What do you think?
The Irrational Decision: How We Gave Computers the Power to Choose for Us
How the computer revolution shaped our conception of rationality—and why human problems require solutions rooted in human intuition, morality, and judgment
In the 1940s, mathematicians set out to design computers that could act as ideal rational agents in the face of uncertainty. The Irrational Decision tells the story of how they settled on a peculiar mathematical definition of rationality in which every decision is a statistical question of risk. Benjamin Recht traces how this quantitative standard came to define our understanding of rationality, looking at the history of optimization, game theory, statistical testing, and machine learning. He explains why, now more than ever, we need to resist efforts by powerful tech interests to drive public policy and essentially rule our lives.
While mathematical rationality has proven valuable in accelerating computers, regulating pharmaceuticals, and deploying electronic commerce, it fails to solve messy human problems and has given rise to a view of a rational world that is not only overquantified but surprisingly limited. Recht shows how these mathematical methods emerged from wartime research and influenced fields ranging from economics to health care, drawing on illuminating examples ranging from diet planning to chess to self-driving cars.
Highlighting both the power and limitations of mathematical rationality, The Irrational Decision reveals why only humans can resolve fundamentally political or value-based questions and proposes a more expansive approach to decision making that is appropriately supported by computational tools yet firmly rooted in human intuition, morality, and judgment.
In the 1940s, mathematicians set out to design computers that could act as ideal rational agents in the face of uncertainty. The Irrational Decision tells the story of how they settled on a peculiar mathematical definition of rationality in which every decision is a statistical question of risk. Benjamin Recht traces how this quantitative standard came to define our understanding of rationality, looking at the history of optimization, game theory, statistical testing, and machine learning. He explains why, now more than ever, we need to resist efforts by powerful tech interests to drive public policy and essentially rule our lives.
While mathematical rationality has proven valuable in accelerating computers, regulating pharmaceuticals, and deploying electronic commerce, it fails to solve messy human problems and has given rise to a view of a rational world that is not only overquantified but surprisingly limited. Recht shows how these mathematical methods emerged from wartime research and influenced fields ranging from economics to health care, drawing on illuminating examples ranging from diet planning to chess to self-driving cars.
Highlighting both the power and limitations of mathematical rationality, The Irrational Decision reveals why only humans can resolve fundamentally political or value-based questions and proposes a more expansive approach to decision making that is appropriately supported by computational tools yet firmly rooted in human intuition, morality, and judgment.
280 pages, Hardcover
Published March 10, 2026
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 9 of 9 reviews
June 16, 2026
I have followed Ben‘s work and his blog for a while and generally enjoy reading his pieces, so I had high hopes going into this. The historical treatment was interesting and nicely complemented some other reading I have done lately on the history of computation. I also generally appreciated the effort of trying to show working as well as failure regimes of the methods and technologies that the books covers, and the book reinforced the perspective that I should freshen up my fundamental stats skills. That being said, I still finished the book being a bit unsure what to walk away with, apart from having ~geschärfte Sinne~ about what certain people market as rational (or optimal) solutions for problems that ought not have rational (or optimal) solutions. No revolution, but a reminder to stay skeptical, 3.5
Read
July 30, 2026What are the main developments that contributes to mathematical rationality:
1) Mathematical optimization - given a model of the world, compute the action that will give the best return/outcome, with the underlying assumptions
2) Game theory - rules based means to codify human interactions but limited applicability in terms of using it to understand complex social and economic interactions. Works when it is a prescriptive model of behavior, not a descriptive one. In the context of game playing, it helps humans improve their game playing skills.
3) Randomized experiments - By tabulating the outcome of the various random/experimental actions, one can then estimate which actions yield the best returns on average. Randomized trials establish a set of rules for making hard decisions, help one understand if the measured benefits are justifiable against the costs and harms. It does not promise decisions will always be correct but having a clear set of rules is better than nothing.
4) Statistical prediction - Machine learning is basically statistical pattern recognition wherewith there's very few things we understand of it today but there are plenty of things we can do with it. We're working with that we don't understand in the hope that we can achieve something that is truly mind blowing.
5 Questions to ask to determine whether a pattern classification problem is solvable:
1) Does conventional wisdom say there should be a classification rule that is stable over time?
2) Can we not write a computer program that does it?
3) Can we collect a lot of data?
4) Are the stakes low if you make an error?
5) Do you have access to a ton of computing power?
A minimum cost diet is ridiculous, just so are social networks optimized to maximize advertising revenue.
Cannot quite agree with this statement that was quoted in the book:
"When we face scientifically unanswerable questions, we need to be open about making decisions based on our values, not on our science and technology."
what happens in a democratic society when values can greatly differ across the population and sometimes can even be polarizing, would not the above approach make choosing the path forward even more challenging? Does it mean that all such decisions end up becoming a political decision?
1) Mathematical optimization - given a model of the world, compute the action that will give the best return/outcome, with the underlying assumptions
2) Game theory - rules based means to codify human interactions but limited applicability in terms of using it to understand complex social and economic interactions. Works when it is a prescriptive model of behavior, not a descriptive one. In the context of game playing, it helps humans improve their game playing skills.
3) Randomized experiments - By tabulating the outcome of the various random/experimental actions, one can then estimate which actions yield the best returns on average. Randomized trials establish a set of rules for making hard decisions, help one understand if the measured benefits are justifiable against the costs and harms. It does not promise decisions will always be correct but having a clear set of rules is better than nothing.
4) Statistical prediction - Machine learning is basically statistical pattern recognition wherewith there's very few things we understand of it today but there are plenty of things we can do with it. We're working with that we don't understand in the hope that we can achieve something that is truly mind blowing.
5 Questions to ask to determine whether a pattern classification problem is solvable:
1) Does conventional wisdom say there should be a classification rule that is stable over time?
2) Can we not write a computer program that does it?
3) Can we collect a lot of data?
4) Are the stakes low if you make an error?
5) Do you have access to a ton of computing power?
A minimum cost diet is ridiculous, just so are social networks optimized to maximize advertising revenue.
Cannot quite agree with this statement that was quoted in the book:
"When we face scientifically unanswerable questions, we need to be open about making decisions based on our values, not on our science and technology."
what happens in a democratic society when values can greatly differ across the population and sometimes can even be polarizing, would not the above approach make choosing the path forward even more challenging? Does it mean that all such decisions end up becoming a political decision?
September 13, 2026
Llibre excepcionalment escrit que fa un repàs per la història de la computació i la teoria de decisió en 4 blocs: optimització, teoria de jocs, RCT i reconeixement de patrons. Això permet al lector fer-se la idea de quines situacions són òptimes per aquests mètodes, majoritàriament quan l'escenari està ben caracteritzat, amb normes clares i una evaluacio fàcilment quantificable. I, alhora, quan no es pot confiar-hi. Quan l'escenari i les regles no són compartides per tots els actors i difícilment evaluable sense proxies.
La reflexió sobre la burocràcia a l'últim capítol molt adient. L'al·legat final a no deixar-nos prendre la capacitat de TRIAR (l'acció final) en contrast DECIDIR (procés mental) que sí que poden fer les màquines, molt bo.
La reflexió sobre la burocràcia a l'últim capítol molt adient. L'al·legat final a no deixar-nos prendre la capacitat de TRIAR (l'acció final) en contrast DECIDIR (procés mental) que sí que poden fer les màquines, molt bo.
April 8, 2026
The Irrational Decision: How We Gave Computers the Power to Choose for Us by Benjamin Recht is a powerful and timely read that challenges how we define “rational” in a world increasingly driven by algorithms. Recht does an excellent job unpacking the history behind machine based decision making and exposing its hidden limitations. What makes this book stand out is its balance it doesn’t reject technology but instead urges us to question our blind trust in it. The examples are clear, relevant, and often unsettling, especially when applied to real-world systems like AI and public policy. By the end, you’re left rethinking not just computers, but your own decision-making process.
May 29, 2026
The limits of mathematical rationality
For the last 80 years a mathematical approach to rationality has dominated economics and many other fields. Computers are very good at optimization when problems are well defined, i.e., the boundaries of acceptable solutions are clearly specified and there is a clear definition of what “optimal” means. But many of the problems that face us are ill-defined and the choice among competing definitions is a choice among different value systems. Computers can decide, but they can’t choose. To understand why, read this book.
For the last 80 years a mathematical approach to rationality has dominated economics and many other fields. Computers are very good at optimization when problems are well defined, i.e., the boundaries of acceptable solutions are clearly specified and there is a clear definition of what “optimal” means. But many of the problems that face us are ill-defined and the choice among competing definitions is a choice among different value systems. Computers can decide, but they can’t choose. To understand why, read this book.
June 8, 2026
I expected it to be about the limits of being able to delegate decision making to AI based on either impossibility of transfer of responsibility or fundamental inability to articulate requirements (because they are contradictory or unknowable beforehand). Instead it tried to make a more trivial objection about technical limits. Technical limits are temporary.
Spends almost half the book on science history that is not relevant to the arguments. Better treatments of it available in many other books. Why not just put some in suggested reading and the concentrate on the argumentation?
Spends almost half the book on science history that is not relevant to the arguments. Better treatments of it available in many other books. Why not just put some in suggested reading and the concentrate on the argumentation?
September 22, 2026
Felt a bit misled by the title here. I wanted to learn about how people in power use computers and data to offload responsibility and accountability for the decisions they make. Instead, this book bizarrely delves deep into the history of statistics and mathematical theory. I think this would've been fine if it was just an introductory chapter, but never actually getting around to the human impact of computer "decision making" felt like a miss to me.
April 7, 2026
learned some history I didn't before but seems more about behavior economics than computers
September 10, 2026
Good but definitely could (and should) have been longer
Displaying 1 - 9 of 9 reviews