Many technical problems involved with Component Based Software Development
might be considered solved these days. Less so are mental reservations both,
developers as well as their managers have with respect to the risks involved
with building software from powerful building blocks. 

This paper addresses the issue of trust developers might extend towards
components. Besides psychological aspects this issue has a technical
aspect. We will focus on the latter. After a brief review of mechanisms to
describe software components and the relationships between descriptions and
certifications, the paper zooms in on "selfdescriptive" methods, i.e. methods
directly involved with the executional properties of software. 

The approach presented in detail is tuned towards fine-grained
distinctions. It lends itself particularly to questions of version
discrimination and software maintenance.