The November 2005 issue of Dr. Dobb's Journal has an interesting article by Dr. Charles Peck, et al., "Building Internet Distributed Computing", Dr. Dobb's Journal, November 2005, pp. 39-41 (not yet available online). The architecture and design are straightforward and show how existing open source packages can be used to distribute intensive scientific computing tasks across multiple, distributed processors.
In addition to being a software architect and developer I'm also interested in psychoanalytic and psychodynamic aspects of human behavior. So I was surprised to read that the metaphors used to describe the architecture included "servers", "mothers", "nannies", and "children". In this model "the mother is responsible for all communication" with the server and the nannies. The nannies monitor the children, and the children do the work. I know it's convenient to use common metaphors to describe computer systems, but using the words "mother", "nanny, and "child", when words like "coordinator", "monitor", and "process", or "task" would do just as well just re-enforces generalized cultural stereotypes. In addition, metaphors like these might not travel well to other cultures, making the software more difficult to read and harder to share.