The unused power of SW safety&dependability analyses

In many domains, when software controls or implements safety critical functionalities, software safety and dependability analyses are needed in order to understand how it may fail, the consequences of these failures and how to avoid or mitigate these failures. But from experience in different projects, when these analyses are performed at the software level, they are not used to their maximum extent: they are always late and their content is often incomplete (just to justify the already assigned criticality level). They are not always considered as bringing safety and dependability requirements into the project. Existing techniques may be used along all software development and used for many different purposes to support the definition, verification and demonstration of software related safety and dependability characteristics. Therefore the questions are: When are these analyses required? Why are they so difficult to be performed? How are they finally used and why in such a limited way so far? This paper will analyze when these analyses are required, what for, which ones are required by ECSS and evaluate how to better use them in Space projects.