None of them were writing to each other. Line up the eulogies and it's the same body in every casket: the legible. Anything you can see clearly enough to measure, a competitor can copy and a model can learn well enough to replace.
This week Sarah Guo gave the survivor a name. The untrainable corner: frontier work whose correctness exists only inside someone's private data, walled off inside a system you have to be allowed into. Her sharpest point is that the real bottleneck is permission, and accountability. A model can be smarter than any person and still has to be let in the door, and someone still has to put their name on what it does.
Here is the part that flips how I read every AI-eats-the-world headline. For that corner, a better model is not a threat. It's a gift. As capability rises, the measurable work commoditizes and falls away, and the value gets pushed up into the shrinking band of work that stays private and accountable. If you're standing in that band, the next frontier model is a sharper tool you point at the workflow you already defend.
I work on this problem at Typeface, so I come at it with a builder's bias. But I've watched it play out firsthand: the deployments that stick are the ones where we did the slow, unglamorous translation and earned the right to sit inside the workflow, regardless of how clever the model underneath.
This week's Foraging maps where these voices agree, where they split on who holds the ground, and the one question none of them answered.
Link in comments.