#AcademicRunPlaylist - 10/18/24

It was a nearly perfect fall day in Boston, and luckily I was able to go out for a decent run while listening to talks for my #AcademicRunPlaylist!

First was an intriguing talk by Vatsal Sharan on using algorithms to understand transformers (and vice versa) at the Simons Institute for the Theory of Computing https://www.youtube.com/watch?v=qHWBk5ewYwI

Next was a nice talk by Robert Lawrence on why industrial policies to revive manufacturing employment are likely to fall flat at the Peterson Institute for International Economics https://www.youtube.com/live/LOgxdMgqjZY?si=Kqw4Fm2hCCCOqoIB&t=485

Next was a fascinating talk by Jon Kleinberg on proving that an algorithm can generate valid strings from an unknown language given a finite training set at the Simons Institute. This is interesting because it has already been proven that algorithms cannot provably learn these languages in general, but Kleinberg shows that generation is a different beast through a delightful proof https://www.youtube.com/watch?v=BrdaVSZBuyU

Next was an excellent talk by Holger Spamann on the role of indirect investor protections in protecting portfolio investors in the US at the QMUL School of Law. Unsurprisingly, Elon Musk comes up a lot... https://www.youtube.com/watch?v=H_Rsvol6mR0

Next was a compelling talk by Sewon Min on the computational and performance benefits of retrieval-based language models at the Simons Institute https://www.youtube.com/watch?v=YGWkSX5zzDQ

Next was "The Victorian Internet" by Tom Standage. Technology interacts with society in unpredictable ways, and arguably one of the first instantiations of a technological revolution that upended norms, business models, and the law was the telegraph. As Standage shows in this masterful book, the advent of the telegraph spawned changes that are shockingly similar to those we've observed with the internet and other modern technologies - utopian prognostications about its effects, new forms of military action, platform business models, network effects, disinformation, and more.

Our distance from the time of the telegraph makes it particularly instructive, and one that economists and technologists alike would be wise to reflect on when pontificating on the unique nature or general applications of a particular new innovation. Highly recommend https://www.bloomsbury.com/us/victorian-internet-9781635573961/

Last was "Streets of Gold" by Ran Abramitzky and Leah Boustan. In this book, Abramitzky and Boustan systematically exorcise most myths and complaints about immigration in the US - that modern immigrants don't integrate into US society like previous waves of immigrants, that immigrants take jobs from native born US workers, that immigrants and their descendants are fiscal drains on the government, and so on.

There's a lot of time spent on the children of immigrants, as they show that this group has very positive economic outcomes relative to their parents. These sections are insightful, and the hypotheses posed about the causes of this success - namely that immigrants tend to take jobs below their true skill level and that they tend to move to economically growing regions - is convincing.

Some major omissions here, however, prevent this book from being a complete home run. Native Americans and colonialism in North America, for example, is literally never mentioned or examined. Most analyses here compare immigrants to white Americans, explicitly because Black Americans have poor economic mobility. This coupled with the nearly complete avoidance of any discussion of the effects of racism (there's only one section that spends time on this), is especially galling. While a consideration of these issues would frankly only make the conclusions of the book stronger, they stain an otherwise solid contribution to the scholarly literature.

https://www.hachettebookgroup.com/titles/ran-abramitzky/streets-of-gold/9781541797826/?lens=publicaffairs