Best Computer Science Books: 43 Picks Recommended by CEOs

Brian Christian and cognitive scientist Tom Griffiths take the limits of time and memory that computers face and show how computer scientists’ solutions can guide human choices, from picking a spouse to tidying an email inbox. Chris Oliver says it helps apply algorithms to buying a house or a car.

Ian Goodfellow, Yoshua Bengio and Aaron Courville build from linear algebra, probability and numerical computation up to convolutional networks, sequence models and deep generative models. Elon Musk calls it the only comprehensive book on the subject, and Nassim Nicholas Taleb praises how directly it handles the math.

Pedro Domingos explains machine learning through its five schools, from connectionism and Bayesian inference to evolutionary computation, and argues they could combine into a single master learner that finds any knowledge in data. Mark Cuban credits it with convincing him that the time for AI had arrived.

Harold Abelson and Gerald Jay Sussman of MIT teach recursion, abstraction and modularity in Scheme, then build interpreters and compilers on register machines, and the text long served MIT’s introductory course. Patrick Collison lists it on his bookshelf, and Santiago Basulto calls it the place to start with fundamentals.

Sebastian Raschka walks readers through coding a large language model with no existing LLM libraries: first a base model, then a text classifier, then a chatbot that follows instructions, all small enough to run on a laptop. Lex Fridman recommends it to beginners and experts alike.

Thomas M. Cover and Joy A. Thomas teach information theory through mathematics, physics and statistics, covering entropy, data compression, channel capacity, rate distortion and Kolmogorov complexity, with problem sets and historical notes in every chapter. It sits among the books Eric Weinstein shared as ones that mattered to him.

Peter Norvig teaches advanced Common Lisp by rebuilding real artificial intelligence programs as case studies, with chapters on debugging, efficiency and the performance problems of working software. It doubles as a course text on AI programming, and Patrick Collison lists it on his bookshelf.

Scott Aaronson, a theoretical computer scientist, moves from logic and set theory through computability, complexity, quantum computing and cryptography, in an informal book based on his own Waterloo course. Patrick Collison lists it on his bookshelf.

Science journalist Anil Ananthaswamy explains the linear algebra, calculus and other mathematics underneath machine learning, holding that safe use of AI requires understanding that math. Tyler Cowen named it a contender on his best nonfiction of 2024 list, while warning that the math will be too much for many readers.

Bruce Schneier’s reference moves from cryptographic protocols and techniques to the algorithms themselves, explaining how RSA and the Data Encryption Standard work and giving implementation advice and source code in C. Dominic Steil recommends it to young people heading into his field.

Jon Bentley builds each essay around a small, nagging problem from working programmers and walks through clever solutions, mixing practical technique with design principles. James Stanley found many a-ha moments in its focus on well-defined problems solved with as little machine time and memory as possible.

Jean-Philippe Aumasson explains the mathematics of modern encryption, from hash functions and block ciphers to RSA, elliptic curves, TLS and post-quantum methods, and uses code samples to show common implementation mistakes. Louis Nyffenegger has it on his current reading list.

Brian W. Kernighan explains for readers with no technical background how hardware, software, networks and the web work, why programming is difficult, and where the limits of computers lie. Eric Schmidt calls it the clearest and simplest account of how the digital world works.

Trevor Hastie, Robert Tibshirani and Jerome Friedman place machine learning and data mining inside one statistical framework, covering boosting, support vector machines, neural networks, random forests and spectral clustering. Nassim Nicholas Taleb calls it the one complete reference on the plumbing behind machine learning.

Kai-Fu Lee, a former executive at Apple, Microsoft and Google, argues that China has drawn level with America in artificial intelligence and forecasts which jobs AI will hit. It is mainly about economics and geopolitics, and Aidan Connolly says it made him think about the rivalry.

Philosopher Nick Bostrom’s book is mainly about what happens if machines surpass human general intelligence, and how to give such a system goals that benefit people. Sam Altman calls it the best thing he has seen on machine intelligence, and Elon Musk urged great care with AI after reading it.

Walter Isaacson traces the computer and the internet through the people who built them, from Ada Lovelace and Alan Turing to Tim Berners-Lee, asking why some inventors succeeded and how teamwork helped. Santiago Basulto says learning how these hackers shaped the tech world had a deep effect on him.

Tracy Kidder follows Tom West’s team at Data General, mostly recent college graduates, racing to design the Eagle minicomputer in the late 1970s. A story of computer engineering rather than theory, it is on Naval Ravikant’s list, and Mohnish Pabrai calls it required reading for product development teams.

Douglas R. Hofstadter weaves Bach, Escher and Gödel together to explore recursion, isomorphism, self-reference and formal systems, inventing two small programming languages. Mark Manson reads it as a study of how paradoxes can explain consciousness, and Seth Godin urged his readers to pick it up.

Claude Shannon’s life is the subject here: his MIT thesis applying Boolean algebra to circuits, the Bell Labs paper that founded information theory, and his juggling machines and unicycles. Cal Newport read a review copy and liked it, and David Senra recommends it.

Garry Kasparov recounts his 1997 match against IBM’s Deep Blue and uses chess to trace the history of machine intelligence, arguing that people can achieve more with intelligent machines than by fearing them. Max Levchin praises its human view of what AI can and cannot do.

Andreas M. Antonopoulos explains bitcoin’s decentralized network, how a transaction moves through it, and how keys, addresses and wallets work, with code snippets for developers. It is mainly about the currency, and Auston Bunsen learned the parts of a wallet address from its fourth chapter.

Clive Thompson studies programmers as a tribe, profiling engineers behind Facebook, Instagram and Google’s AI efforts and weighing the politics of code. It is about programming culture more than computer science, and Kevin Kelly says it explains what programmers do and why.

Frederick P. Brooks Jr. draws on managing IBM’s System/360 and OS/360 for essays on the people problems of building software, including Brooks’s law about late projects and the second-system effect. Mainly about managing software teams, it is a book Jeff Bezos has recommended.

MIT cosmologist Max Tegmark weighs what artificial intelligence could do to life on Earth, from AI weapons and lost jobs to twelve possible futures, arguing that misaligned goals rather than malice are the danger. It is mainly about consequences, and Barack Obama put it on his best of 2018 list.

Neuroscientist and computer engineer Jeff Hawkins proposes that the brain builds hundreds of thousands of maplike models of the world, and uses the theory to consider the future of artificial intelligence. Bill Gates found its theory of what a true AI might take fascinating.

Wharton professor Kartik Hosanagar, drawing on his own work designing algorithms, explains how they shape what people buy, read and date, why they go rogue and the biases they create. It is mainly about AI in society and proposes an Algorithmic Bill of Rights, which Nandan Nilekani urges acting on.

Los Angeles Times correspondent Michael A. Hiltzik tells how Xerox PARC invented the laser printer, the graphical interface and an early personal computer, then saw its corporate parent turn them down. It is mainly the history of a research lab, and Patrick Collison lists it on his bookshelf.

Douglas R. Hofstadter gathers his early 1980s magazine columns on self-reference, analogy and the prisoner’s dilemma, including three chapters on the Lisp language, all asking how human and machine thinking work. Patrick Collison lists it on his bookshelf.

Computer scientist Seymour Papert argues for teaching children computer literacy with the LOGO language and its Turtle graphics, claiming that habits like debugging can reshape how they learn every subject. It is mainly about education, and Patrick Collison lists it on his bookshelf.

Richard W. Hamming turned a Naval Postgraduate School course into a study of how great ideas arise, returning to Shannon’s information theory, Grace Hopper’s programming and his own error-correcting codes. It is mainly about research method, and Patrick Collison lists it on his bookshelf.

M. Mitchell Waldrop tells how J.C.R. Licklider, from a Pentagon post in 1962, pictured personal, interactive computers and an Intergalactic Network, then steered funding toward that vision. It is a history rather than a technical text, and Patrick Collison lists it on his bookshelf.

David Kahn’s history of codes and ciphers runs from ancient Egypt through war, diplomacy and espionage, and a later edition adds Ultra and computer cryptography. It is mainly a history, and Tyler Cowen says reading it as a boy taught him that problems exist to be solved.

Ray Kurzweil describes the neocortex as a stack of some 300 million pattern recognizers and proposes building artificial intelligence the same way, drawing on techniques from his speech recognition work. It is mainly a theory of mind, and Naveen Jain values what it explains about the brain.

Brian W. Kernighan and Rob Pike explain the Unix philosophy of small programs that combine, working through the shell, grep, sed, AWK, C programs and a yacc parser for a small language. It is mainly a programming manual, and Tim O’Reilly admires its clarity and its tools philosophy.

John Brockman gathers 25 essays answering Norbert Wiener’s warning about machines that learn and decide, from contributors both alarmed and unworried about artificial intelligence. It is mainly about the future of AI, and Gerd Leonhard recommends it as reflections from leading scientific minds.

Paul E. Ceruzzi follows computing from the earliest electronic digital machines to the arrival of the web, organized around four turning points such as the first personal computers and networking after 1985. Joan Boixados recommends it to readers who care about computers and history.

Arvind Narayanan and three coauthors explain how Bitcoin and its block chain function, how secure and anonymous its users are, and how to engineer software for the network, with online programming assignments. It is mainly about digital currency, and Antonio Eram has it on his reading list.

Mainly a handbook on software construction, Steve McConnell’s book covers coding, debugging, detailed design and unit testing, down to naming variables and deciding when a subroutine is worth writing. It is the practical side of the field, and Santiago Basulto pairs it with SICP as a starting point.

Peter Seibel interviews 15 programmers, among them Ken Thompson, Donald Knuth and Brendan Eich, on how they learned, debug, read code and choose languages. It is about the craft of programming, and Santiago Basulto calls it full of inspiring stories.

George B. Dyson traces thinkers from Thomas Hobbes to John von Neumann who anticipated artificial intelligence and artificial life, and argues that programs and global networks now form an evolutionary arena. It is mainly a history of ideas, and Antonio Eram recommends it.

Jeff Hawkins, creator of the PalmPilot, proposes a memory-prediction theory of the cortex, contends that today’s computers are not intelligent, and sketches how intelligent machines could be built. Joan Boixados notes how closely it ties to computer science and artificial intelligence.

Astronomer Cliff Stoll recounts how a 75-cent accounting error at Lawrence Berkeley National Laboratory led him to an intruder exploiting a GNU Emacs flaw, whom he traced to West Germany. James Stanley sees it as the earliest recorded account of a computer abused by a remote intruder.