I view that the phenomena of having a  "consensus"is not science, but rather management.

The Dilbert comics are entirely too close to organizational reality.
Pointy haired bosses,
      i.e. folk who are highly paid to manage things
      where they really don't have a clue,
are entirely too common. 

Unfortunately, having little real knowledge or competence in an area where they need to make a high return/loss decision, they receive conflicting input from the "knowledge folk".  They are forced to assess, not the result/data/science, but the perceived competence of the specific individuals.  Additionally, a PHB needs to have CYA to keep their own, well paid, positions, in the case of failure, yet they need to project that they are "right":

The "knowledge"individuals then, could argue technical merit but not be either effective nor useful.
The result tends towards a "belly-bucking contest"to see who is the "dominant' one and thus the "correct"result.

If one remembers that there are big dollars riding on the result:
    position, reputation, authority, etc,
one is then not surprise with bullying, going with the crowd (more CYA), sleight of hand, etc.
=>"Consensus"is just CYA writ large where other course-correcting feedbacks are suppressed.

The solution?
There really isn't one for human organizations.
Darwinian selection tends to occur but the collateral damage can be huge as the belly-bucking incumbents fight dirty to stay in the money and have little actual risk of personal loss.