Some Books Were Read: 2024
Here are some notes on a few books I enjoyed in 2024, not neccesarily books published that year.
The Wool Trilogy, Hugh Howey. Re-reading as a side effect of the TV show, Silo. It’s as good as I remember from back in 2018. Dystopian, post-apocalyptic sic-fi is extremely well worn ground and it’s hard to do anything new. But Howey creates engaging characters including his protagonist Jules who to begin with, has no real interest in going beyond the immense yet claustrophic silo and even less interested in its social ranks. Howey makes you feel the claustrophobia and confinement of the place, a physical representation of the oppressive and limited mental life the inhabitants are allowed. The mystery of the outside is revealed with good pacing as Jules is reluctantly dragged towards it. If you like the show, the books are worth it, with one positive difference being more complex and interesting antagonists and characters.
The Mistborn Trilogy, Brandon Sanderson. What happened was a few years back my wife and I were looking for for a fantasy book/series for our daughter, and on examining this white box set, a very nice young man came up to us to tell us how good Mistborn and Brandon Sanderson were, and yes, this is would be both a good series and a good introduction to his work. Then, earlier this year I came across a writer on yt, called Brandon Sanderson, playing Elden Ring’s DLC and thought—I recognize that name. Learning a bit more about him, Sanderson seems to be something of a legend in terms of organization and productivity—he’s certainly an immensely prolific writer and maintains a public presence and a kind of roadmap that seems almost like a full-time gig in itself. He also teaches/taught writing. All quite impressive. And so yeah, why not read that series of his we picked up? And while fantasy tends to be not much my thing, I I did enjoy very much enjoy it. The allomancy magic system is pretty clever and the world of Scadrial felt extremely well thought out. The writing is lean and straightforward, more like action/utilitarian prose, which I liked and the first book at least is quite plot/story driven, again more like an action thriller than fantasy.
Understanding Deep Learning, Simon J.D. Prince . Not a light book by any means, the author describes it as graduate level, but I’d argue it’s more accessible than rightly well-regarded texts that came before it, and acts as an up to date replacement for Deep Learning. This is an extremely well written and well organised text with an excellent companion site [1]. Chapters 2-7, coming in at around 100 pages, is the best introduction to neural networks in print that I have read. The math is ramped and presented alongside visual representations, very welcome for non-math types like myself. Definitely rewarding to anyone willing to spend time working through it—the 'toy example' of back propagation in chapter 7 for example is worth time spent with a pen and paper as to how these things learn. Chapter 12 covers transformers—if you've read What Is ChatGPT Doing … and Why Does It Work? , and wanted to dig a bit more, this is a good next thing to read. In my Goodreads activity I marked this as being read from Jan 1 to Dec 31st. That’s because it’s one of those you can read but maybe never really finish. In that sense it reminds of work like Structure and Interpretation of Computer Programs. I’ll be spending a good amount of time with it in 2025 as well.
Foundations of Deep Reinforcement Learning, Laura Graesser, Wah Loon Keng. This covers a lot of ground. Like Understanding Deep Learning has for deep learning, it has become the default introduction for reinforcement learning for me, replacing Sutton & Barto , and its mix of code, algorithms and math is very welcome. Why pick reinforcement learning for a book? It's important to real world LLM usage ,via techniques such as reinforcement learning from human feedback (RLHF). As well that, reinforcement learning has a pretty broad range including experimentation (bandits are a simple/special-case form of reinforcement learning, kind of, and a general form of A/B testing, kind of). In that sense you get a lot of functional leverage as an engineer by having a grasp of the topic. Personally though, I find reinforcement learning fascinating. Even OpenAI, way back, started out with reinforcement learning and released the Gym framework, which is used in this book.
The Nice House on the Lake, Deluxe Edition, James Tynion, Álvaro Martínez Bueno. Walter invites his friends to a lake house, whereupon the world ends, Walter is not what he seems and they can’t leave the house. And then things get even stranger. It’s an odd low key horror story of people living in relative comfort while trapped with each other and the enigmatic Walter, their common friend. The art style is good (including how the panel layouts drive the story), it has a slightly saturated and blocky coloration that loosely reminded me of night scenes in John Wick, Monkey Man, or Bladerunner 2049, and the covers have night time flash fill effect that’s suitably eerie. It’s a slow build with the story chapters jumping around in time and between characters. The book novel comes alternate issue covers, the initial pitch for the series, details on the characters, and concept sketches. I don’t read a huge amount of graphic novels, and don’t keep up with the state of the medium any ore, but I will say this is easily one of the best I’ve ever read. The concept, art, story structure, pacing, and writing are all seriously impressive. A second cycle, The Nice House by the Sea, started in 2024, and I’m looking forward to picking up as a deluxe edition as well eventually. But I might not be able to wait that long—those covers, my goodness.
The Dirty South, John Connolly. Number 18 (18!) in the Charlie Parker series. This one is a time shift, set not long after the events in book one and sees Parker investigating a brutal string of murders in a dying Arkansas town called Cargill. The antagonists and locals are particularly well written, maybe the strongest character set in the series: Evander Griffin, the local police chief could carry his own book, a potential teacher and guide to Parker as they reluctantly partner to investigate the cases. At time it feels like a modern western, when a man walks into town, as much as a crime thriller. The local politics surrounding the deal the town needs to survive is well realized and Cargill feels like a real place. We are back to the younger, broken and menacing character of the early novels dealing with varying degrees of compromised, bad and useless people. It’s a welcome and needed reset, the recent books while good (and good by John Connolly means excellent by any standard), were leaning heavy on the supernatural.
The Secret Hours, Mick Herron. A standalone novel set in the world Slough House, intrigue and mendacity in the secret service and cabinet offices set in the present and in the mid-nineties, after the end of the cold war, highlighting the organizational decrepitude and increasing uselessness of spies and spooks. The vehicle is a for-show investigation into corruption. And it’s just as good an anything in the Slough House series, the only slight caveat being it can be a bit hard to follow the twisty plotlines and the time shifts. Herron is one of, if not the best of, spoken and inner dialog writers in modern fiction—his pen has acid for ink. And so it’s no surprise on reflection that Slow Horses has been a runway hit TV show, given the books have such savage wit.
Stubborn Attachments, Tyler Cowan. I think it’s well worth reading [2] regardless whether you agree or not with Cowan’s worldview and presumptions (I gather not everyone does agree). Growth broadly, is good, as is wealth, gated by certain moral imperatives deemed a common sense. In this book-modulo-treatise, he outlines why that should be and the consequences. It has a useful overview of an idea, a growing one as best as I can tell, that current generations should not overly discount the needs of future generations and Cowan believes we discount the future too much (you see this applied in Kim Stanley Robinson’s The Ministry for the Future). It’s strong on economic doctrine, while light on policy: how much discount, how much distribution, how much wealth and their application are left unanswered, but I think that’s ok for a (reasonably brief) treatise on oughts. Cowan gets bucketed as a liberal but I tend to find him hyper-rational, in the sense of high internal cohesion and willingness to follow things to logical conclusions, and am much less interested in his categorization. It helps that he’s a clear and lively writer, even when the concepts are inherently complicated and abstruse. Pro or con on the ideas, the book if nothing else, will help steelman and clarify your own personal reasoning on growth and wealth accumulation.
[1] UDL, astonishingly, is freely available as a PDF
[2] Stripe Press produce wonderful, beautiful books.