2. Mining Hubs: speaking of mining hubs, GHash.io and the others all depend on physical hardware to do the mining. In the event the center is compromised or physical hardware damaged there's not much of a failover for this. Unlike data which can be backed up in a redundant data center in another city, these hubs are susceptible to many types of disaster.
Miners should have set up a few failover pools (easily done with cgminer or bfgminer), so that when the primary pool is down, the hashrate will be pointed to secondary pools automatically.
So, problems on one or a few public pools shouldn't cause too much problem...
I'm not referring to miners pointing their hardware to a pool, I'm referring to the large scale operations that are housing multiple mining hardware in one location. The farms that rent/sell hashing contracts and the type like the one shown in this video
http://www.businessinsider.com/worlds-largest-bitcoin-mining-operation-2014-3. The one in this video houses 1 petahash of mining power. While it is a private venture, as long as the security & health of the total network relies on it then it's at least a concern of public nature. If the site was destroyed, taken offline for any reason, or even seized by force (whether governmental or otherwise) even insurance wouldn't be able to cover replacement of the FPGA's/ASIC's as they're perpetually on back order it seems.
So, just to clarify, I'm not meaning folks should make sure they have failover pools listed in their mining scripts. Just that farms for mining are here and they'll do nothing but get bigger further centralizing the network power. Since hardware actually needs to be replaced rather than backed up somewhere it brings with it an inherent risk of failure.