I agree, Doug. But I'm also sensitive to the challenge WikiTree faces. Even as oversimplified as the current "Confirmed with DNA" requirements are, we not infrequently see members expressing difficulty understanding and following them. So it's a rock and a hard place: marketing versus accuracy.
Irrespective of the privacy issue I described, I also strongly feel autosomal triangulations among 7th and 8th cousins as described in this thread are, scientifically, highly improbable with any degree of accuracy.
I've used the terrible analogy before that working with autosomal DNA for genealogy is a bit like medicine. Anybody can disinfect a scraped knee and put a Band-Aid on it. Using our typical DNA reported results out through 2nd cousins--great-grandparent MRCAs--is like that.
By 3rd cousins, approximately 8% to 9% of them will share no meaningful amount of DNA with us so our matching "range" effectively starts at zero. It takes a bit more background in biology and more genealogical research to deal with them to a degree of certainty. Now we're at the level of first-aid where we have a laceration rather than just a scrape.
When we get to 4th cousins, only about 48% of them will share a reasonably detectable amount of DNA with us, and the theoretical average sharing is down to 0.196%, or roughly 13cM. Now we need some fairly extensive first-aid training because we have to splint a broken bone and control any bleeding, and we'll also need an experienced physician to set the break and manage treatment. By 3g-grandparent MRCAs we're already at a level of complexity that really can't be distilled down to a simple, standardized if/then procedure.
By 5th cousins, we're scrubbed, gloved, masked, and in the operating room surgically removing an appendix. No Bactine spray and Band-Aids now. You gotta know what you're doing. Any two 5th cousins will be a detectable match only about 15% of the time. Simple statistics, then, tells us that finding any three 5th cousins who match each other would be a probability of around 0.0225, or 2.3% of the time. That places the odds of it happening at approximately 44-to-1.
But wait; there's more.
For a valid triangulation we need at least three cousins not just sharing DNA: they need to share some of the same DNA from the same MRCA. This means significant, overlapping segments that can be confidently traced back to the MRCA. That changes the game. Instead of a 15% chance of finding two 5th cousins who fit the criteria, we need to halve that to 7.5% because any two children of the MRCA couple (except identical twins) will carry only about 50% of the same parental DNA. That drops our probability of finding an accurate triangulation for three cousins to 0.005625, or about 0.57%.
Because each level of full cousinship decreases the expected amount of shared DNA by a multiple of 0.25--e.g., 3rd cousins will share only about 1/4 as much DNA as 2nd cousins--we go really quickly from "anybody can apply a Band-Aid" to "we've got a full ER surgical team on this and the prognosis is not good." By 6th cousins--5g-grandparents--the probability of success is already down to 4.41x10-4, or 0.000441, 0.044%...odds of 2,273-to-1. And we haven't even introduced pedigree collapse into the mix.
Things get quite complex quite quickly. There is no way that WikiTree can establish a simple criteria or procedure (and hope to retain a measure of probable accuracy) for autosomal triangulation among distant cousins. One size does not fit all.
I think the best solution, though it would likely be unpopular, would be to restrict the current triangulation instructions to MRCAs no greater than 4g-grandparents. Going back six generations would be permitted, but autosomal "Confirmed with DNA" could no farther.
An alternative would be to increase the minimum shared segment threshold to 15cM, preferably even a bit higher. There has never been anything remotely magical about a 7cM cutoff. I believe the origin of that minimum was a misreading of work done over a decade ago by John Walden where he found that, when using phased results from one parent, a reported 7cM segment survived the phasing 63% of the time. But what's key is that when the phased results from both parents were considered, 7cM segments proved to be false 58% of the time.
We have a large number of "Confirmed with DNA" status markers on WikiTree that rely on autosomal triangulation among distant cousins. My guess is that nearly all of the ones using 6th cousins or greater cannot stand up to close scrutiny. But I also think the ship has already sailed and that it's unlikely we'll see a significant revision to the triangulation policies.