Phil Murphy of Forrester recently published The Application Knowledge Deficit (subscription required). I won't reproduce his numbers here - it's not publicly available to non-subscribers - I would make two points:
Almost no-one relied on reading the source code to understand an existing system
This contrasts with the use of a business rules management system that allows far better understanding of what is being done with business users, as well as analysts, able to read and understand the business logic.
Very little code mining or transformation is being done
Part of me wonders if this is because the target system is still being coded by hand. If the developers knew that they had a way to manage the business rules in the new system, would they be more interested in finding out what the business rules were in the old system?
If organizations are going to free up any budget for innovation they must reduce their maintenance costs. While not all the code in a legacy application should be replace with decision services, some should be. Doing so not only allows the possibility of keeping large chunks of the legacy application working longer (because the high maintenance cost pieces are now managed properly), it also takes advantage of more modern thinking by using the right kind of technology to build each kind of service. "You can't code your way into the future" as a Gartner analyst once said. If you build the new systems the same way you built the old systems, nothing is every going to change. I blogged about business rules in the context of another of Phil's pieces here and about how Application Maintenance can be improved with business rules here. I have also blogged about using business rules to improve old applications on my other blog.