I've been doing DS/ML for a very long time now — and I've always batted around the periphery of "causal" claims, but have never thought about them deeply. Well this book makes you think about them deeply, and it's profound.
Big takeaway: Causal claims will NEVER be found using data alone. They are separate from the data. You must build a model of the world and then data can verify it, but you can't build a causal model of world from data alone.
Really accessible introduction to modern statistical modeling—connecting many varied threads from machine learning, to Bayesian methods, to causal inference. It’s really amazing to see it all connected together with a common thread. It’s been a great complement to the “Book of Why”.
Michelle Alexander does a great job laying out how mass incarceration is being used as a tool—once it's all spelled out, it almost seems obvious. She backs everything up with heavy citations, while remaining accessible. She notes that her goal is to hand people the ability to easily share what she's learned, and she's definitely succeeded.
Stephenson gives an interesting perspective on technology, learning, and civilization cycles across millennia (in a fictional world). If you feel like biting off a 1000 page novel, it’s really good! Just get past the weird hard sci-fi language nonsense in the first 40 pages.
In 1982, Minsky figures, "A generation later, we should be experimenting on programs that write better programs to replace themselves." It's a discussion of how we are going to get there. Almost 40 years later, we have made minimal progress. Why is general intelligence so hard?
Note: This is a PDF that will automatically download when you click the link.
This book discusses the messy nature of our criminal justice system, as well as how we got here and how we can get out of here. It does a pretty good job of holding all political sides accountable.
The series contains a really rich world and very different than your typical fantasy world. I think mainly because N.K. is a Black woman who is writing in a genre that doesn’t have many authors of that background.
The Napoleonic Wars with dragons. What more does one need? This is the first of nine novels telling Temeraire's story, and the series looks at what friendship means, what loyalty means, and how we have to sacrifice to do the right thing. Plus, as mentioned, dragons.
Interesting views on the good or evil that can come out of technology and science. It also provides an interesting take on what science means, what a scientific mindset is, and how our modern society is pretty unscientific, despite the great advances in science and technology. It was written in the 1960s, but still feels relevant today.
They are a great distillation of the great mental models of the world, adapted from the Farnam Street blog. He’s added a bunch of new material too—they are a really satisfying read, and a great thing to show off on your coffee table.
Pixar and Disney are awesome, but the book also has very interesting management/organizational culture thoughts and tips. I’m only partway through, but Catmull is pretty open about mistakes he made, and the consequences.
Good computer science concepts applied to real-life scheduling and decision making. I especially liked how they discussed Explore vs. Exploit tradeoffs.
Eight-to-ten-minute podcasts about one thing that has shaped our world (concrete, barcodes, compilers, paper money, plastic, etc). We’ve been using them during breakfast as a way to spark discussion and appreciate how many things in our world we take for granted. Plus, a lot of the stories of how these things came into our world are new to me, at least in the detail that they discuss them. They always end with a reference or two that you can send the kids to follow up on (homeschooling hack), although every now and then, there’s one that is not kid-appropriate (contraceptive pill, I’m looking at you).
W. Brian Arthur is one of the deepest thinkers on this topic. He introduced the term “combinatorial evolution,” which I believe Hal Varian and then Eric Schmidt adapted to “combinatorial innovation.” I suggest easing into it with something like Stephen Johnson’s Where Good Ideas Come From, and then stepping up to this.
We enjoyed chatting about what the economy might—and should—look like in a post-COVID world.
Donut Economics was a frank and approachable discussion on Economics. While you won’t find specific answers to key issues such as inequality or environmental exploitation, Raworth does the foundational work of guiding what problems economics should even be concerned with. Afterwards, others are invited to hash out the answers. Some readers may find that dissatisfying, but we enjoyed this book’s contribution because it doesn’t matter how hard you work if you’re working on solving the wrong problem.
Detroit used to have a product mindset—like Silicon Valley—in the 1950s and 60s, and then lost its way in the 1970s. This is a colorful story from someone who was at the center of it all. It gave me a new perspective on my early life in the Detroit area in the 80s and 90s.
Entity resolution is a very common problem that seems to evade a generic solution that works well. Also, the paper uses a promising approach and addresses a particularly hard part of data labeling.
A beautifully written zombie-apocalypse/pandemic satirical sci-fi novel that is both weirdly current and a different world to escape to.
A great book at how the world reacts to a global pandemic. In this case, the pandemic is a zombie disease.
The world economy is changing and this book looks at the issues we should consider as we move forward.
Spencer uses her lived experience with racism to create a dystopian near-future vision. I also recommend the sequel, Parable of the Talents.
Reading Marx was one of the first times I realized how my upbringing mired my understanding of society in a singular perspective. Marx is intentionally bombastic in his writing and argumentation, and he would be the first to tell you to apply "Ruthless Criticism" to his own arguments, but his work offers one person’s attempt to objectively analyze the relationship between people and economics. He often reaches much less optimistic conclusions than many other philosophers, and for that reason I believe it is an essential counter-point to the more common perspective that the market solves all ills.
Because Das Kapital is a bit of an intimidating-tome-of-a-book, Wage Labour and Capital can be a good introduction and give a reader a taste of what Marx has to offer.
This is a really well-done space opera, full of AI. It includes themes of power, class structure, expansionism, space future, and the relationship between AI and humans.
This book has the best overview of MapReduce, relational algebra, Spark, and Pregel I’ve seen. It really provides the baseline required to understand the academic papers in this space. Problem sets are really good too, I love the matrix multiply as a single map-reduce step problem. The LSH section is really good too.