Blockchain-like protocols are flourishing. Maybe not surpris-
ingly, the differences among these protocols are often subtle
and difficult to understand. More importantly, it is often
unclear what the weaknesses of each of these protocols are
and how easily they can be attacked. The goal of this paper is
to shed light on the important differences between blockchain
protocols and the impact these differences can have in terms of
their vulnerabilities. We cover well-studied protocols ranging
from those inspired from the distributed systems literature
(e.g., PBFT), to recent research prototypes (e.g., ByzCoin or
Algorand), including the popular Bitcoin protocol.
Towards reaching our goal, we first precisely define the
problem that these protocols seek to solve. Then we propose
a unifying scheme that captures, at a high level, the behavior
of any blockchain protocol. Interestingly, this scheme is also
sufficiently low level to highlight the important differences be-
tween these protocols. We show that blockchain-like protocols
can be differentiated according to their degree of indulgence—
i.e., tolerance towards node misbehavior or towards network
asynchrony—which translates into the different vulnerabilities
of each of these protocols.