Consensus requires third parties to enter into consensus, and trusted third parties are security holes. For example, the security of the network's value integrity is left to the consensus among developers when we talk about BIP, or specifically regarding the Knots vs. Core issue. There is no true immutability. That's why I talk about security holes derived even from fiat regulatory issues that can lead to a legal ban on the very actors involved in "consensus." And ultimately, science itself is not based on consensus, despite what fiat academic bureaucrats would like.