The writers say that currently the "practices" on researchers in computational science are far from good. Scientific work should be reproducible and materials and instruments should have open and verifiable provenance. In computational science this is not (always) the case.

The writers point out that most departments feel that the basic methodology and skills should be developed, but in practice it is difficult to fix the problem.

As a key example, basic programming methodology is often lacking: understandable coding, version tracking, scripting languages, debuggers etc. We should develop the skills so that "computational scientists [...] can modify their programs continuously as their questions are answered."