Over at ArchaeoBlog, Dr. Cagle is about to (re?)visit the survey method debates. This should be required reading for any archaeologist who conducts survey, especially should that archaeologist also plan surveys.
That said, the abstract to the 20-year old Lightfoot article summarizes the situation as it stands today. I hope that this changes in the near future, now that geophysical techniques are becoming more common. I doubt that they’ll ever replace shovel testing as the survey method of choice, at least not in the near future. The types of data identified by shovel tests are different than those generally found through geophysical survey. I suspect that at some point in the near future, it might be unthinkable to use only one type of survey when conducting large-scale survey. Instead, by using the two types, the different types of data would complement each other to provide more accurate identification of sites and site boundaries.