Experimental data can correct modeling errors
In the implementation phase of a molecular systems engineering effort, developers will be able to obtain experimental data on the behavior of components (for example, the energetics and reliability of mechanosynthetic steps). In this phase of development, designers can use even inaccurate calculations to screen proposed components and processes, rejecting some as unworkable, validating some as almost certainly workable, and classifying the remainder as possibly workable. Experiments can then evaluate designs in the latter categories, reducing uncertainties and enabling subsequent system-level engineering to proceed with more accurate data. This greater knowledge can be used both to select specific component designs from an initial pool of candidates and to shrink the required error margins in design at the system level.