Difficulty drop DoS concern (consensus)

[trosos]

Hi,

If I understand the Chia Consensus correctly, there is a theoretical possibility for an epoch to contain no transaction blocks.
In this scenario, how will the next difficulty be computed?

Although unlikely, such situation can happen if the difficulty suddenly drops due to a drop in network capacity. Such network capacity drop seems quite conceivable if we imagine some world-wide technical problems (e.g. implementation issue as the one which happened at block height 255108, or some kind of DoS attack, or a global power outage). When the network capacity raises back (causing the next epoch to contain so many blocks that none of them is eligible to be a TX block), there is nothing worse than not having a means to increase the difficulty back to normal: under such circumstances, the network would be practically unable to create any TX block any more.

As a natural solution, one might imagine a consensus where the difficulty is increased by factor of DIFFICULTY_CHANGE_MAX_FACTOR whenever the previous epoch contains no TX blocks.

[trosos]

I missed the fact that the first block of an epoch must be a transaction block, because there are no overflow blocks at epoch boundaries.

I am sorry for the noise.