Just to give some context...
In the context of the Original Post, the proposed merge requests came from "trusted" people" but the source of the proposed merge request should not carry that much weight. It merely reflects someone identified two profiles as duplicates.
The potential problems caused by merging them verse not merging them should be weighed. In this scenario, it appears that the profiles are the same person with no conflicts as to the vital statistics. This disagreement is limited to the biography not containing sufficient informatoin to support the profile data.
When you merge these two, I would include a section "Disputed" that contains the biography section that is in dispute and some explanation as to the nature of that dispute.
I believe "backdooring" is when, instead of merging two profiles, you change the associated profiles so that it reflects what the merge would have reflected. But this process is exactly what the person should have done initially. Instead of creating a duplicate profile, they should have edited the associated profiles referencing the profile that is now in dispute.
Leaving the profiles separate causes problems also. Not merging stops the flow of DNA information from crossing over from one side to the other. DNA may help to resolve some of these conflicts, but you can't compare with someone you don't know exists.
If I also try to add a profile and I find the one not documented, I might for example, connect the associated profiles to the one not well documented, not knowing the other duplicate existed.
The introduction of DNA has made increased the need to merge duplicate records. The profile I found may not have any DNA testers associated with it while the other may have many. I will probably find this out over time, but it would be better if I know about it sooner. I may being doing some needless work during he time in between.
IMO, the most difficult scenario involves disagreements as to parentage. which is why I suggested the earliest profile as a guideline. But this guideline applies to those profiles that have conflicting supporting evidence, which I am guessing is uncommon. Most will probably not have any supporting evidence, or one will have supporting evidence, and the other will not. I would use the first in wins as a guideline for the former, and the best evidence wins for the later.
In my opinion, the reasons to merge carries more weight than not merging.