Routing updates are split in half to reduce the update time.
Information learned from one source is not distributed back to that source.
New route information must be learned from multiple sources to be accepted.
The time between updates is split in half to speed convergence.
New route information is suppressed until the system has converged.