Currently all users are being encouraged to avoid sending their transactions to public mempools in order to avoid predatory front-running and sandwich attacks.  Users are being encouraged to send their transactions through OFAs or directly to block builders, or RPC providers that relay transactions to block builders.

We would be interested in proposals to better understand how this is currently working in order to test the assumptions of this approach.  For instance, a proposal addressing this topic can explore whether all transactions that are sent to private channels need to be sent there?  If not, what percentage dont’ need to be sent?

There is an assumption that sending transactions to block builders will protect you from having your transaction exploited.  Is this something that can be tested?  How many transactions are being sent to a single block builder instead of multiple block builders, which introduces latency of transaction inclusion and encourages centralization. Is there data that can suggest the reasons for this behavior and potentially point to a direction to mitigate it?