One interesting item from recent Y-DNA tests is data on another cousin from the lineage of 'your' David's son & dau-in-law wf Sarah Elizabeth Mobley. I expected a perfect match with his 2nd cousin once removed, per the paper trail. But didn't get it. Still in the proper group, but whereas I was hoping for a one or two-marker branch indicator, nothing at all showed up different from the group profile. Of course, the two tests I was comparing had been made several years apart by two different test companies using two different standards and a different set of markers that wasn't a perfect match, so maybe they might have been a closer match than I saw. I don't really know without throwing a lot more money and time at it.
All of which goes to suggest that administrators of surname projects probably need to pay very, very close attention to (and track) items that those who avail themselves of the testing process don't often need (or have the ability and/or money) to find out and track. It also could have been that one test service did not test one of the markers that might be a potential branch indicator and that the ones they did test might have indicated a more recent a change in only one of the cousin's lines. Without more testing and money poured into it, probably not very likely now to find out, one test service having now sold out to the other. Re-testing or expanding the number of markers tested gets as expensive as a new test on a newly discovered candidate, considering the current costs and economic conditions.