In part because easily understandable methods for reporting very small probabilities are not in place, journalists naturally err on the side of portraying science as a matter of consensus rather than as a matter of nature observed. As long as we treat readers as "consumers of the consensus" (rather than as independent observers), we'll be doing more than overlooking the mission all of us have in observing the world around. For example, impressions of science as consensus open it up: (i) to the objections of postmodernists as yet another form of idea hegemony, and (ii) to religious enthusiasts as a place to find air time for their consensus too. As Martin Gardner and Michael Crichton (among many) have pointed out, the record of scientific consensus even as an indicator of developing scientific pathways has been far from stellar.
If on the other hand we get in the habit of teaching science not as "how to describe the world" but as "how to observe the world", e.g. through inquiry, discovery (e.g. Piaget, modeling workshop) and empirical observation exercises, not only does pedagogy improve but the foregoing objections vanish. Companies can confidently hire all the scientific observers trained in intelligent design that they need, and philosophers can rail against the scientific elite and not come close to having the clout of the elite's real nemesis: the amazing world around us.
I should add that clear distinctions between science (as paradigms tuned to our interface with the extra-cultural world around) and religion (as beliefs impacting one's behavior within culture) or politics (as idea-sets guiding interaction between families) flow naturally from integrated views of correlation-based complexity. The latter's Bayesian roots can also help journalists to better communicate the very small probabilities with which we began this discussion.