While the topic is a bit divisive, the statistical technique highlighted in this post is really cool:

https://wattsupwiththat.com/2022/...

I didn't read the whole paper but why would you assume that there is always one kink in the data.

The meaning of linear regression is also easy to understand — the number of disasters has increased between 1970 and 2020. What meaningful information does the kink add? Regardless, extrapolating with either model is stupid.

@electrineer I think it's to show potential trends that might go unspotted. The keyword being "potential".