There’s likely some cases of convergent evolution, but I’m not sure this is settled for all trees.

I’d suspect that at least some trees with last common ancestors that are shrubs have re-enabled genes that enable the tree phenotype rather than independently evolving the tree phenotype.

But still really cool (and maybe turns on exactly how much evolution your consider needs to happen before it’s convergent evolution.)

First off, thanks for the discussion, and this is an important question to ask as Democrats unpack why they lost. It boils down to “does moving to the central position gain more voters in the end or not?”

The argument I’m making applies in reverse too. What if there was polling data that indicated that this issue shifted the seven percent who were las likely to vote for Harris into not voting for Harris or not voting at all?

If that’s the case, then their decision becomes more understandable.

The challenge with this poll question is that it doesn’t ask whether this issue changes a potential voter from someone who wouldn’t have voted for Harris into someone who would have voted for Harris. It asks if they are more likely to vote for Harris.

For example, I was already highly likely to vote for Harris, but her being more emphatically against the genocide would still have made me even more likely to vote for her.

To make the case that she should have used this poll to change her position, you have to look at the pre-existing likelihood that someone would vote for her and see whether this issue brought them over that threshold. (For example, what fraction of the 35% voted in the primary and the midterm election? Were they already planning on voting? Who were they planning on voting for if not Harris?)

Bonus points if your static site sends a 503 with a retry after header.

Exactly.

The general approach is to use interpretable models where you can understand how the model works and what features it uses to discriminate, but that doesn’t work for all ML approaches (and even when it does our understanding is incomplete.)

Maybe not the hardest, but still challenging. Unknown biases in training data are a challenge in any experimental design. Opaque ML frequently makes them more challenging to discover.