Identifying a clear definition for what constitutes a consensus is pretty difficult. Age of the question, competing answers, and large numbers of downvotes can all confuse the issue.
I think its a bit easier to identify when there isn't a clear consensus.
If a meta discussion has no answer with a high (5 or more) positive score, or two contradictory answers that have high (5 or more) positive scores within 5 points of each other, then we should consider it to be an issue for which there is no clear consensus.
The converse doesn't necessarily apply to identifying a consensus, though. For example, if a question is asked, and within one day, a single answer garners a high positive score, I'd hesitate to consider that consensus simply because there hasn't been enough time to see if an opposing view has any support.
Meta doesn't always get the visibility it needs to really establish a consensus. Sometimes people who participate heavily on the site may not pay enough attention to what goes on in meta. A highly upvoted meta answer could indicate that something is clearly allowed, yet members of the community may continue to vote to close. In these cases, those with the privilege of voting to reopen should do so, and also leave a comment pointing users to the appropriate meta discussion. Doing so may result in competing answers, or downvotes on the existing answer, but it will provide a clearer understanding of overall community sentiment.
Note this may change as we grow
agree