I’m happy Dwarkesh got to interview Robert Caro for several reasons.
First, I suspect it’ll prompt a few people to actually read Caro’s biographies of Robert Moses and Lyndon Johnson. They’re astonishing works of journalistic art that deserve all the attention they can get. And they’re especially relevant in this historical moment of vast change and unease.
Second, it’s rekindled my excitement for the fifth book in The Years of Lyndon Johnson, which covers Vietnam and LBJ’s presidency. That release will be cause to pause most things in my life and make the time and space to devotionally read.
Third, I co-sign the idea that floated around X (which I now can’t find) that Stripe Press should move heaven and earth to publish the full, one million plus word version of The Power Broker. If anyone has the clout to get the Stripe folks to try and pull that off, it’s Dwarkesh.
Finally, the interview crystallised something I’ve noticed over the past year or so, especially in interviews of the “deep research” genre. I’ll call it performative questioning.
A lot of people have realised that, in some sense, the impact any given interview is likely to have correlates with the quality of the questions asked. In true Goodhart’s law style, that’s now the thing that gets optimised for. Dwarkesh mostly dodges that failure mode, but he gets close.
My suspicion is that people drawn to constant, interrogative reasoning, like Dwarkesh, do most of their thinking with an LLM as a sparring partner. Performative questioning is that pattern carried into a live room: it’s turn-based interaction, with an undercurrent of inquiry one-upmanship there to prove to the interviewee and the audience that the interviewer is sufficiently smart and deep.
On balance, it’s a net gain. It forces prep and pre-thinking that carve through the obvious questions and angles before the conversation even starts. But Caro himself is a reminder of what it can miss. In Working, he describes writing “SU” (shut up) in his notebook when an interviewee goes quiet, so he doesn’t rush in to fill the silence.
Sometimes the best question is no question at all. I just hope we don’t get stuck in a local maximum where everyone is trying to ask the best question and forgets that an interview is more than the sum of its Q and A.