In my opinion it was not an unrealistic assumption.
You make an unrealistic assumption that only experienced QA testers would vote
for packages. But anybody can post a link on IRC, web or wherever and ask people
to vote. How can you guarantee that this did not already happen for other packages?
I think the solution with the least effort would be to add a QA checklist link next to
the votes and ask people to follow it for voting.

I also believe that is a bad a idea that positive votes can outvote bugs/problems.
If there really is a problem with package, then the maintainer should fix and promote it to testing again.

Your negative vote is about a flaw in the maemo.org interface and not about a problem
with the package. For this particular package, testing for function is probably the only
thing that needs to be tested as it is otherwise the same as the Nokia PR package.