Thanks @fbslo
For instance, I don't think witnesses (or whatever name we give those running the Atomic Plugins) should get a fee based on number of transfer transactions they sign. This could easily be manipulated with a flood of low value transfers executed when a particular witness is scheduled to sign.
This is a very good point.
We are not only considering the number of transactions. We are looking into other parameters like uptime via a "statistics object". The idea is to observe the behavior and extend the parameters - we are running an extended public TESTNET to iron out this.
Further we are limiting/capping the daily rewards so even if someone decides to attack, the transaction fee on Etherium or Bitcoin networks will prevent such an attack. The daily reward limits are changeable and the committee members can do so.
Suggestions welcome. If there are new ideas we can adapt them.