There are a lot of promises made about RCV, but there are many reasons to think it (a) does next to nothing to improve representation in government, and (b) the drawbacks outweigh any nominal gain in accuracy.
I’d start with the fact that even FairVote recognizes RCV elections agree with our current system at least 93% of the time. I say “at least” because strategic voting in our current system has similarities with the instant runoff model: Whether it is the voter or the IRV system, support for less viable candidates is moved to more viable candidates.
Second, simulations of RCV and FPTP suggest they share a similar ideological distribution of winners. This, in combination with the above real-world data, should call into question the efficacy of RCV as a proposed solution… if it is even worth the effort.
Lastly, there are much better alternatives we should be considering, such as Approval voting or STAR voting. One of the problems RCV shares with our current system is an electability bias, which tends to correlate with the amount of money a campaign can get. A recent post-election survey in Chicago suggests that candidates with large war-chests perform disproportionately well under both FPTP and RCV. On the other hand, Approval and STAR practically eliminated the correlation.
You mentioned you want systems which create consensus and pull the country back from extreme division. This is what Approval voting and STAR voting are designed to do. They should be center-stage in this discussion. Not RCV.