Hello Sam
You’ve got a good eye for detail. The RFC 2328 does indeed say that the cost should be equal to the largest cost of any of the component networks. However, in practice, most OSPF implementations, including Cisco’s, use the cost of the subnet with the lowest cost that falls within the summary range when calculating the cost for an OSPF summary route. This is done to ensure that traffic is routed through the most efficient path.
So, while the RFC does state one thing, the actual implementation can vary depending on the vendor. In this case, the lesson and your lab findings are accurate for Cisco’s implementation of OSPF.
This example emphasizes the fact that vendors are not obligated to adhere absolutely to the definitions of protocols in the RFCs. However, those factors that enable compatibility between devices of different vendors are more important as far as ensuring interoperability goes. Vendors are free to tweak their own implementations, and Cisco seems to do this quite often.
I hope this has been helpful!
Laz