Research in blockchain has been held back because we don't have good ways to compare max throughputs for different protocols. Pleased to announce a new approach with Kartik Nayak (@kartik1507) and Nibesh Strestha (@nibeshrestha2). 1/

The basic problem is that standard models don't take into account upload/download capacity. For example, in the `synchronous' model, any message of any size arrives within some fixed time. 2/

This means we have no way to formally analyse many performance claims: are DAG-based protocols actually better than single-proposer protocols like PBFT? Does the linear message complexity of Hotstuff actually reduce latency? 3/

Our "Pipes" approach allows one to calculate latency as a function of network bandwidth, network delays, the number of processors, and the incoming transaction rate. 4/

Each protocol has a "latency bottleneck": an incoming transaction rate at which latency becomes unbounded over the protocol execution, i.e., a maximum throughput that the protocol can handle without unbounded latency. 5/

The paper is available here: . Experiments to be added soon... 6/