As the first step, we added to cM (semi)exhaustive and random search strategy. We would like to add various focused search techniques starting from our probabilistic approach (see FOTP2005 TKMP2011 FT2010 or validating other existing techniques such as genetic search, hill-climbing, etc.
We prefer probabilistic approach because it is very intuitive and allows to quickly reduce dimensions in complex multi-dimensional spaces and focus search on areas with most likely speedup. To some extent, it is similar (or possibly equivalent) to ANOVA, therefore we may be interested to connect cM to off-the-shelf ANOVA exploration engine.