First of all, thanks to Michael for beginning an extremely useful discussion. Having sources to identify the parents, spouses, and children and their relationships is at once the gold standard and the minimum expectation for good genealogy.
The discussion has quickly moved to "How do we know we're there?" -- which really has two components -- "How does a reader know"? and "How does a computer know"? Unfortunately, these are two very different questions.
A major reason I prefer WikiTree as a site is that WikiTree has narrative biographies where other sites have "notes". What I want to see on a profile is a narrative biography with inline sources that has enough information about a person that I can get some small understanding of who that person was and what they did as well as what their family looked like. When there is some dispute or controversy, I want the source to contain enough information that I can evaluate its value or reliability compared to some other source. I think the very things that I value most are the things that are most difficult for a computer to evaluate.
Because basically the computer is looking for similarities. If it is programmed to look for xxx and if I have entered xxx twice and xxy once, the computer will respond that I have been correct two times and incorrect once. This is an extremely valuable ability, and the whole database errors effort is based upon it. An enormous number of profiles currently have things missing which ought to be there, and the DB errors project lets us know. The DB errors project can even spot errors in my beloved narratives; I prefer sub-paragraphs, and DB errors, without evaluating the content of my writing, can spot where I mismatched a heading with ==Before and ===After.
What troubles me somewhat about this discussion is the thought that we should curtail the styles of narratives and sources in order to make things easier for a computer to spot what is correct or not correct. A well-written, documented narrative on a profile linked to other profiles is what makes WikiTree stand out from other sites with (1) narratives but no organized system of links, like Wikipedia, and other sites with (2) organized systems of linked names like Geni or My Heritage, where the absence of sources or documentation makes them almost entirely useless from a quality perspective. We need to expand upon WikiTree's unique features, not diminish them!
We may want to focus some of our quality efforts on a subset of profiles, such as pre-1500. The Magna Carta project set a standard for well-researched profiles which didn't qualify for a badge until they had been reviewed by a second person. It would be possible, I'm sure, to add a place for three checkmarks in the Edit frame where a human being could check off whether there are sources for parents, spouses, and children. If you wanted this checked only by a reviewer, the computer could be programmed that the check marks must be added by someone who had not previously made any changes to the profile. This could make sense for pre-1500 profiles which have many people working on them -- and for which sourcing is crucially important, but most of us cannot find or credibly interpret the original Latin-language sources. In theory it would make as much sense for the profile of my grandfather who was born in 1872, but in practice on WikiTree we have much more resistance to inline sourcing for more recent profiles -- witness the G2G conversation where it was proposed that the computer simply check for the presence of a </ref>, the ending of any inline source!