Lost Worlds using actual geography
A convenient plateau playing the role of refugium for an implausible assortment of otherwise extinct species won't work but I think I have read someone else's idea that does and forgot about reading it. It almost certainly derails human history, though.



The change is that instead of Australia, Antarctica and South America being marsupial heaven [1], the tri-continent area is inhabited by a surviving population of archosaurs and not just birds and crocodilians. Obviously this would be a small subset of the archosaurs that were around before 65 million BP but 65 million years is a long enough time for lots of interesting radiation and specialization. Before the Great American Interchange, the archosaurs in South America would develop in splendid isolation and the Wallace Gap would allow the archosaurs in Australia to do the same. The ones in Antarctica would be screwed for the most part once it got cold down there.

While it would be interesting to think about humans encountering what I will call for the sake of convenience honkers, once the Great American Interchange gets going, species that spread to North America may then spread to Asia and ever farther. It would be pretty easy for this to prevent humans from ever evolving and very difficult to get anything resembling our history. No Linnaeus expounding on the similarities between 12-ton neo-dinosaurian bipedal apex predators and the common chicken, sad to say.

I seem to recall Harrison used the exact opposite of this: hominids evolved in the New World for some reason while dinosaurs kept the Old.

1: Yeah, placental mammals were there too.