the Economist’s Money Talks has the episode about how the money works for broadway shows: TL;DR: broadway revenue on shows. Each show is like an investment — you put in millions of dollars to put it on stage, with a small probability it becomes a hit and earns huge return. However, post-pandemic the return distribution has shifted a little so that this business model becomes less sustainable.x ...

The night started with Kodály’s Dances of Galánta, then Hadelich joined for Dvořák Violin Concerto, lastly Brahms’s Symphony No. 2. The digital program book which contains full, detailed, amazingly written notes can be found here. My friend says that someone said that unpaid intern hold those pens. I’m not violin nor Brahms expert. But it was definitely a splendid night that even the 48-minute super long Brahms symphony failed to bore me. ...

The Grande Pas Classique is one of the most academic and technically challenging piece among the bellet repertoire. [The choreographer, Victor Gzovsky] was born in Russia and studied ballet in St Petersburg under the famous Mariinsky ballerina Evgenia Sokolova. In 1925, he left for Berlin, but later moved to France and after WWII he even held a post of the ballet master at the Paris Opera. In the 1950s – 1970s, Gsovsky worked in Munich, Dusseldorf and Hamburg, where he staged a number of productions. ...

The U of C Symphony Orchestra is rehearsing Ouvertüre zur Fledermaus for Halloween these days. I play piccolo for this piece. It saves a lot of practice time (piccolo can easily play fast). Though the piece sounds easy, the rubato of the piece takes a lot of rehearsals to align everyone on track. Fledermaus overture is such a ubiquitous piece for all music festivals that it’s yelling for parody. Tom and Jerry has one episode dedicated to it — doesn’t the background look like the Schönbrunn Palace? ...

Here’s my own class note from first two weeks of Professor Kamenica’s Topics in Informational Economics course. Math Prelims: Comparing Two Distributions Second-Order Stochastic Dominance (SOSD) Definition. Given $\mathbf{X}, \mathbf{Y} \in \mathcal{X}^n$, we say that $\mathbf{X}$ second-order stochastically dominates $\mathbf{Y}$, written as $\mathbf{X} \ge_{sosd} \mathbf{Y}$, , if for every increasing and concave function $u : \mathbb{R}^n \to \mathbb{R}$, $$ \int u(x), dF_{\mathbf{X}}(x) \ge \int u(x), dF_{\mathbf{Y}}(x). $$ Mean-Preserving Spread (MPS) Definition. Given $\mathbf{X}, \mathbf{Y} \in \mathcal{X}^n$, we say that $\mathbf{X}$ and $\mathbf{Y}$ are equal in distribution, denoted $\mathbf{X} \overset{d}{=} \mathbf{Y}$, iff. $F_{\mathbf{X}}(\cdot) = F_{\mathbf{Y}}(\cdot)$ for their CDFs. ...

Predicting the next viral song on TikTok is almost as hard, if not even more, compared to interest rates fluctuations. Question: can we treat TikTok viral music as a pure random selection? Regardless of the music-pop mechanism, John Williams would have a seat reserved for him at the TikTok Hall of Fame. One of his composition for starwars is now the constant popular song on TikTok — does the following music ring any bells? ...

Bloomberg has the story The AI Boom Needs a Market for Compute. It claims that still now AI companies who need to buy computing resources have to reach a sales representative: In the clunky current system, the chief operating officer of an AI company calls up Amazon Web Services Inc. or CoreWeave and asks for a price. — Bloomberg And, … The godfather of spectrum auctions Paul Milgrom and his company, Auctionomics Inc., have teamed up with OneChronos Markets, a company that builds smart exchanges, to do for compute what Milgrom did for spectrum. — Bloomberg ...

I’ve just been assigned a new piece to practice as part of the University of Chicago Chamber Music Ensemble — Francis Poulenc’s Sonate pour flûte et piano (FP 164). The piece was premiered by Jean-Pierre Rampal — and by coincidence, I’ve also been reading his biography lately! Rampal had such a wonderful presence — full of humor, tolerance, and resilience. He was both humble and brilliant. Rather than trying to rephrase and risk losing the warmth of his voice, I’ll just share a passage from his book directly: ...

A brilliant blend of economic modeling and reinforcement learning.

At the Conference on Frontiers in Machine Learning and Economics (2025), Professor Tamara Broderick from MIT gave a brilliant keynote, based on the paper An Automatic Finite-Sample Robustness Metric. It’s a metric that measures how sensitive are conclusions to the removal of just a small fraction of data? Motivation Even dropping a vanishingly small proportion of observations can matter. We should worry if removing a fraction $\alpha \in (0,1)$ of the data: ...