol

Some of these questions may be trivial, and easy to answer or have already an answer. This is just a list of things to keep in mind.

Is the topology of the network already defined?
If not, is it part of the dispersal research to also define the network topology?
Or will the network topology be defined somewhere else altogether?
How many dispersal nodes are we targeting?
Are dispersal nodes under our control? Or are they decentralized/permissionless?
Do dispersal nodes need to be incentivized? How?
Erasure coded chunks are not necessarily encrypted - do we need (optional) encryption?
Dispersal nodes are NOT long-term storage node - or are they? Could they be, if not?
How many are “enough” confirmations?
Recent vs obsolete data - what to store?
Semi-AVID: Can the setup be run for nodes joining later? Or is it required for every node “at the beginning of time”
Redundancy?
Reliability (can chunks get lost - retry?)
Frequency?
Security? Can everyone connect to the storage nodes? Retrieve from storage nodes?