Comment by justicz

(I work for Algorand) This is false. The BFT protocol that Algorand uses guarantees that each round may have exactly one block (up to our security assumptions with regard to honest stake); this is one of the protocol’s “safety” guarantees. In the case that a large number of participating nodes crash, the network will refuse to produce a block until a sufficient amount of stake begins participating again. Nodes that fall behind can use the standard catchup protocol — the one that nodes new to the network make use of — to pick up where they left off.