I'm very glad that the screener(s) is (are) not bound by majority rule. As Skip and others have mentioned, perhaps only one voter sees something incorrect (or in the case of cover scans, askew), and by that lone negative the screener sees the problem and rejects the submission. By properly stating the objection when voting no, the voter gives the submitter real information to resubmit the contribution better than before.

Alternatively, a submission which has more negative votes than positive, while the screener accepts the submission, means the screener finds the objections overcome by the positive aspects of the contribution.

This is a good system. We submit stuff, trying to follow the rules and make the database better. Then we look at these submissions, and express our opinions as to their value. The screener looks at both the submission and the votes (and, very importantly, the comments on both yes and no votes) and judges whether to release the contribution or reject it.

Even if something incorrect accidentally gets released as part of a database update, we can always go back and correct it -- creating a half ping pong effect (a ping?) to compensate for any overcorrections and return the data to correctness.

The safeguards will win out eventually. The process leads the data along the path toward perfection.