Comment by obel1x

Which part of the architecture causes the latency? The author says it's between 2 and 13 seconds and is faster the more users there are - why would that be the case?