Great, thanks for the explanation!
Let me see if I get the gist of how the network works. All the DLUX network data (token balances, etc.) is stored as a state file. Each node hosts a copy of the latest version of this state file and makes the file accessible through the IPFS network. When a user makes an operation, the frontend would allow writing to Hive (custom json) and to the DLUX state file. The DLUX nodes then (I guess) verify that the Hive operation was successfully completed on Hive, and also verify that the transaction is valid as per the DLUX network rules. If so, they update the state file to include the new operation. The nodes also of course allow all performing all sorts of operations (loans, etc.) that aren't possible on Hive alone. Is that more or less accurate?
Got a question. How are node operators compensated for running a server and keeping the network integrity? One thing I'm wondering about is the susceptibility of the network to collusion between node operators - say if a given token's price increases a lot and there aren't that many nodes, the node operators may find it more profitable to collude, distribute the token balances among themselves and quickly sell it on exchanges. How does the network defend against this, considering that there would likely be highly volatile tokens with low total supply on it?