Suggested model: Feynman diagrams.
Many people have seen these squiggly little diagrams of subatomic physical processes, such as for example x-ray radiation scattering off of an electron. Each individual diagram has an order, related to the number of nodes in the diagram.
In one domain of physics, quantum electrodynamics, each higher order contributes a smaller incremental amount of accuracy to the calculation. Roughly, one might imagine first order gives 90% accuracy, second order gives 99% accuracy, third order givers 99.9% accuracy, etc. In this world, you don't expect the second-order processes to blow up in your face. (technically, this is due to small coupling constants)
However, in other domains (quantum chromodynamics, which describes processes inside the proton, neutron, quarks and gluons etc.), the second-order processes can be just as, if not more, influential than first-order processes. Now you have a problem if you ignore second-order processes, as the first order calculation might only give you 10% accuracy, the second gives 20% accuracy and so on.
When second (or third) order processes are not insignificant, that's where they can return to bite you in the seat if you ignore them, aka 'blowback'. The difficulty then lies in determining whether or not such second order processes have this potential effect in the system of question.