1) If everyone knew the IP addresses of BPs, then clearly this would put them at risk of attack. But do BPs themselves know each others IP addresses? In order to speed up relay of blocks to each other and reduce time to confirmation (although I guess then they could attack each other or give away IP address of others etc.)
2) What is the number one thing that prevents the reduction of block interval and/or confirmation time? I guess since transaction validation is fast it would be network latency, but it would be nice if someone could lay out the relative contribution of different factors.
3) How often are block producers decided on? (although I guess this is fairly stable unless there is good reason to vote otherwise). I ask as, once BPs are decided on, then if a block came from a block producer and is validated by most or all of them under bft then we know it is valid so we need not wait for network to catch up (hence low block time, and hence Q1 on whether bps are better connected)
4) This leads to the question of whether there is a greater time delay when votes for bps are counted? When there is potential for change, for the above to be correct the whole network must agree who the bps are - in practice it is unlikely to be a problem but if the list of bps completely changes then I guess part of the network could trust a different bp set