It allows you to run brokers of different versions in one cluster, to avoid a downtime of the cluster during the upgrade. Before you upgrade any broker, you set the inter.broker.protocol.version to the existing version on all brokers. Then you start upgrading broker by broker, the newer broker will still use the 'old' protocol to communicate with the other brokers. This keeps the cluster functional during the time when just some brokers are updated.
Once all brokers are upgraded, you change the inter.broker.protocol.version to the new version and restart them one by one.
More details here: https://kafka.apache.org/documentation/#upgrade