What can be done to reduce self-censorship in socially moderated media?

Forbes recently had an interesting article
about the failures of eBay’s feedback system. Many socially-moderated
systems like eBay’s have a significant flaw: users are afraid of
stating their true opinions of other users in feedback comments, as
they expect the other user will respond with retaliatory negative
feedback. The end result is the majority of users end up having near
perfect ratings, even if they do have a very tainted auction history,
as the “social norm” becomes maintaining the status quo.

A related problem occurs on a macro scale at many socially-organized news/discussion sites (what marketers call Web 2.0 applications).
If a user submits a news story which some members disagree with, that
news story is voted down. After the article submitter sees this occur,
they are less likely to continue submitting news stories which are
against the grain, as “what’s the use, it will just get voted down
again”.

A “social norm” is therefore unintentionally established on the
website which gets re-enforced over time. The enforcement of this
social norm soon results in the polarization of the community,
ultimately leading once “neutral” sites to become clearly left vs right
politically, capitalistic vs socialistic economically, or most commonly
pro-Apple vs everyone else (I’m a part-time Apple user btw).

Whether or not this polarization is actually a bad thing is another
discussion onto itself. The greater question is, if you own and operate
a socially-organized site, are you willing to accept the possible
alienation of some of your users?

Common sense would dictate that you should try to avoid alienating
your customers at all times. So what can be done to prevent social
norms from developing on your social media / web 2.0 / interactive
website?