I suppose it depends which conferences you attend. I can tell you from
personal experience (and the angry phone calls...) that papers from
"famous" people do get rejected, frequently. Are you arguing that there
should be no peer review, given that it fails on occasion or because
some people are unethical? Indeed, peer review can help catch the case
of intentional or otherwise duplication of earlier work.
The problem is that a patent provides a presumption of validity. Thus,
even if the patent ignores prior non-patent art, the creator of that
prior (published) art has to spend time and money defending himself in
court. Ultimately, that favors the big corporation with a standing army
of lawyers over any small or not-for-profit outfit. I'm not worried
about the stupid ideas (yes, there's a patent on including a list of
destination network addresses in a packet header to do multicast), but
rather the previously published and slightly non-obvious results.
Henning Schulzrinne http://www.cs.columbia.edu/~hgs