It seems like most of those issues could be solved with a distributed orchestration layer, the same kind used to power SETI@home, BOINC, crypto mining, RC5 decryption back in the day, etc. E.g. if the administration of billing, compute, and storage were decentralized and each home node could drop in and drop out as necessary.<p>AWS itself has nodes that can be preempted by higher paying users, no, with barely a few seconds to shut down your workflows?<p>You shouldn't misrepresent your home lab as an actual hosting business with staff and a data center and insurance and all that, but there still ought to be a way to loan out idle resources on an ephemeral basis.