They aren't here yet so let's not act like they'll be here before we have Turing complete scripting.
If the DACs are here already then they can provide liquidity to each other.
There is an advantage to a blockchain with its own internal market maker - speed. A blockchain market maker can react with 0 latency to market changes, whereas MM bots will always lag.
Yeah but then you cannot spread risk and benefit from the decentralization. Load balancing isn't just about speed but about risk. If one node goes down another node can take on the load of the node that went down until the new load can be brought back up.
This minimizes the downtime of any node in the network. Typically this applies to virtual machines but I don't see why it couldn't to a certain extent apply to blockchains. I don't see why our transactions for example should only execute on one blockchain when any blockchain which can meet our order could in theory take our order.
Also we have ACCT and aren't even using it. What is the reason not to use the feature if it's already built in?
I suppose the way to answer this question is to do some tests, some simulation where we have different test chains which attempt to do it with ACCT. If an internal bot is proven to be more efficient and more reliable then use the internal bot but I think the internal bot wouldn't be long term more reliable.