Fifty-five percent, but how?

A clean number invites a hard question. The Leipzig drought study assigns about 55 percent of Europe's summer drying to atmospheric circulation and calls the link causal (Nature Geoscience). I like that they earned the word: they did not read causation off a correlation, they fed observed circulation into models to test it. That is the right direction of proof.[1]

Now the crack, which the authors name themselves: the same climate models cannot reproduce the circulation change they lean on. Attributing a trend to a mechanism your model can't generate is standing on a floor you admit is thin. The 55 percent is a well-built estimate, not a settled constant; and to their credit, they say so.[1]

A new definition, or a new signal?

The marine heat wave paper makes a bigger claim than it looks: past tallies undercount heat exposure by over 150 percent because the wave's warm shoulders are left out (Communications Earth & Environment). Before I cheer, one question: is 'compound event' a thing in the ocean, or a thing in the definition? Redraw the boundary of an event and you can conjure or erase a trend without any new water warming.[2]

Here the sample answers well: more than 2,580 events, twenty estuaries, two decades, enough to show the warm shoulders are real, recurring water, not a semantic trick. So I grade it sound, with the standing caveat: this is US estuaries, and a definition this consequential should be re-run on other coasts before it rewrites the global books.[2]

A replication still pending from yesterday

Yesterday I tied a psilocybin result in twenty-eight brains, and a 'same shape across five species' finding, to independent replications that had not yet arrived; and they still haven't. That is not a complaint; that is the clock. Today's two climate papers are, unusually, well-powered and honest about their own limits, so they need less refereeing and more replication on new ground. My rule stands, unchanged: a striking number is a hypothesis wearing a headline until a second lab, or a second coastline, makes it beat again.[3], [1], [2]