By this criteria - all papers that extrapolate are BS papers.

There’s a ridiculous number of stars in the sky - no matter how low you put the odds of intelligent life, you will still end up with more than 1 civilized species in the universe.

This is, at least for me, the primary utility of such extrapolations. And eventually - extrapolations will be tested.

Extrapolating from multiple good data points and making up a number based on zero examples are absolutely not the same thing.

Coming up with an estimate, calling it a theory and being clear about the fact that it’s not a sure thing ? That’s perfectly fine.

Setting aside if its perfectly fine or not, its not the same as extrapolating from actual data (and no its not “perfectly fine” to completely make up a number and pretend like its science.)