Interesting perspective on preferring structural (“generative”) over reduced-form (“discriminative”) models:
The mindset for many in the “data science” scene, especially in finance, is “how can I use machine learning tools to discover structure in my data?” I’d caution against this approach, advocating instead a mindset of “what structure might give rise to the data that I observe?”
And related to this by Scott Alexander on automation and artificial intelligence:
One theme that kept coming up was that most modern machine learning algorithms aren’t “transparent” – they can’t give reasons for their choices, and it’s difficult for humans to read them off of the connection weights that form their “brains”. This becomes especially awkward if you’re using the AI for something important. Imagine a future inmate asking why he was denied parole, and the answer being “nobody knows and it’s impossible to find out even in principle”.
Was a mortality rate of 3 percent per month for German fighter pilots in WW2 a lot? This is from Ferguson’s Kissinger biography (emphasis added):
By the end, the war had cost the lives of at least 5.2 million German servicemen—nearly three in every ten men mobilized—and more than 2.4 million German civilians. Total mortality approached 10 percent of Germany’s prewar population. To a remarkable extent, these casualties were inflicted in the final year of the war. More German soldiers lost their lives in the last twelve months of fighting than in the whole of the rest of the war.
Given this extreme malleability of the R runtime it is a legitimate question: “why R hasn’t fractured into a million incompatible domain specific languages and died?”
From Timothy Taylor’s recommendations in the new issue of the JEP, “Seven Facts on Noncognitive Skills from Education to the Labor Market”.
The “British Newspaper Archive Blog”:
My Grandma died when I was young, but in later years my Grandad told me that her parents, my great-grandparents, had gone to Canada and had promised to call for their daughters when they were settled. They never did.
afinetheorem (Kevin Bryan) breaks 7 years of “economic research only”