Library

The history of molecular biology from roughly the 1930s to 1970s. Be warned, this is not a breeze to read, but it is one of the most remarkable science books I ever picked up. Reading this is about as close as you can get to having a front-row seat to how scientists go from ignorance and confusion to insight and unifying theory without being one yourself. I want to do this one justice and write more about it at some point, ping me if you see this after 2026 is over and I still haven’t!

Maybe more meme than book at this point. For the gist, read vgr’s post about the central idea of “legibility”. He hints at parallels in tech (waterfall project planning), but I often find myself thinking about this when someone wants to add 20 strictly-validated form fields across 15 rigid workflow steps in some tool, hoping it will make reality conform to the platonic ideal of the process, but most likely just making people hate their job and put everything important in the unstructured “notes” field instead.

An extremely formal theory of design, the formality of which Alexander himself apparently called misguided later on. Design, according to this, is about eliminating misfit between the form of what you are designing and its context (IIRC he explicitly said you can only solve this via negativa). To eliminate all misfit, you map out which of your design variables interact, break the graph down into densely connected but externally loosely coupled subsets, solve each of these in isolation, then put the design solutions together into a whole. It seems extremely obvious that you are not going to do this set-/graph-theoretically for any real-world design problem, but I find the intuition behind the approach interesting and useful anyway. He also claims that traditional cultures have done this “unselfconsciously” by way of cultural evolution, but “modern” designers are operating “selfconsciously” (more rooted in rational attempts at problem-solving) — I don’t find this part particularly relevant, a traditional culture hasn’t had to optimize global logistics or design a CPU, so it’s only natural that the approaches differ.

Should be read with two things in mind: First, it deliberately focuses on the “systems” part of management (largely to the exclusion of interpersonal aspects). Second, it comes from a particular context that has many properties that won’t apply to most organizations (extremely fast-scaling venture-backed tech startups). With these caveats, I liked it precisely because of how tactical and systems-focused it was. A lot of the material has previously been covered on the author’s blog. Some favorites include the point that teams are an abstraction, and his playbook on tackling migrations, which applies well beyond software.

Highly specific to its US/Silicon Valley context as well as very tactical and concrete, much like the other Larson book I read. Good insight into what staff roles actually look like (I wasn’t aware of the breadth/types before reading this). The “Operating at Staff” chapter had useful and actionable advice for any leadership role that included at least some engineering or even technical product work. I previously encountered similar topics in the context of companies about as far from SV big tech culture as you could get while still touching computers, and I think it would have been a very handy reference there or anywhere in between. Mostly skipped the concrete career advice as it wasn’t immediately relevant to me, so I can’t judge its usefulness.

Less a guide for trading, more a tour of the associated worldview. I’ve encountered much of this already over the years by way of economics, interviews with people with a finance background, and “general online exposure”. This is the most concise summary of this kind of reasoning and the associated mental models I’ve seen so far. Recommended reading, even if you keep your money in a cheap broad index fund, as you probably should and which reading this will make you more likely to do.

This is the canonical book explaining timeless and domain-agnostic principles of good design. Unfortunately it also seems to fail to apply its own insight to writing and is terribly disorganized. Looking past that, I still found it worthwhile as an introduction.

Norman argues against calls for making tools and technology simple. He distinguishes between things being complex and things being complicated. Complexity, in his view, is an inherent feature of many domains and tasks, or simply of reality, and any tool worth anything in dealing with reality has to match that complexity to be useful. Good design, then, does not simplify the inherent complexity of reality (it can’t), it makes it easier for humans to make sense of and engage with it. It’s been a while since I read this and the details are hazy, and the complicated/complex thing is semantic nitpicking, but regardless of the words you choose the distinction feels like a really important one and matters a lot if you want to design anything useful.

Duttweiler was the founder of Switzerland’s largest supermarket chain, which (of course) is organized as a federation of cooperatives. Randomly found this in my university’s free library one day. The writing is so-so, and far from impartial, but I stuck with it and was rewarded. The origin story of Migros, and Duttweiler’s various other ventures, are a pretty inspiring example of entrepreneurial spirit winning against incumbents trying everything to entrench the comfortable-for-them status quo, majorly improving the situation for consumers. I also did not fully realize before how weirdly protectionist and cronyist Switzerland used to be, e.g. enacting a law (German) that prohibited chains from opening new stores or expanding existing ones.

Roth concerns himself mainly with the design of matching markets. Unlike “normal” markets, the focus is on both sides choosing who they can or want to transact with — students and colleges, employees and employers, organ donors and recipients, etc. The book outlines some common failure modes of matching markets: 1. There’s not enough participants on either side (lack of “thickness”). 2. There’s too much evaluation to be done (“congestion”) and search costs go through the roof. 3. Markets “unravel” because decisions get pushed too early, before participants can adequately evaluate. 4. Mechanisms make it unsafe for participants to express their true preferences, so they have to game the market to get a good outcome. His main fixes are centralized clearinghouses which aggregate participants on both sides (thickness) and enforce simultaneous clearing (no unraveling), and making market mechanisms strategyproof so people are incentivized to reveal their true preferences. Importantly, this only works if the market produces stable outcomes that people don’t want to defect from — otherwise, they will circumvent the process. Roth also introduces repugnance as a reason why some markets (e.g. organ donations) can’t use prices as an allocation mechanism.

Sacks has recently been somewhat disgraced for apparently making up a lot of stuff in his case studies. Luckily I mainly knew him as the extremely colorful and unhinged character of this and his other memoir, so I don’t think my view shifted much and I’d probably still recommend reading this if you like the genre.