Reliability prediction and 'Deepwater Horizon'; lessons learned

My guess is that only very few of the readers are not aware of what is currently happening in the Gulf of Mexico. Of course the most important effects of this disaster are on the societal level. The loss of lives, the harm to the environment and the impact on the people, directly and indirectly involved, make this event a true disaster. Apart from these, there are also issues. Many professionals fear that this disaster will also backfire on the profession of Reliability Engineering. The oil industry is subject to very strict standards that explicitly prescribe the use of certain models with corresponding parameters according to which a disaster like this, statistically speaking, should not have happened. Still, things did happen. For this reason, many people are currently criticizing the profession of Reliability Engineering; in the press journalists publicly question the value of the models, the parameters and often the entire approach used in predicting and mitigating events like this. In order to get better insight into the reasons for mismatch between model-based predictions and actual field events it may be useful to look slightly deeper into the backgrounds of the models used. One of the problems that professional reliability engineers face is that the models used for complex cutting-edge systems very often have a high degree of uncertainty. On the one hand, the empirical field data is often lacking; on the other hand, for economic reasons, systems are often operated at the limits of their capabilities upto a level where the validity of the models that are available can be questioned. Both these facts may cause uncertainties of very different natures. In order to manage these uncertainties in an adequate manner knowledge about this is essential: