The IPSO framework allows us to optimally design experiments and surveys. We discuss the utility of IPSO with a simplified 10-parameter MCMC D-optimization of a dark energy survey. The resulting optimal number of redshift bins is typically two or three, all situated at z < 2. By exploiting optimization we show how the statistical power of the survey is significantly enhanced. Experiment design is aided by the richness of the figure of merit landscape that shows strong degeneracies, which means one can impose secondary optimization criteria at little cost. For example, one may choose either to maximally test a single model (e.g., ΛCDM) or to get the best model-independent constraints possible (e.g., on a whole space of dark energy models). Such bifurcations point to a future where cosmological experiments become increasingly specialized and optimization increasingly important.