We must do check list testing, but too few teams evaluate and evolve their check lists. Do they still make sense? Do they still provide the best value? Some things, e.g. virus checking, we must do because although the risk may be miniscule, the failure consequence (and negative publicity) are astronomical.
“Strategic” testing is a deep strategy insight into where to look for the bugs, and the most efficient tools for doing the digging.
The principal problem with Strategic Testing as described, is that most teams lack the data to make decisions upon. We gather the wrong info. When we tried to compare Model Based testing with traditional hand crafted testing, we could say that the Model method found X bugs in Y weeks, but we were hard pressed to find comparable data for hand crafted testing. And it gets more complicated as you look at priority of the bugs found, etc.
Some of our instincts about testing turn out to misguided.
There were several enjoyable talks for me this year at PSNQC. The proceedings are at http://www.pnsqc.org/proceedings/pnsqc2004.pdf
Is all pairs testing really better than random?
See PNSQC talk Pairwise Testing: A Best Practice That Isn’t
Are other partitioning methods, e.g. based on code coverage, really better than random? See
V. N. Nair, et al, “A Statistical Assessment of some Software Testing Strategies and Application of Experimental Design Techinques”, Statistica Sinica 8(1998), 165-184
Simeon C. Ntafos, “On Comparisons of Random, Partition, and Proportional Partition Testing” IEEE TRANSACTIONS ON SOFTWARE ENGINEERING, VOL. 27, NO. 10, OCTOBER 2001
So the statistics are HARD. But most teams could do MUCH better in the analysis realm.
See PNSQC_2004 "Becoming a Better Tester"