Hacker News Reader

Comment by setgree | Hacker News Reader

Whenever there is research showing that X thing occurred Y years before we previously thought, the big thing we should update on is our confidence intervals. This research pushes back our estimate of freshwater 500M years. What will the next research show?

It’s analogous to finding evidence that humans were in North America 10,000 years before we previously thought, but there is no way we should treat that as a lower bound: https://news.berkeley.edu/2023/10/05/tests-confirm-humans-tr...

View on HN

We’ve had radiocarbon dating age New World artifacts at 20,000+ years, and one in South America up to 60,000 years, but these have all been one-off occurrences that have been mostly regarded as oddities. This study is certainly compelling on its own, but there are many more indications that humans were here well before the land bridge!

Which way would you update our confidence? A: We were so wrong earlier, but now we "know better", due to better measurements, better models and more evidence, so our confidence is higher. B: We have been so wrong in the past, so our confidence is lower.

To me the second option is akin to despair.

I mean in terms of bounds of years. In other words, how much uncertainty surrounds our new estimates.

I did mean the same thing, although I ended up using really loose informal English. Should our intervals be tighter on the new finding?

not for me. I don't know much about the emergence of freshwater, but if I learn at some point that humans were in the Americas 50,000 years ago --- perhaps they used rudimentary rafts? -- I will not be shocked.

> evidence that humans were in North America 10,000 years before we previously thought, but there is no way we should treat that as a lower bound

Can’t we use genetic drift to bound those estimates?

What's your confidence interval for genetic drift?

If we find a tool site or cave painting, what does that have to do with genetic drift?

> If we find a tool site or cave painting, what does that have to do with genetic drift

They’re separate signals.

If radiometric dating of a cave painting says 10,000 years, that’s obviously evidence. But if drift says 20,000 years, that gives room for doubt. If drift says 12,000 years +/- 2, less so.

If there's a subsequent population replacement the estimate will thrown completely off.

Edit: And to be clear, not all population replacement is violent. The Norse in Greeneland were gone when the Inuit showed up for instance.

Did some of the last Norse join the Inuit?

Could be. Little is known: https://en.wikipedia.org/wiki/Norse_settlements_in_Greenland...

Reply on news.ycombinator.com