As Ancestry admits:
Keep in mind--The possible relationship information we show is only as accurate as the member-contributed family tree information found in OneWorldTree. We are unable to verify whether these results are in fact accurate. Of course, that's part of the fun of family history--digging into the research.I won't argue with that. It is fun. When I first discovered that my maternal grandmother allegedly was related to such diverse individuals as Flannery OConnor, Werner Von Braun, George W. Bush, Ray Bradbury, Elvis Presley, Amelia Earhart, Shirley Temple, JP Morgan, and to top it off, a direct descendant of Geoffrey Chaucer, I didn't know quite how to respond. One relative, when I mentioned it, came up with a one-word colorful description of what he thought of it all - bullshit.
I wasn't quite ready to dismiss all of it. I knew her ancestry went back to 18th century New York, and once you're in that early colonial melting pot, a lot of families are intermingling. It's not impossible to imagine a handful of the proposed lineages were actually correct. The question was which ones were true, and which ones were like Thoreau and Catherine?
Since then I have discovered my connection with the Sweezey/Swayze clan is rather well documented. Tonight I returned to OneWorldTree and started going through the 35 famous people they list, and discovered that a couple of the descendancies come through the Sweezeys. I was able to "confirm" the connections at Sweezey.net. Browsing through the site, and seeing notes on various entries, they appear to be concerned with accuracy.
So I feel fairly certain that US President, Benjamin Harrison (through the wife of William Henry Harrison) is on my tree. Fifth cousins, three times removed. Elvis Presley may also be my cousin, but he doesn't appear at Sweezey.net, and the Sweezeys don't appear on the Presley genealogy at Genealogy.com. There is one link in his OneWorldTree ancestral chain that is questionable. A recent unofficial genealogy that purports to go back fifty generations contains the missing link, but it's the only place I've found it (besides OneWorldTree) and it seems of questionable veracity.
So right now, with 35 data points, it seems OWT is somewhere between 2.8% and 5.7% accurate. Maybe over time I'll discover it has a better record.