There are always going to be witnesses since we are all around the World.

But what I was thinking was more akin to the maintenance of stability in case half or 75% of our Top20 witnesses get cut off for any reason.

Well, I don't know many details about all internal mechanisms present here, but I once saw a comment about Hive possibly having a mechanism that automatically shutdowns_{(by automatically setting the disable key or by setting said witness for a temporary suspension)} any witness that misses too many blocks in a row. I actually like that.

I far as I've seen, this mechanism doesn't seem to be working as we can see several active witnesses that never migrated from Steem missing all blocks since March/2020. But perhaps said mechanism is just for the Security Council_{(aka. the Top 20 witnesses)}, but I don't think said mechanism actually exists.

If said mechanism doesn't exists and developers feel the necessity to implement such a thing, My suggestion are as follows:

• No witness can miss more than 30 sequential assigned blocks after last witness update. This is only half an hour of instability.
  • If a witness misses the 29 sequential blocks, the 30^st missed will trigger a virtual operation that suspends block production for said witness.
  • The witness can always remove the "suspended" status by sending a price feed or a witness update as "proof-of-life" to the network.
  • Whenever the network receives a proof-of-life, the countdown restarts no matter how many sequential blocks the witness already missed before reaching 30.
    • If witness misses 29 blocks before successfully signing 1, the witness needs to miss another 30 blocks to be suspended.
• If, let's say, over 80% current Top20 got temporary replaced by another temporary Top20 due to suspensions, no major policies can be approved by the replacements, like a Hard Fork until the temporary top20 get the actual votes for the Top20 positions either by getting from the community or by waiting for the official Top20 votes to expire on their own.

I know it's very unlikely for developers to prioritize or even consider the idea of making Hive an "Apocalypse-proof" blockchain, but I find this an interesting thought exercise.