It's about what constitutes knowledge.
In science, the experimenter's lab book contains authentic data, but no knowledge worth having. The experimenter might write up the experiment and publish it, with his conclusions. But original research papers are often flawed. The conclusions turn out not to be fully justified, one way or another. The papers often report the latest original research on health, diet etc, and contradict themselves weekly. Not good Wikipedia material.
Then you get review papers, where somebody does no original research, but they assess previous published work and try to figure out what can really be learned from it all. These are "secondary" sources. There you might find some knowledge.
On Wikipedia, you aren't supposed to publish your own research, or do your own review. You're supposed to base articles on published reviews, or textbooks etc, which are even further downstream. Then you know the claims being made have stood up to examination.
In genealogy, it's easy to use scanned images of registers and go wrong. The registered data might be inaccurate, or you might be looking at the wrong person, or you might be misunderstanding the implications.
What should be better in principle is a compiled pedigree - say a journal article - where somebody has already collected several pieces of interlocking data and put them together into a story that appears to makes sense, though they might have had to put an interpretation on the raw data that isn't the most obvious. Hopefully somebody who knows what they're doing and is honest, but nobody's perfect.
We'd tend to call that a secondary source, but in Wikipedia terms, that would be original research. Those articles are often flawed.
You're on safer ground when the published work has been around a bit and has been re-examined critically. This doesn't mean that a royal descent given by Browning becomes solid when a few people have copied it into their Great Bloggins Family books.
Books like Complete Peerage and the Richardson books are built on top of previous work. They've assessed the value of it, discovered the contradictions between versions, discounted the obvious handwaving. What gets through should have the best chance of being right. Those are the kind of sources that Wikipedia wants you to use.
But of course there's only a small fraction of WikiTree where we can do that. Most of the genealogy hasn't been done yet, not properly, and there isn't a big body of accepted knowledge out there to build an encyclopedia-type resource on. Most of WikiTree is about posting your own original research in the primary data, and would be banned on Wikipedia.