Oddbean new post about | logout
 That would be a tricky consensus change. I think it would be easier to implement additional safety limitations per Rusty's Great Script Restoration project. 
 Thanks for the response! I would have thought it was just policy and not consensus since the problematic block has not been validated in full yet. Given a valid new block header, my node will ignore any other header of the same height, even though it does not know that the first block received is fully valid, just that the header is. Assuming an implementation in which blocks of the same height can be validated in parallel, all this means is changing what the node does in the event of a tie from “first seen” to “first validated”. What am I missing?