If you have 100 hours available for testing, you can use static analysis to find 90% of the bugs and spend the rest of your time on the 10% that require deeper insight. Or you can waste 90% of your time being a human compiler, manually cross-checking symbols. Which is going to result in more reliable software?
That assumes that:
- 1) Static analysis will find enough problems to save 90% of your testing time
- 2) There is no additional cost in development time in manually handling all type conversions
And there are a lot of reasons to believe that neither of those are true.