What do you think?
Rate this book
The Model Thinker: What You Need to Know to Make Data Work for You
How anyone can become a data ninja
From the stock market to genomics laboratories, census figures to marketing email blasts, we are awash with data. But as anyone who has ever opened up a spreadsheet packed with seemingly infinite lines of data knows, numbers aren't enough: we need to know how to make those numbers talk. In The Model Thinker, social scientist Scott E. Page shows us the mathematical, statistical, and computational models--from linear regression to random walks and far beyond--that can turn anyone into a genius. At the core of the book is Page's "many-model paradigm," which shows the reader how to apply multiple models to organize the data, leading to wiser choices, more accurate predictions, and more robust designs. The Model Thinker provides a toolkit for business people, students, scientists, pollsters, and bloggers to make them better, clearer thinkers, able to leverage data and information to their advantage.
From the stock market to genomics laboratories, census figures to marketing email blasts, we are awash with data. But as anyone who has ever opened up a spreadsheet packed with seemingly infinite lines of data knows, numbers aren't enough: we need to know how to make those numbers talk. In The Model Thinker, social scientist Scott E. Page shows us the mathematical, statistical, and computational models--from linear regression to random walks and far beyond--that can turn anyone into a genius. At the core of the book is Page's "many-model paradigm," which shows the reader how to apply multiple models to organize the data, leading to wiser choices, more accurate predictions, and more robust designs. The Model Thinker provides a toolkit for business people, students, scientists, pollsters, and bloggers to make them better, clearer thinkers, able to leverage data and information to their advantage.
466 pages, Kindle Edition
First published November 27, 2018
1194 people are currently reading
6467 people want to read
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 30 of 93 reviews
April 18, 2019
As a professional Business Intelligence Analyst (BIA) this is the perfect non-fiction book for my desk. I do data aggregation, reporting, and analysis at my day job. And we are constantly trying to determine what the best correlation, representation or model is for analysis the data available to us.
Scott E. Page starts us off talking about WHY. This is an often overlooked piece of any business work. The why. Why do we do something? Why do we care? Why use X over Y? And so on...
In this case Page is able to eloquently argue that in today's world of "big data" we need to be more aware of what options are available to us for analysis. It's no longer appropriate to use one model to analysis a problem. Instead we need to leverage the multi-model approach and look at complex issues, problems or phenomena (as he calls them) from many angles. As the world has gotten more complex, our data has gotten larger and more granular; which means we need to look at it from many different perspectives.
Seriously Detailed
I'm on a team of BIA's, many more senior than I, and so I chatted with them about the core concepts in Page's book and we all agreed on one thing. It's comprehensive! If there is a major model not represented in The Model Thinker I'd be shocked. Page does a great job of touching on 30+ models and giving three very specific things for each:
1) The definition and general usage
2) The actual mathematical breakdown
3) A real-life, relevant example of using the model.
The best part of reading any portion of Page's epic selection of models is easily the examples. From health care to criminal form to food quality to the stock market to population growth; we are given applicable scenarios to understand the nuances of each model and why it's the best choice. Plus there are tons of little tidbits and facts that are fun to share at cocktail parties (or if you're me, in random elevator conversations) in here!
Creativity
There is one thing that really surprised me in The Model Thinker and that is Page's emphasis on creativity. As someone who has a BA in Communications, Marketing & Design, and used to work in the magazine industry, I was surprised to find that my current Analytical career had such a basis in creativity. Reflecting upon Page's statements and my job I realized that he is right. When we program/code, develop visualizations or infer outcomes from data; we are looking at something and creatively manipulating it. Perhaps this explains how I went from an Art Director to a Analyst in one lifetime.
Overall
This is a book that I will be purchasing for my office shelf. I have already gone back to my eARC copy multiple times to look things up and to continue learning the models. It's not a book you will likely read cover to cover at any given time. But the first 50 pages of introduction and concepts are superb and well worth reading in order. After that you can jump around to the models you are most interested in, or if you're looking for the right solution for data crunching, read the intros to each chapter to determine if there might be applicable use to your situation. I know that The Model Thinker has already been picked up by some mainstream Universities and Colleges as a required textbook and certainly I can see why. In one book you gain the knowledge of hundreds of years worth of calculations and analyzation. Whether you currently work as a Data or Business Analyst, have a desire to learn to use big data, are a programmer or just a geek that loves graphs; The Model Thinker is likely to fill a void, you didn't even know existed, by giving you more models, examples and calculations than you will ever need.
To read this and more of my reviews visit my blog at Epic Reading
Please note: I received an eARC of this book from the publisher via NetGalley. This is an honest and unbiased review.
Scott E. Page starts us off talking about WHY. This is an often overlooked piece of any business work. The why. Why do we do something? Why do we care? Why use X over Y? And so on...
In this case Page is able to eloquently argue that in today's world of "big data" we need to be more aware of what options are available to us for analysis. It's no longer appropriate to use one model to analysis a problem. Instead we need to leverage the multi-model approach and look at complex issues, problems or phenomena (as he calls them) from many angles. As the world has gotten more complex, our data has gotten larger and more granular; which means we need to look at it from many different perspectives.
Seriously Detailed
I'm on a team of BIA's, many more senior than I, and so I chatted with them about the core concepts in Page's book and we all agreed on one thing. It's comprehensive! If there is a major model not represented in The Model Thinker I'd be shocked. Page does a great job of touching on 30+ models and giving three very specific things for each:
1) The definition and general usage
2) The actual mathematical breakdown
3) A real-life, relevant example of using the model.
The best part of reading any portion of Page's epic selection of models is easily the examples. From health care to criminal form to food quality to the stock market to population growth; we are given applicable scenarios to understand the nuances of each model and why it's the best choice. Plus there are tons of little tidbits and facts that are fun to share at cocktail parties (or if you're me, in random elevator conversations) in here!
Creativity
There is one thing that really surprised me in The Model Thinker and that is Page's emphasis on creativity. As someone who has a BA in Communications, Marketing & Design, and used to work in the magazine industry, I was surprised to find that my current Analytical career had such a basis in creativity. Reflecting upon Page's statements and my job I realized that he is right. When we program/code, develop visualizations or infer outcomes from data; we are looking at something and creatively manipulating it. Perhaps this explains how I went from an Art Director to a Analyst in one lifetime.
Overall
This is a book that I will be purchasing for my office shelf. I have already gone back to my eARC copy multiple times to look things up and to continue learning the models. It's not a book you will likely read cover to cover at any given time. But the first 50 pages of introduction and concepts are superb and well worth reading in order. After that you can jump around to the models you are most interested in, or if you're looking for the right solution for data crunching, read the intros to each chapter to determine if there might be applicable use to your situation. I know that The Model Thinker has already been picked up by some mainstream Universities and Colleges as a required textbook and certainly I can see why. In one book you gain the knowledge of hundreds of years worth of calculations and analyzation. Whether you currently work as a Data or Business Analyst, have a desire to learn to use big data, are a programmer or just a geek that loves graphs; The Model Thinker is likely to fill a void, you didn't even know existed, by giving you more models, examples and calculations than you will ever need.
To read this and more of my reviews visit my blog at Epic Reading
Please note: I received an eARC of this book from the publisher via NetGalley. This is an honest and unbiased review.
October 2, 2019
Doesn’t provide as much math as a textbook and doesn’t provide as much substance as a story/novel. Caught in between doing neither well.
October 26, 2018
Professor Page certainly introduces some interesting concepts in The Model Thinker. His overall aim here is to get you to use multiple models in your thinking, and the plethora of models provided will be an aid to any person out there.
That said, I found it rather difficult to get through. My attention wandered far more than I thought it would. I think that is just the nature of non-fiction sometimes. Even when interested in a subject, a book on it is not always enjoyable. Thus, my star rating should be taken to reflect an emotional reaction to this book rather than a reaction to its content or construction.
If you've ever taken one of Professor Page's classes on models and/or model thinking, you'll be right at home in this book.
That said, I found it rather difficult to get through. My attention wandered far more than I thought it would. I think that is just the nature of non-fiction sometimes. Even when interested in a subject, a book on it is not always enjoyable. Thus, my star rating should be taken to reflect an emotional reaction to this book rather than a reaction to its content or construction.
If you've ever taken one of Professor Page's classes on models and/or model thinking, you'll be right at home in this book.
November 25, 2018
I really like the subject of this book. Model thinking is one of the best subjects I have taken in Coursera. The concepts are really useful and practical. It helps you to frame your thinking about numerous things in our world.
I would not give this book a 5 stars because I think this subject is best convey through other format like videos, lectures, or course. The topic is a bit complex especially for those who ate not comfortable with numerical reasoning. By putting it in book format it limited its audience.
Since it is already written as a book, my suggestion would be to have more illustrations instead of text. This would aid in easy understanding.
Aside from illustrations, more practical examples would help.
I would not give this book a 5 stars because I think this subject is best convey through other format like videos, lectures, or course. The topic is a bit complex especially for those who ate not comfortable with numerical reasoning. By putting it in book format it limited its audience.
Since it is already written as a book, my suggestion would be to have more illustrations instead of text. This would aid in easy understanding.
Aside from illustrations, more practical examples would help.
June 21, 2020
Here are my key takeaways from the book. I’ve found that the earlier chapters were more interesting to me as there were more statistical concepts at the start. There are many game theory models in the later chapters:
-Many-model approach: The book relies on a many-model approach. A given problem can usually be solved with several models. Also, different problems require different models. (pp. 5-6) Condorcet jury theorem: Many models are better than one model in a jury problem. If a judge is right more often than not, a decision will be more accurate if there are multiple jurors. (p. 28) Diversity prediction theorem: Many-model error = Average-model error - diversity of model predictions. (pp. 29-30)
-The wisdom hierarchy: Data, information, knowledge, wisdom. Data is just raw data. Information is categorical data. For example, region-labeled temperature data is information. Knowledge is “justified true belief” (Plato) and contains additional relational information such as correlations, causation, logic, and conditions for the models to hold. Knowledge is organized information. Models are knowledge. Wisdom is the “ability to identify and apply relevant knowledge”. (pp. 7-8)
-Models can be divided into embodiment models (such as a discounted cash flow model), analogous models (such as a valuation multiple model), and alternative reality models (such as the game of life; they model hypothetical scenarios and not an actual process). (pp. 13-14)
-The seven uses of models: REDCAPE (p. 15):
--Reason (identify conditions and deduce logical implications);
--explain;
--design processes, institutions, policies, and rules;
--communicate: Clear sets of inputs and outputs and corresponding definitions of all inputs and outputs that can be agreed upon. (pp. 20-21)
--act;
--predict;
--explore.
-Bagging means “bootstrap aggregation” and constructs many models. Many datasets are drawn and models are fitted. The average of the models is the final model. (p. 42)
-Rule-based models work without loss functions. (p. 53)
-Models produce an equilibrium, cycles, randomness, or complexity. (p. 56)
-Rational-choice models and zero-intelligence models provide upper and lower bounds for potential outcomes. (p. 58)
-Multiplication of random variables leads to lognormal distributions. Addition of random variables leads to normal distributions. (p. 60) “Lognormal distributions lack symmetry because products of numbers larger than 1 grow faster than sums ... and multiples of numbers less than 1 decrease faster than .. sums. If we multiply sets of twenty random variables with values uniformly distributed between zero and 10, their product will consist of many outcomes near zero and some large outcomes, creating the skewed distribution shown in 5.2.” (p. 66) Long-tailed distributions require non-independence. Those often come in the form of positive feedback. Sales, fires, city populations. (p. 69) Power-law distributions have many small events. (p. 70) Power laws with an exponent of 2 or less lack a well-defined mean. (p. 71)
-Sampling random variables: \sigma_\mu = \frac{\sigma}{\sqrt{N}} and \sigma_{\sum} = \sigma\sqrt{N} (p. 63)
-Zipf’s law: “The special case of power laws with exponents equal to 2 are known as Zipf distributions. For power laws with exponents of two, an event’s rank times its probability will equal a constant, a regularity known as Zipf’s law. Words satisfy Zipf’s law. The most common English word, the, occurs 7% of the time. The second most common word, of, occurs 3.5% of the time. Notice that its rank, 2, times its frequency of 3.5% equals 7%.” Event Rank * Event Size = Constant (p. 72)
-Self-organized criticality, forest fire model: Self-organized criticality leads to power-law distributions for the sizes of forest fires. “The key assumptions for self-organization to critical states is that pressure increases smoothly, like water flowing into the lake, and that pressure decreases in bursts, including possibly large events. (pp. 74-75)
-Projects with large budgets over run out of control. This is because the random variables, i.e., the budgets for individual stages and parts of a project, are positively dependent on each other. If one part encounters problems, the next step will also run into budgetary problems. (pp. 78-79)
-Models of value and power: LOTB and Shapley values. LOTB is the last-on-the-bus value. How much would a participant add to the system if he joined last? Shapley values are the average value of a participant for each possible condition under which the participant could enter the system. (p. 107)
-In a network model, a “node’s betweenness score equals the percentage of minimal paths that go through a node.” (p. 119)
-Friendship paradox: “One analysis of friendships on Facebook found that the average person has around two hundred friends and their friends, on average, have more than six hundred friends.” (p. 124)
-Network robustness. “Here, we consider the question of how the size of the largest connected component of the network, the giant component, changes as nodes randomly fail… In the random network, the size of the giant component falls linearly at first. At a critical value where the probability of an edge equals 1 divided by the number of nodes, the size of the largest component falls to an arbitrarily small proportion of the original network size. The small-world network shows no such abrupt change. A majority of t connections exist within the geographic clusters.” (p. 127)
-”Most consumer goods and information spread through both diffusion and broadcast. Our next model, the bass model, combines the two processes in a single model.” (p. 136)
-SIR model: The susceptible-infected-recovered model has a built-in saturation assumption. (p. 137)
-The minimum percentage of immunity in a population, the vaccination threshold, depends on the basic reproduction number (R_0), which says how many people get infected by each person that is newly infected. The vaccination threshold is: V \geq \frac{R_0 - 1}{R_0}. Polio has a R_0 of 6, so the vaccine must cover ⅚ of the population.(p. 138-139)
-Superspreaders are key to epidemics. This is due to degree squaring. Superspreaders have more contacts to others, which exposes them to more risk for getting a disease and also increases the chances to spread the disease in the same manner. A superspreader with 3x more social contact will therefore be 3x more likely and 3x more dangerous when infected, making him 9x more dangerous overall. (pp. 139-140).
-R_0 and the vaccination threshold are contextual tipping points. Small changes in the context (environment) can change outcomes significantly. The current focus in public discussions around R_0 and vaccinations show that there is awareness for the criticality of these measures. (p. 141)
-Grossman and Stiglitz paradox: If investors believe in the efficient market hypothesis, they stop analyzing, making markets inefficient. (p. 160)
-Agent-based models are computer programs that model each agent individually. (p. 213)
-Hedonic attributes: more is always better. Spatial attributes: sweet spots that are different among agents. (pp. 227-228)
-Voronoi neighborhoods help to see which characteristics a product should have to optimally fit a certain customer group. (p. 231)
-”A one-dimensional model implies that candidates position at the median. Higher-dimensional models imply that they should not. Which type of model do we believe? We should place complete faith in neither model, but instead gain insights from both… We should instead expect complexity, an endless dance of competition for votes through coalition building.” (p. 234)
-Models of cooperation: In this example: a lone cooperator cannot spawn an additional cooperator, but two adjacent cooperators can. It follows that a small cluster of cooperative nodes surrounded by empty cells could expand into open nodes. Therefore, regions of cooperation can emerge from a handful of cooperators. (p. 264) Group selection is another way to achieve efficient cooperation. Let groups form. The best group (the one with many cooperators) will emerge as a winner and make copies of itself or otherwise project its system on the rest of the population. This can be easier than to just have one large population that has to be made cooperative. (p. 265)
-Many-model approach: The book relies on a many-model approach. A given problem can usually be solved with several models. Also, different problems require different models. (pp. 5-6) Condorcet jury theorem: Many models are better than one model in a jury problem. If a judge is right more often than not, a decision will be more accurate if there are multiple jurors. (p. 28) Diversity prediction theorem: Many-model error = Average-model error - diversity of model predictions. (pp. 29-30)
-The wisdom hierarchy: Data, information, knowledge, wisdom. Data is just raw data. Information is categorical data. For example, region-labeled temperature data is information. Knowledge is “justified true belief” (Plato) and contains additional relational information such as correlations, causation, logic, and conditions for the models to hold. Knowledge is organized information. Models are knowledge. Wisdom is the “ability to identify and apply relevant knowledge”. (pp. 7-8)
-Models can be divided into embodiment models (such as a discounted cash flow model), analogous models (such as a valuation multiple model), and alternative reality models (such as the game of life; they model hypothetical scenarios and not an actual process). (pp. 13-14)
-The seven uses of models: REDCAPE (p. 15):
--Reason (identify conditions and deduce logical implications);
--explain;
--design processes, institutions, policies, and rules;
--communicate: Clear sets of inputs and outputs and corresponding definitions of all inputs and outputs that can be agreed upon. (pp. 20-21)
--act;
--predict;
--explore.
-Bagging means “bootstrap aggregation” and constructs many models. Many datasets are drawn and models are fitted. The average of the models is the final model. (p. 42)
-Rule-based models work without loss functions. (p. 53)
-Models produce an equilibrium, cycles, randomness, or complexity. (p. 56)
-Rational-choice models and zero-intelligence models provide upper and lower bounds for potential outcomes. (p. 58)
-Multiplication of random variables leads to lognormal distributions. Addition of random variables leads to normal distributions. (p. 60) “Lognormal distributions lack symmetry because products of numbers larger than 1 grow faster than sums ... and multiples of numbers less than 1 decrease faster than .. sums. If we multiply sets of twenty random variables with values uniformly distributed between zero and 10, their product will consist of many outcomes near zero and some large outcomes, creating the skewed distribution shown in 5.2.” (p. 66) Long-tailed distributions require non-independence. Those often come in the form of positive feedback. Sales, fires, city populations. (p. 69) Power-law distributions have many small events. (p. 70) Power laws with an exponent of 2 or less lack a well-defined mean. (p. 71)
-Sampling random variables: \sigma_\mu = \frac{\sigma}{\sqrt{N}} and \sigma_{\sum} = \sigma\sqrt{N} (p. 63)
-Zipf’s law: “The special case of power laws with exponents equal to 2 are known as Zipf distributions. For power laws with exponents of two, an event’s rank times its probability will equal a constant, a regularity known as Zipf’s law. Words satisfy Zipf’s law. The most common English word, the, occurs 7% of the time. The second most common word, of, occurs 3.5% of the time. Notice that its rank, 2, times its frequency of 3.5% equals 7%.” Event Rank * Event Size = Constant (p. 72)
-Self-organized criticality, forest fire model: Self-organized criticality leads to power-law distributions for the sizes of forest fires. “The key assumptions for self-organization to critical states is that pressure increases smoothly, like water flowing into the lake, and that pressure decreases in bursts, including possibly large events. (pp. 74-75)
-Projects with large budgets over run out of control. This is because the random variables, i.e., the budgets for individual stages and parts of a project, are positively dependent on each other. If one part encounters problems, the next step will also run into budgetary problems. (pp. 78-79)
-Models of value and power: LOTB and Shapley values. LOTB is the last-on-the-bus value. How much would a participant add to the system if he joined last? Shapley values are the average value of a participant for each possible condition under which the participant could enter the system. (p. 107)
-In a network model, a “node’s betweenness score equals the percentage of minimal paths that go through a node.” (p. 119)
-Friendship paradox: “One analysis of friendships on Facebook found that the average person has around two hundred friends and their friends, on average, have more than six hundred friends.” (p. 124)
-Network robustness. “Here, we consider the question of how the size of the largest connected component of the network, the giant component, changes as nodes randomly fail… In the random network, the size of the giant component falls linearly at first. At a critical value where the probability of an edge equals 1 divided by the number of nodes, the size of the largest component falls to an arbitrarily small proportion of the original network size. The small-world network shows no such abrupt change. A majority of t connections exist within the geographic clusters.” (p. 127)
-”Most consumer goods and information spread through both diffusion and broadcast. Our next model, the bass model, combines the two processes in a single model.” (p. 136)
-SIR model: The susceptible-infected-recovered model has a built-in saturation assumption. (p. 137)
-The minimum percentage of immunity in a population, the vaccination threshold, depends on the basic reproduction number (R_0), which says how many people get infected by each person that is newly infected. The vaccination threshold is: V \geq \frac{R_0 - 1}{R_0}. Polio has a R_0 of 6, so the vaccine must cover ⅚ of the population.(p. 138-139)
-Superspreaders are key to epidemics. This is due to degree squaring. Superspreaders have more contacts to others, which exposes them to more risk for getting a disease and also increases the chances to spread the disease in the same manner. A superspreader with 3x more social contact will therefore be 3x more likely and 3x more dangerous when infected, making him 9x more dangerous overall. (pp. 139-140).
-R_0 and the vaccination threshold are contextual tipping points. Small changes in the context (environment) can change outcomes significantly. The current focus in public discussions around R_0 and vaccinations show that there is awareness for the criticality of these measures. (p. 141)
-Grossman and Stiglitz paradox: If investors believe in the efficient market hypothesis, they stop analyzing, making markets inefficient. (p. 160)
-Agent-based models are computer programs that model each agent individually. (p. 213)
-Hedonic attributes: more is always better. Spatial attributes: sweet spots that are different among agents. (pp. 227-228)
-Voronoi neighborhoods help to see which characteristics a product should have to optimally fit a certain customer group. (p. 231)
-”A one-dimensional model implies that candidates position at the median. Higher-dimensional models imply that they should not. Which type of model do we believe? We should place complete faith in neither model, but instead gain insights from both… We should instead expect complexity, an endless dance of competition for votes through coalition building.” (p. 234)
-Models of cooperation: In this example: a lone cooperator cannot spawn an additional cooperator, but two adjacent cooperators can. It follows that a small cluster of cooperative nodes surrounded by empty cells could expand into open nodes. Therefore, regions of cooperation can emerge from a handful of cooperators. (p. 264) Group selection is another way to achieve efficient cooperation. Let groups form. The best group (the one with many cooperators) will emerge as a winner and make copies of itself or otherwise project its system on the rest of the population. This can be easier than to just have one large population that has to be made cooperative. (p. 265)
January 28, 2020
OK. So, this is a beast which is neither fish nor fowl. It’s somewhere between popular science and a coursebook.
Many model thinking is so hot right now (see Munger, Parish etc etc.). This book kinda fits into that genre.
On the other hand, it is also a much more serious treatment of how to apply analytic models, and it’s almost a textbook for his course: https://www.coursera.org/learn/model-...
That said, the book doesn’t get into the nitty gritty of how to use each approach. That’s not necessarily a problem – learning how to use an analytic technique ‘for realz’ takes time.
What I liked: I’m a researcher/ data analyst. My stats skills are pretty solid, but are very much grounded in a psychology background (almost everything I do belongs to the family of regression/ structural equation models). What this did is blow things wide open and offer a lot of other possible views.
The book uses a variety of examples to make it’s point.
He also makes the argument of multiple models as offering complementary views and tools (a sort of pluralist approach).
My favourite parts: The illustrations of using multiple models to offer complementary explanations of the same phenomenon, e.g. the Global Financial Crisis, the Cuban missile crisis, and rising social inequality.
Minor objections: Page is clearly a strong advocate for combining models (many model thinking). The book could maybe do with a little more discussion on the practicalities of combining models, and also potential errors/ drawbacks.
Given Page talks about causality a bit, I think it could have done with its own chapter. Selfishly, I’d love to see his take on Judea Pearl’s work, and how he compares and contrasts it to other approaches.
TBH, my biggest objection is that some of the images and equations came through with very poor image quality on the kindle. That problem is not unique to this book, but it’s still a bit annoying.
All that said, it was good enough for me to enrol in the Coursera version.
Many model thinking is so hot right now (see Munger, Parish etc etc.). This book kinda fits into that genre.
On the other hand, it is also a much more serious treatment of how to apply analytic models, and it’s almost a textbook for his course: https://www.coursera.org/learn/model-...
That said, the book doesn’t get into the nitty gritty of how to use each approach. That’s not necessarily a problem – learning how to use an analytic technique ‘for realz’ takes time.
What I liked: I’m a researcher/ data analyst. My stats skills are pretty solid, but are very much grounded in a psychology background (almost everything I do belongs to the family of regression/ structural equation models). What this did is blow things wide open and offer a lot of other possible views.
The book uses a variety of examples to make it’s point.
He also makes the argument of multiple models as offering complementary views and tools (a sort of pluralist approach).
My favourite parts: The illustrations of using multiple models to offer complementary explanations of the same phenomenon, e.g. the Global Financial Crisis, the Cuban missile crisis, and rising social inequality.
Minor objections: Page is clearly a strong advocate for combining models (many model thinking). The book could maybe do with a little more discussion on the practicalities of combining models, and also potential errors/ drawbacks.
Given Page talks about causality a bit, I think it could have done with its own chapter. Selfishly, I’d love to see his take on Judea Pearl’s work, and how he compares and contrasts it to other approaches.
TBH, my biggest objection is that some of the images and equations came through with very poor image quality on the kindle. That problem is not unique to this book, but it’s still a bit annoying.
All that said, it was good enough for me to enrol in the Coursera version.
January 20, 2019
The title of this book is a little misleading. While the author provides a clear, informative overview of the different types of models and the kinds of problems they are regularly used to solve, he never touches on examples of where you might want to apply them. I was expecting a little more guidance on what this information means to me.
In spite of that, I could see this being very useful, and it did make me want to read more on some of the topics covered here.
In spite of that, I could see this being very useful, and it did make me want to read more on some of the topics covered here.
March 22, 2020
Fascinating read. It managed to explain a wide range of intense math/statistical topics in an approachable way. It is definitely more technical than a pop science book (Sapiens, Sixth Extinction) but more accessible than a textbook... take from that what you will. The true test will be if I retained any of it for use in daily/life. I certainly hope so.
June 9, 2019
Charlie Munger (the billionaire investor/partner of Warren Buffett) said: "The first rule is that you’ve got to have multiple models—because if you just have one or two that you’re using, the nature of human psychology is such that you’ll torture reality so that it fits your models, or at least you’ll think it does. You become the equivalent of a chiropractor who, of course, is the great boob in medicine.
It’s like the old saying, “To the man with only a hammer, every problem looks like a nail.” And of course, that’s the way the chiropractor goes about practicing medicine. But that’s a perfectly disastrous way to think and a perfectly disastrous way to operate in the world. So you’ve got to have multiple models."
We solve problems every day and our approach to problems is biased based on what we've learned
and experienced so far. The saying "To the man with only a hammer, every problem looks like a nail" is an apposite summarization of the above behavior. The problem with this approach is that it can be dangerous and result in completely opposite outcomes deviating from the expected. When workers join a new organization and the conventional statement in the meeting room is: "It's always been done this way." We have to give the problem its due respect and action in an appropriate way. This is where model thinking comes into play; Pague's book performs a comprehensive review of all the major models that can be used to explain how the world works - from policy decisions by governments/disease prevention organizations to equilibriums in game theory to almost even time distribution of parents' time with their kids. :)
In short, once we validate (with available historical data) that a model can be fit to a problem, the model will be able to predict the future state of the system. In the machine learning world, this can be used to predict what books you might like to read, people you might want to date, etc., In another world, models can predict the eventual state of a system that might appear random at first. For example: Pague introduces to the Balancing Process in the chapter on Path Dependence. Imagine a urn containing 1 white ball and 1 grey ball. You pick a ball at random: if the ball is white, you add the white ball back but adding 1 more gray ball and vice versa - this is the balancing process to ensure you will likely see a grey ball in the next turn with higher probability. Repeating this process indefinitely shows that the system will converge to an equilibrium state consisting of almost equal proportions of white balls and grey balls in the urn! The real life example of this is parenting with 2 kids. If you spend time with your elder one in the morning, you might most likely spend time with the younger one in the evening.
The book is an exhaustive summary of many models that could be applied at home, work and government organizations. It was intriguing to get a peek into how centers like CDC model disease outbreaks and control the spread (these come under network models which also covers social network phenomena like Facebook). The book is more textbook style narrative although Pague annotates every model with simple numerical examples (to illustrate the models) and relevant real world applications of the models. What surprised me was the plethora of scenarios where models could be applied and in some cases, either the same model could be applied to many domains or many models can be combined to explain one scenario (thus the many model approach referred to as model ensemble).
Pague's book is deep, dense, highly practical and very helpful to get clarity about how the world operates. Most of this is rooted in consumer psychology and rational behavior. Some of the models/chapters might need extra research to get a better understanding and that's where I felt Pague could have compressed too much into one book. IMO, the book could have been better off as a 2 part series but nevertheless, it's highly worth your time particularly if you are in the knowledge industry. Go for it!
It’s like the old saying, “To the man with only a hammer, every problem looks like a nail.” And of course, that’s the way the chiropractor goes about practicing medicine. But that’s a perfectly disastrous way to think and a perfectly disastrous way to operate in the world. So you’ve got to have multiple models."
We solve problems every day and our approach to problems is biased based on what we've learned
and experienced so far. The saying "To the man with only a hammer, every problem looks like a nail" is an apposite summarization of the above behavior. The problem with this approach is that it can be dangerous and result in completely opposite outcomes deviating from the expected. When workers join a new organization and the conventional statement in the meeting room is: "It's always been done this way." We have to give the problem its due respect and action in an appropriate way. This is where model thinking comes into play; Pague's book performs a comprehensive review of all the major models that can be used to explain how the world works - from policy decisions by governments/disease prevention organizations to equilibriums in game theory to almost even time distribution of parents' time with their kids. :)
In short, once we validate (with available historical data) that a model can be fit to a problem, the model will be able to predict the future state of the system. In the machine learning world, this can be used to predict what books you might like to read, people you might want to date, etc., In another world, models can predict the eventual state of a system that might appear random at first. For example: Pague introduces to the Balancing Process in the chapter on Path Dependence. Imagine a urn containing 1 white ball and 1 grey ball. You pick a ball at random: if the ball is white, you add the white ball back but adding 1 more gray ball and vice versa - this is the balancing process to ensure you will likely see a grey ball in the next turn with higher probability. Repeating this process indefinitely shows that the system will converge to an equilibrium state consisting of almost equal proportions of white balls and grey balls in the urn! The real life example of this is parenting with 2 kids. If you spend time with your elder one in the morning, you might most likely spend time with the younger one in the evening.
The book is an exhaustive summary of many models that could be applied at home, work and government organizations. It was intriguing to get a peek into how centers like CDC model disease outbreaks and control the spread (these come under network models which also covers social network phenomena like Facebook). The book is more textbook style narrative although Pague annotates every model with simple numerical examples (to illustrate the models) and relevant real world applications of the models. What surprised me was the plethora of scenarios where models could be applied and in some cases, either the same model could be applied to many domains or many models can be combined to explain one scenario (thus the many model approach referred to as model ensemble).
Pague's book is deep, dense, highly practical and very helpful to get clarity about how the world operates. Most of this is rooted in consumer psychology and rational behavior. Some of the models/chapters might need extra research to get a better understanding and that's where I felt Pague could have compressed too much into one book. IMO, the book could have been better off as a 2 part series but nevertheless, it's highly worth your time particularly if you are in the knowledge industry. Go for it!
August 4, 2020
Книга повторяющая отличный курс на Курсере по Модельному мышлению. Непростое чтение, но для тех, кто хочет разбираться в мат моделях неизбежное. Я проходил курс, поэтому многое уже было знакомо. Сложнее с применением, хоть и очень хочется использовать
April 22, 2021
After using Scott Page's previous work on path dependence models during my bachelor thesis, I was really excited when this book was suggested during my master. A beautiful book! It took me some time to go through all the introduced models but it was worth it!
April 12, 2020
A somewhat dry, but unique text, that concisely presents a worldview it took me years of piecemeal study to stitch together from bits of economics, behavioral science, computer science, biology, complex systems, and beyond. You might think of it as “computational social science”, though it certainly extends further.
It does a very nice job of pointing to real-world scenarios that can be usefully analyzed by each model, and highlights topics of equity, diversity, and other pro-social values, specifically. Recommended if you’re interested in a broad set of models with which to understand how the world works, and the levers at our disposal to improve it.
It does a very nice job of pointing to real-world scenarios that can be usefully analyzed by each model, and highlights topics of equity, diversity, and other pro-social values, specifically. Recommended if you’re interested in a broad set of models with which to understand how the world works, and the levers at our disposal to improve it.
April 7, 2021
Read all but the 2nd-to-last and 3rd-to-last chapters, because I ran out of time before the book was due back at the library. I did read the final chapter. I found the book to be very dense, but interesting at times. The first 100 pages were tedious and largely review for me. There were some interesting nuggets from game theory, signaling models, random walks, and spatial vs. hedonic models. All in all, like an encyclopedia that would be best digested in 6+weeks, each chapter lulling you to sleep. Or a reference book to come back to and skim to find an idea.
September 10, 2021
Nice collection of models. If you have some behavioural science modelling experience a lot will be familiar. A rough sample: game theory stuff, epidemiological models, Markov models, some statistical distributions, learning models, network models. I found the explanations well done, the chapters are relatively short and lean more in the direction of intuition and application, but the basic equations are always featured and explained as well.
I'd recommend skipping the first two chapters, at least for me they were a drag and I feel like they can be summarized like this kinda obvious point: Don't rely on any single model, they almost always simplify too much, at least use multiple models.
I really liked this quote from Jean Piaget, apparently from 1968: Knowing reality means constructing systems of transformations that correspond, more or less adequately, to reality.
And I think I found a mistake in his definition of Simpson's paradox?
Shouldn't the average number of women of the total population just be a weighted average of the average numbers of women in all subpopulations, and therefore not be able to fall below 50%? Anybody can explain what I'm missing?
I'd recommend skipping the first two chapters, at least for me they were a drag and I feel like they can be summarized like this kinda obvious point: Don't rely on any single model, they almost always simplify too much, at least use multiple models.
I really liked this quote from Jean Piaget, apparently from 1968: Knowing reality means constructing systems of transformations that correspond, more or less adequately, to reality.
And I think I found a mistake in his definition of Simpson's paradox?
Logic can also reveal paradoxes. Using models we can show the possibility of each subpopulation containing a larger percentage of women than men but the total population containing a larger percentage of men, a phenomenon (Simpson’s paradox).
Shouldn't the average number of women of the total population just be a weighted average of the average numbers of women in all subpopulations, and therefore not be able to fall below 50%? Anybody can explain what I'm missing?
December 24, 2020
This books offers insight into so many disciplines by giving a primer of so many models and concepts, developed by different people for different problems encountered in the real world. While it can only be an introductory reading due to the vast number of models presented, it is still very dense and at times very technical. So be prepared to do some skimming if you want to get through it. But your horizon will be broadened.
October 12, 2024
insightful
March 10, 2022
This is a decent introduction to several modeling techniques. Its strength is its weakness: It delivers very short, descriptive introductions to the models, which means that a lot of important information is left out. Balancing brevity, adequate rigor, and readability for a book on modeling techniques would certainly be a tough task. Some chapters are more successful at this than others. Hopefully that helps you to decide if this is a good fit depending on what you're seeking.
April 23, 2019
This book is awesome. First, it heightened my appreciation for math. I've never had the opportunity to learn about models in such great detail and extensiveness. The book explains formally many models used to understand several problems and phenomenons that happen in the current world. The models help you understand in a mathematical and formal way some thoughts you may have intuitively and explain it. Other times, applying the same model to different problems makes you understand things you would never have understood if you didn't see the problem from that perspective.
I cannot say I understood the book fully because there are some math equations I'm not prepared to work with 😛, I need refreshers on Differential Calculus to understand some mathematical arguments.
I want to re-read this book to understand it better, and start a notebook with my notes on the book, but first a refresher on Math.
The models I found easier to understand an apply are:
- Markov Chain models
- System Dynamics
- Vote models
- Game models
It helped me understand how to model several things mathematically.
It also helped me to understand the four results a system could
The first pages of the book state the case of learning many thinking models in order to better understand the world, and throughout all the book the author repeats the case with examples of why it is important to learn several models applicable to the problem. I've learned thoroughly how applying different models to the same problem can you give several points of views, excluding details that are not important in that specific view and highlighting the most relevant aspects.
Finally, the humbling component of the book: there's so much to learn that is difficult to understand and there are things of the world we could not understand even with all our advanced knowledge.
I cannot say I understood the book fully because there are some math equations I'm not prepared to work with 😛, I need refreshers on Differential Calculus to understand some mathematical arguments.
I want to re-read this book to understand it better, and start a notebook with my notes on the book, but first a refresher on Math.
The models I found easier to understand an apply are:
- Markov Chain models
- System Dynamics
- Vote models
- Game models
It helped me understand how to model several things mathematically.
It also helped me to understand the four results a system could
The first pages of the book state the case of learning many thinking models in order to better understand the world, and throughout all the book the author repeats the case with examples of why it is important to learn several models applicable to the problem. I've learned thoroughly how applying different models to the same problem can you give several points of views, excluding details that are not important in that specific view and highlighting the most relevant aspects.
Finally, the humbling component of the book: there's so much to learn that is difficult to understand and there are things of the world we could not understand even with all our advanced knowledge.
June 27, 2024
Building on past book recommendations focussed on Statistical Thinking or Critical Thinking skills comes this recommendation of “The Model Thinker”.
I am rarely as impressed and challenged by a book as this one. It is a tour de force of mathematical and statistical models worth understanding and considering. Beyond that, it applies each to real-life problems and brings to life their relevance for the worlds of commerce and public policy decisions. An ethos pervades the entire book; the need to apply many models to enable better thinking about issues.
So, in this post, I will share what I found within “The Model Thinker” by Scott E. Page and why I recommend it as a text to all data & analytics leaders with sufficient maths knowledge. It could be a great aid to continuing to develop, nurture and hone both your knowledge of models and your application. Let’s get started and dig into what you can find in this treasure trove of a book...
Why we need Many Model Thinkers
Let me first give you some background on the author, to establish his credibility as your teacher (via this well-written book). Dr. Scott Page is the Leonid Hurwicz collegiate professor of Complex Systems, Political Science & Economics at the University of Michigan. In addition to that very long job title, he’s also a faculty member of the Santa Fe Institute and teaches at a number of other universities. He has written five books on topics including diversity and complexity theory. But this appears to be his best to date, certainly the most relevant to all mathematically grounded data leaders.
Within the opening three chapters of his book, Scott makes the case for why we need models and many of them. His examples bring to life the twin problems that we need models (not just data) and that any one model is an imperfect representation of the real world (or overfitted). He makes a great case for the benefits of what he calls many model thinking. By this, he means having the knowledge & skills to apply a variety of relevant models to a single problem.
The vast majority of the rest of this book introduces the reader to a broad portfolio of models to consider. In each of the next 24 chapters, Scott introduces a new class of models. He explains the thinking behind that approach, the mathematics of some models in that class and how they could be applied. Here the book excels many others on the market. It combines sufficient mathematical depth to understand and try the model yourself plus guided applications to real-world problems.
Which models are covered in The Model Thinker
Perhaps you’re scratching your head thinking, how can he fill 24 chapters with different classes of models? In your day-to-day work, it might feel like beyond significance testing, regression models, random forest and occasionally clustering – what other models matter? If so, prepare to have your memory jogged (if you studied Maths or Statistics degrees) or your eyes opened. There is a big wide world of mathematical models out there that you may not have considered.
48
Book Reviews to cross post to Amazon & Goodreads
Are you a Many Model Thinker? Here's how...
To bring that to life (at just a high level) here are some of the topics covered:
1. Categorisation Models & Model Error Decomposition Theorem
2. Rational-Actor Model, Prospect Theory & Rule-Based models of human behaviour 3. Normal Distribution Models & Central Limit Theorem
4. Power-Law Distribution Models & Preferential Attachment Model
5. Linear Models & Multi-Variate Regression Models
6. Concavity & Convexity, plus related Growth Models
7. Models of Value & Power (inc. Shapley Value)
8. Network Models (inc. explaining 6 degrees of separation model)
9. Broadcast, Diffusion & Contagion Models (inc. SIR Model)
10. Entropy Models
11. Random Walk Models (inc. Bernoulli Urn Model)
12. Path Dependence Models (inc. Polya & Balancing processes) 13. Local Interaction Models (inc. cooperation in Game Theory) 14. Lyapunov Models & Equilibria
15. Markov Models (inc. Markov Decision Models)
16. Systems Dynamic Models
17. Threshold Models with feedback (inc. Schelling models)
18. Spatial & Hedonic Choice Models
19. Game Theory Models (zero-sum, sequential & continuous) 20. Models of Cooperation (inc. Clustering Bootstraps Coop)
21. Collection Action Problems (inc. altruism & congestion)
22. Mechanism Design (inc. Revenue Equivalence Theorem)
23. Signalling Models (discrete & continuous)
24. Learning Models (inc. learning reinforcement)
25. Multi-Armed Bandit problems (inc. Bernoulli & Bayesian) 26. Rugged Landscape Models (inc. NK Model)
Phew! I know! I’m tired from just typing the above & refreshing my memory on what I’ve read. No doubt, if you wanted to go slowly & apply all you learnt above this, should count as a degree-level course. However, the magic of this book is with all that detail and example equations, it is still very readable. You can enjoy so much theory because of how each is applied and brought to life.
Application is still what counts with your Models
That last comment highlights the main reason that I recommend this book so highly. It will help mathematically adept data leaders apply their knowledge. Challenging us to not slip back into lazy ways and to consider only one or a few models when approaching familiar problems. Helping us notice the true complexity behind seeking to model human decision-making & behaviour.
Such examples and application advice permeate the whole book. They come to a crescendo in the final chapter. In chapter 29, Scott turns his attention to 3 major challenges in the world today. Opioid addiction, the COVID-19 pandemic & growing inequality. For each of those challenges, he reveals how the use of multiple models can improve both thinking and policymaking. The book is worth reading for
49
Book Reviews to cross post to Amazon & Goodreads
Are you a Many Model Thinker? Here's how...
this chapter alone, it really helps you to better critically assess past actions & future options.
I hope the above review was useful and inspires you to invest in this book. After an initial read, it could serve you well as a reference. A guide to help you recall other models to consider & how they might apply to what you are seeking to understand. But, even if you don’t buy Scott’s book, do take up the challenge to adopt many model thinking. If you are currently basing your forecast or explanation of human behaviour on just one model, try another. Consider the other dynamics which could be at play. Experiment with using multiple models to give you a multi-dimensional view of that complexity.
I am rarely as impressed and challenged by a book as this one. It is a tour de force of mathematical and statistical models worth understanding and considering. Beyond that, it applies each to real-life problems and brings to life their relevance for the worlds of commerce and public policy decisions. An ethos pervades the entire book; the need to apply many models to enable better thinking about issues.
So, in this post, I will share what I found within “The Model Thinker” by Scott E. Page and why I recommend it as a text to all data & analytics leaders with sufficient maths knowledge. It could be a great aid to continuing to develop, nurture and hone both your knowledge of models and your application. Let’s get started and dig into what you can find in this treasure trove of a book...
Why we need Many Model Thinkers
Let me first give you some background on the author, to establish his credibility as your teacher (via this well-written book). Dr. Scott Page is the Leonid Hurwicz collegiate professor of Complex Systems, Political Science & Economics at the University of Michigan. In addition to that very long job title, he’s also a faculty member of the Santa Fe Institute and teaches at a number of other universities. He has written five books on topics including diversity and complexity theory. But this appears to be his best to date, certainly the most relevant to all mathematically grounded data leaders.
Within the opening three chapters of his book, Scott makes the case for why we need models and many of them. His examples bring to life the twin problems that we need models (not just data) and that any one model is an imperfect representation of the real world (or overfitted). He makes a great case for the benefits of what he calls many model thinking. By this, he means having the knowledge & skills to apply a variety of relevant models to a single problem.
The vast majority of the rest of this book introduces the reader to a broad portfolio of models to consider. In each of the next 24 chapters, Scott introduces a new class of models. He explains the thinking behind that approach, the mathematics of some models in that class and how they could be applied. Here the book excels many others on the market. It combines sufficient mathematical depth to understand and try the model yourself plus guided applications to real-world problems.
Which models are covered in The Model Thinker
Perhaps you’re scratching your head thinking, how can he fill 24 chapters with different classes of models? In your day-to-day work, it might feel like beyond significance testing, regression models, random forest and occasionally clustering – what other models matter? If so, prepare to have your memory jogged (if you studied Maths or Statistics degrees) or your eyes opened. There is a big wide world of mathematical models out there that you may not have considered.
48
Book Reviews to cross post to Amazon & Goodreads
Are you a Many Model Thinker? Here's how...
To bring that to life (at just a high level) here are some of the topics covered:
1. Categorisation Models & Model Error Decomposition Theorem
2. Rational-Actor Model, Prospect Theory & Rule-Based models of human behaviour 3. Normal Distribution Models & Central Limit Theorem
4. Power-Law Distribution Models & Preferential Attachment Model
5. Linear Models & Multi-Variate Regression Models
6. Concavity & Convexity, plus related Growth Models
7. Models of Value & Power (inc. Shapley Value)
8. Network Models (inc. explaining 6 degrees of separation model)
9. Broadcast, Diffusion & Contagion Models (inc. SIR Model)
10. Entropy Models
11. Random Walk Models (inc. Bernoulli Urn Model)
12. Path Dependence Models (inc. Polya & Balancing processes) 13. Local Interaction Models (inc. cooperation in Game Theory) 14. Lyapunov Models & Equilibria
15. Markov Models (inc. Markov Decision Models)
16. Systems Dynamic Models
17. Threshold Models with feedback (inc. Schelling models)
18. Spatial & Hedonic Choice Models
19. Game Theory Models (zero-sum, sequential & continuous) 20. Models of Cooperation (inc. Clustering Bootstraps Coop)
21. Collection Action Problems (inc. altruism & congestion)
22. Mechanism Design (inc. Revenue Equivalence Theorem)
23. Signalling Models (discrete & continuous)
24. Learning Models (inc. learning reinforcement)
25. Multi-Armed Bandit problems (inc. Bernoulli & Bayesian) 26. Rugged Landscape Models (inc. NK Model)
Phew! I know! I’m tired from just typing the above & refreshing my memory on what I’ve read. No doubt, if you wanted to go slowly & apply all you learnt above this, should count as a degree-level course. However, the magic of this book is with all that detail and example equations, it is still very readable. You can enjoy so much theory because of how each is applied and brought to life.
Application is still what counts with your Models
That last comment highlights the main reason that I recommend this book so highly. It will help mathematically adept data leaders apply their knowledge. Challenging us to not slip back into lazy ways and to consider only one or a few models when approaching familiar problems. Helping us notice the true complexity behind seeking to model human decision-making & behaviour.
Such examples and application advice permeate the whole book. They come to a crescendo in the final chapter. In chapter 29, Scott turns his attention to 3 major challenges in the world today. Opioid addiction, the COVID-19 pandemic & growing inequality. For each of those challenges, he reveals how the use of multiple models can improve both thinking and policymaking. The book is worth reading for
49
Book Reviews to cross post to Amazon & Goodreads
Are you a Many Model Thinker? Here's how...
this chapter alone, it really helps you to better critically assess past actions & future options.
I hope the above review was useful and inspires you to invest in this book. After an initial read, it could serve you well as a reference. A guide to help you recall other models to consider & how they might apply to what you are seeking to understand. But, even if you don’t buy Scott’s book, do take up the challenge to adopt many model thinking. If you are currently basing your forecast or explanation of human behaviour on just one model, try another. Consider the other dynamics which could be at play. Experiment with using multiple models to give you a multi-dimensional view of that complexity.
February 22, 2022
The good: the outline, the breadth, the sheer volume of material. The reader gets bits of everything.
The bad: skimming and no really deep dives. The bits are way too tiny.
The not-nearly detailed enough: - math and the know-how. I don't think many newbies will manage to do many (or even any) of these models basing on just this one skinny volume. But then again, a girl can dream of a book that has it all, right?
The really good take-outs driver: 29 model groups are conversationally discussed, most of these are a mix of some more models that are really broadly used in a lot of industries. A lot of motivation for usage is quite creatively and readably layered out. Nice work!
The bad: skimming and no really deep dives. The bits are way too tiny.
The not-nearly detailed enough: - math and the know-how. I don't think many newbies will manage to do many (or even any) of these models basing on just this one skinny volume. But then again, a girl can dream of a book that has it all, right?
The really good take-outs driver: 29 model groups are conversationally discussed, most of these are a mix of some more models that are really broadly used in a lot of industries. A lot of motivation for usage is quite creatively and readably layered out. Nice work!
February 16, 2019
This is a great source of wisdom. 25 something general prediction models with simple math explanation, good examples and comprehensive insights. Many model approach reasonable as nothing else in this world. Invaluable food for thought!
Read
September 1, 2021In an increasingly complex world, we need models to make sense of the perplexing systems around us. They can help us to explain the world, to create new designs, and to predict what’s coming next – but only if we’re careful to apply the right ones. In fact, to optimize our results, we should try to tackle a problem using as many diverse and relevant models as we can.
---
The key message here is: Modeling humans is a thorny endeavor.
People can be troublesome: if that’s true when it comes to real life, then it’s doubly true when it comes to modeling.
Unlike bowling balls or weather cycles, humans have agency. We have minds of our own and we use them to make decisions. We experience social pressure, we have different preferences, and we often make mistakes. Sometimes, we might even learn from those mistakes.
That’s what makes us fascinating – but from the point of view of modeling, it also makes us frustrating. So does that mean it’s hopeless to try to model human behavior?
Whenever we try to model human behavior, there are some choices and assumptions that we can’t avoid. The first decision we have to make is between two different ways of picturing it – as rule-based or as rational.
Rule-based behavior can be broken down into two main types: fixed and adaptive. Fixed rules don’t evolve as time passes and circumstances change. If we were to formulate a fixed rule in words, we might come up with something like “If the conversation lulls for longer than 20 seconds, change the topic.”
An adaptive rule, on the other hand, changes and evolves in response to new circumstances and novel information. An adaptive rule might allow silences to go on for longer than 20 seconds if it became apparent that these conversational lulls ultimately led to better discussions.
By contrast, the rational-actor model of human behavior assumes that people make rational decisions in order to achieve optimal outcomes. Instead of following set rules, they calculate what the best action is in any given situation and act on that information.
Think of someone buying a house and calmly weighing up their options: the number of bedrooms each house has, the view out the kitchen windows, even the schools in each neighborhood.
Neither rule-based nor rational-actor models work for every situation. When choices are simple or made by sophisticated decision-makers, they're likely to be rational ones. But when a choice is fairly low-stakes, like what color coat to buy, it’s likely that people will apply fixed rules. In others, like deciding who to trust in a delicate negotiation, people might apply adaptive rules.
When it comes to modeling humans, we can rarely hope for complete accuracy. But choosing the right models can make our predictions, designs, and explanations far more accurate.
---
The key message here is: Modeling humans is a thorny endeavor.
People can be troublesome: if that’s true when it comes to real life, then it’s doubly true when it comes to modeling.
Unlike bowling balls or weather cycles, humans have agency. We have minds of our own and we use them to make decisions. We experience social pressure, we have different preferences, and we often make mistakes. Sometimes, we might even learn from those mistakes.
That’s what makes us fascinating – but from the point of view of modeling, it also makes us frustrating. So does that mean it’s hopeless to try to model human behavior?
Whenever we try to model human behavior, there are some choices and assumptions that we can’t avoid. The first decision we have to make is between two different ways of picturing it – as rule-based or as rational.
Rule-based behavior can be broken down into two main types: fixed and adaptive. Fixed rules don’t evolve as time passes and circumstances change. If we were to formulate a fixed rule in words, we might come up with something like “If the conversation lulls for longer than 20 seconds, change the topic.”
An adaptive rule, on the other hand, changes and evolves in response to new circumstances and novel information. An adaptive rule might allow silences to go on for longer than 20 seconds if it became apparent that these conversational lulls ultimately led to better discussions.
By contrast, the rational-actor model of human behavior assumes that people make rational decisions in order to achieve optimal outcomes. Instead of following set rules, they calculate what the best action is in any given situation and act on that information.
Think of someone buying a house and calmly weighing up their options: the number of bedrooms each house has, the view out the kitchen windows, even the schools in each neighborhood.
Neither rule-based nor rational-actor models work for every situation. When choices are simple or made by sophisticated decision-makers, they're likely to be rational ones. But when a choice is fairly low-stakes, like what color coat to buy, it’s likely that people will apply fixed rules. In others, like deciding who to trust in a delicate negotiation, people might apply adaptive rules.
When it comes to modeling humans, we can rarely hope for complete accuracy. But choosing the right models can make our predictions, designs, and explanations far more accurate.
November 16, 2019
The case made by the author is that social systems are very complex and often have vast numbers of what could be causal forces at play. In physics, chemistry and biology, sciences from which social sciences emerge we have often had few enough variables that relatively simple models have been constructed since the enlightenment that are so surefooted at predicting fairly granular aspects of the future that we call them laws. The social sciences (economics, psychology, political science, sociology, etc.) that are emergent from the physical sciences and have complexity that makes creating model that consistently predict future outcomes correctly enough to be called laws difficult at best. Such models would likely be too complex for our minds to use effectively.
So what to do? Should we examine with a single model that is more detailed and more complex. The author makes a strong case that often this leads to less understanding and is impossible to calibrate meaningfully. But there is a proven way to make more accurate predictions. If we simplify a problem in the social sciences we can model it and still gain valuable insight. And then we can simplify the the model in several other contexts and model it in those ways. The prime insight that I obtained from the book is that if we simplify a complex system problem in the context of different forces we gain a much better understanding of a system and can gain a better understanding of its future behavior and make better decisions.
The Model Thinker then goes on to explain and give examples of many types of model types. These include but are not limited to: linear models, power models, network models, broadcast models, diffusion models contagion models, path dependent models, local interaction models, Markov models, systems dynamics models, game theory models, cooperation models, collective action models, etc. Examples of problems are analyzed with every model introduced. Of course even though the math shows the power of this processes, most of us don’t have intuition related to more complex math. The author finishes the book in a way that addresses this. Two complex system problems, opioid addiction and wealth inequality are examined with multiple models to show the power of this approach.
So what to do? Should we examine with a single model that is more detailed and more complex. The author makes a strong case that often this leads to less understanding and is impossible to calibrate meaningfully. But there is a proven way to make more accurate predictions. If we simplify a problem in the social sciences we can model it and still gain valuable insight. And then we can simplify the the model in several other contexts and model it in those ways. The prime insight that I obtained from the book is that if we simplify a complex system problem in the context of different forces we gain a much better understanding of a system and can gain a better understanding of its future behavior and make better decisions.
The Model Thinker then goes on to explain and give examples of many types of model types. These include but are not limited to: linear models, power models, network models, broadcast models, diffusion models contagion models, path dependent models, local interaction models, Markov models, systems dynamics models, game theory models, cooperation models, collective action models, etc. Examples of problems are analyzed with every model introduced. Of course even though the math shows the power of this processes, most of us don’t have intuition related to more complex math. The author finishes the book in a way that addresses this. Two complex system problems, opioid addiction and wealth inequality are examined with multiple models to show the power of this approach.
January 30, 2022
Models have not been this fashionable since Linda Evangelista, Naomi Campbell, and Claudia Schiffer were strutting down catwalks in the 90's. The rise of big data, and high-profile debate over models for climate change and COVID have made mathematical models increasingly important. The Model Thinker does an admirable job explaining many different models, making a convincing case that using many models to investigate issues from different perspectives is a sensible approach.
Page provides 29 chapters, most of which deal with a separate set of models, moving from concepts familiar to high school stats students like power laws and normal distributions through Lyapunov functions to game theory and rugged landscapes. If you've ever wondered what pseudo-intellectual writers mean when they mention entropy or Markov models, this book is an excellent place to start.
Each chapter neatly explains the concept behind each model, the underlying maths, and examples of how they work in practice. The text heavy approach means you need to keep your wits around you - the more complex models require less mathematical-minded readers to re-read and check the text carefully. More liberal use of graphs and illustrations would have been beneficial, but the avoidance of too much maths will be welcomed by many. Given the speed with which some techniques are fine-tuned and adopted, this is a wise approach lest the book becomes too swiftly outdated.
The Model Thinker deserves a place on the desks of anyone working with or interested in data. The final chapter shows why modelling is vital to understanding epidemics, drug addiction, and inequality. We need people with a practical, nuanced understanding of how to best use data to understand the world and society. No single book can provide all the answers, but The Model Thinker is a valuable starting point.
Page provides 29 chapters, most of which deal with a separate set of models, moving from concepts familiar to high school stats students like power laws and normal distributions through Lyapunov functions to game theory and rugged landscapes. If you've ever wondered what pseudo-intellectual writers mean when they mention entropy or Markov models, this book is an excellent place to start.
Each chapter neatly explains the concept behind each model, the underlying maths, and examples of how they work in practice. The text heavy approach means you need to keep your wits around you - the more complex models require less mathematical-minded readers to re-read and check the text carefully. More liberal use of graphs and illustrations would have been beneficial, but the avoidance of too much maths will be welcomed by many. Given the speed with which some techniques are fine-tuned and adopted, this is a wise approach lest the book becomes too swiftly outdated.
The Model Thinker deserves a place on the desks of anyone working with or interested in data. The final chapter shows why modelling is vital to understanding epidemics, drug addiction, and inequality. We need people with a practical, nuanced understanding of how to best use data to understand the world and society. No single book can provide all the answers, but The Model Thinker is a valuable starting point.
September 28, 2019
This is a thorough coverage of various types of models that can help us apply to problems. A strong foundation in mathematics is helpful, as most of these models are spelled out formally via math, but not necessarily required. This book advocates for, importantly, a "many-models" approach to problem analysis: many people approach problems myopically, applying the models and intuitions that are most familiar to them to a problem, convincing themselves they have it all figured out - with unwarranted confidence. The author, I think correctly, argues for humility in approaching problems, and the importance of bringing many analytical tools to bear on every issue, realizing that "all models are wrong" but "many models are useful".
The introductory chapters, basically spelling out what I said above, are the most strong. The rest of the book follows with one model per chapter, and is a bit more hit-or-miss: some chapters are strong and I feel like I learned a lot, others are more esoteric and not quite as useful. Each chapter does come with real-world examples applying the model, which is helpful. With so much content to cover in a relatively short book, each model gets very cursory treatment, though. This is either good or bad depending on what you're trying to get out of the book. I, for one, appreciated being exposed to *many* different ways of thinking, even if each model does not go into much depth.
This is a book that helps expand your landscape of "knowing what there is to know", and in that light I highly recommend it. But I would potentially suggest not reading too carefully into some of the chapters that may not strike your interest.
The introductory chapters, basically spelling out what I said above, are the most strong. The rest of the book follows with one model per chapter, and is a bit more hit-or-miss: some chapters are strong and I feel like I learned a lot, others are more esoteric and not quite as useful. Each chapter does come with real-world examples applying the model, which is helpful. With so much content to cover in a relatively short book, each model gets very cursory treatment, though. This is either good or bad depending on what you're trying to get out of the book. I, for one, appreciated being exposed to *many* different ways of thinking, even if each model does not go into much depth.
This is a book that helps expand your landscape of "knowing what there is to know", and in that light I highly recommend it. But I would potentially suggest not reading too carefully into some of the chapters that may not strike your interest.
May 16, 2020
A tough read - not one you can simply breeze through - a revelation nonetheless, once you have managed to trundle through the first 50 pages. The Model Thinker spans across multiple disciplines, throwing light on the most critical models yet discovered or constructed for various contexts. After laying down the basics of a model, Prof. Page highlights the insights gleaned from the model's applications in the germane contexts.
Prof. Page takes care not to delve too deep into each model, for then, the book would inevitably have run into multiple volumes, and also would have become inaccessible to a bigger audience. Needless to say, most of the models might need of the reader a solid mathematical grounding or at least the intuition. Those with the intuition might have to think deeper through each explanation to truly understand the implications.
Even if you could not really grasp all the maths, I would still recommend you to focus on the takeaways highlighted for each model, for those could really help guide you and your larger group's decisions in the contexts that you might experience in personal, public or professional lives.
In the end though, the book stresses most on the importance of applying multiple (appropriate) models to a problem or situation to cater to any of the REDCAPE uses (Reason, Explain, Design, Communicate, Act, Predict, Explore). Also, since 'all models are wrong', it follows that using ensembles leads to more accurate results than would have emerged from using the individual models (wisdom of the crowds).
Prof. Page takes care not to delve too deep into each model, for then, the book would inevitably have run into multiple volumes, and also would have become inaccessible to a bigger audience. Needless to say, most of the models might need of the reader a solid mathematical grounding or at least the intuition. Those with the intuition might have to think deeper through each explanation to truly understand the implications.
Even if you could not really grasp all the maths, I would still recommend you to focus on the takeaways highlighted for each model, for those could really help guide you and your larger group's decisions in the contexts that you might experience in personal, public or professional lives.
In the end though, the book stresses most on the importance of applying multiple (appropriate) models to a problem or situation to cater to any of the REDCAPE uses (Reason, Explain, Design, Communicate, Act, Predict, Explore). Also, since 'all models are wrong', it follows that using ensembles leads to more accurate results than would have emerged from using the individual models (wisdom of the crowds).
September 19, 2019
How to make sense the data? Use model!
No model is bad. One model is probably good. Many models are better.
Dunia ini kian hari semakin kompleks. Dan for most of the time since cognitive revolution, hanya manusia yang bisa "mengerti" kompleksitas dunia. Manusia melakukan itu dengan membangun "model" di dalam kepalanya. Model itu mungkin cuman intuisi/feeling yang tidak ada bentuk formal matematisnya, tapi sejauh ini terbukti cukup efektif untuk mengendalikan dunia.
Sekarang, dunia yang kompleks ini semakin menjadi lebih kompleks lagi.
Kabar baiknya, kita punya data. Abundance of data. Kabar buruknya, kemampuan kognitif kita mungkin tak bisa lagi mencerna abundance of data tersebut. Kabar baiknya, kita punya komputer dengan kekuatan kalkulasi yang luar biasa. Kabar buruknya, tanpa model yang representatif, kekuatan komputer tadi tidak akan bisa digunakan untuk memproses data secara benar.
That's why we need models! Pertama untuk memahami kompleksitas dunia secara kualitatif. Kedua agar bisa memproses data secara kuantitatif dengan metode yang tepat, which is fitting data to the models.
Dan, satu model tidak cukup. Kita perlu mengkombinasikan beragam model. Mungkin akan lebih kompleks, tapi kita akan mendapatkan pemahaman yang lebih baik.
This book is quite hard for a non-fanatic-math-geek. Tapi gambaran ide setiap model ku rasa cukup mudah dipahami dengan catatan konsentrasi penuh saat membacanya! Wkk
No model is bad. One model is probably good. Many models are better.
Dunia ini kian hari semakin kompleks. Dan for most of the time since cognitive revolution, hanya manusia yang bisa "mengerti" kompleksitas dunia. Manusia melakukan itu dengan membangun "model" di dalam kepalanya. Model itu mungkin cuman intuisi/feeling yang tidak ada bentuk formal matematisnya, tapi sejauh ini terbukti cukup efektif untuk mengendalikan dunia.
Sekarang, dunia yang kompleks ini semakin menjadi lebih kompleks lagi.
Kabar baiknya, kita punya data. Abundance of data. Kabar buruknya, kemampuan kognitif kita mungkin tak bisa lagi mencerna abundance of data tersebut. Kabar baiknya, kita punya komputer dengan kekuatan kalkulasi yang luar biasa. Kabar buruknya, tanpa model yang representatif, kekuatan komputer tadi tidak akan bisa digunakan untuk memproses data secara benar.
That's why we need models! Pertama untuk memahami kompleksitas dunia secara kualitatif. Kedua agar bisa memproses data secara kuantitatif dengan metode yang tepat, which is fitting data to the models.
Dan, satu model tidak cukup. Kita perlu mengkombinasikan beragam model. Mungkin akan lebih kompleks, tapi kita akan mendapatkan pemahaman yang lebih baik.
This book is quite hard for a non-fanatic-math-geek. Tapi gambaran ide setiap model ku rasa cukup mudah dipahami dengan catatan konsentrasi penuh saat membacanya! Wkk
December 13, 2020
Interesting and revealing, a guide to help think on today's complex problems
What's a model? How to use a model? Why to use a model? Any qualitative approach to apply a model? Only use 1 model or various? These are the questions that these book look to help to answer. It is fascinating as depending your area of expertise some model will be recognizable and others not. Maybe someone can say that many other models are not included, or not clear the difference what a model is, is the Standard Atomic model a "model"? Not described here however it is. Same with many others. In any case it is very informative and overeating to understand the why: the more diverse approach to think about a problem, the better answers or guidelines you can get. Some are most difficult to apply unless further study or practice like bandit models or NK models in my case. What about machine learning? Algorithms are also a model? It is a super interesting book to start this thinking approach, and I recommend to do also his course in Coursera: Model Thinking. Ideally to read paired with Algorithms to Live by: different style, different Algorithms (or model?) But completes this vision of learning to apply formal models to analyze a problem and think on different potential solutions.
What's a model? How to use a model? Why to use a model? Any qualitative approach to apply a model? Only use 1 model or various? These are the questions that these book look to help to answer. It is fascinating as depending your area of expertise some model will be recognizable and others not. Maybe someone can say that many other models are not included, or not clear the difference what a model is, is the Standard Atomic model a "model"? Not described here however it is. Same with many others. In any case it is very informative and overeating to understand the why: the more diverse approach to think about a problem, the better answers or guidelines you can get. Some are most difficult to apply unless further study or practice like bandit models or NK models in my case. What about machine learning? Algorithms are also a model? It is a super interesting book to start this thinking approach, and I recommend to do also his course in Coursera: Model Thinking. Ideally to read paired with Algorithms to Live by: different style, different Algorithms (or model?) But completes this vision of learning to apply formal models to analyze a problem and think on different potential solutions.
July 20, 2024
The author does a great job of doing high level overview of different modeling approaches we can use to understand/explore/discover the world around us and find answers to a variety of questions through modeling.
It is definitely high level and I think it gives a great overview of what models are, how they could be used, and then gives you surface level understanding of each with decent examples.
The main ideas are walk away from the book:
- models are wrong, sure, but they are definitely useful. If used right, they make us wiser
- more sophistication and complexity doesn't mean "better". It might be more useful to aim for a variety independent models that can give a different perspective. And if you find that they converge on the answer, you might have something
- also a variety and collection of simple books offer great performance and accurate prediction/appropriate answer, always prefer many to one model
- loved the discussion on normal and power law distributions and what that means in general terms
- dynamic systems modeling offer a way to visualize flow of actions into resources/usage, etc
Anyway, great book if you work in ML, or basically any modeling domain
It is definitely high level and I think it gives a great overview of what models are, how they could be used, and then gives you surface level understanding of each with decent examples.
The main ideas are walk away from the book:
- models are wrong, sure, but they are definitely useful. If used right, they make us wiser
- more sophistication and complexity doesn't mean "better". It might be more useful to aim for a variety independent models that can give a different perspective. And if you find that they converge on the answer, you might have something
- also a variety and collection of simple books offer great performance and accurate prediction/appropriate answer, always prefer many to one model
- loved the discussion on normal and power law distributions and what that means in general terms
- dynamic systems modeling offer a way to visualize flow of actions into resources/usage, etc
Anyway, great book if you work in ML, or basically any modeling domain
August 21, 2021
The agenda of this book is to convince the reader that using multiple models to attempt to understand a situation beats using a single model for most problems of interest.
He does so in a way that lets you enjoy the beauty of models and the surprise of the unexpected results. Very little math is involved, and what there is could be skipped easily.
The book as a whole will only be of interest, I think, to people who love to see how systems in the world work and don’t work. I was engrossed in it immediately but, after several days, set it aside and didn’t come back to it for a few months. It was no problem to pick it up and carry on. It’s semi-cookbook structure lends it to that.
If you liked Freakonomics, or any book of that sort, this is that with more depth and breadth.
The seven-model examination of the causes of wealth/income inequality is extremely enlightening and worth the price of the book. It takes only 11 pages. It will give you a very different perspective and is worth reading, even if you have to get the book from the library or read it at your local bookstore. Starts on page 343! Not one mention of politics.
He does so in a way that lets you enjoy the beauty of models and the surprise of the unexpected results. Very little math is involved, and what there is could be skipped easily.
The book as a whole will only be of interest, I think, to people who love to see how systems in the world work and don’t work. I was engrossed in it immediately but, after several days, set it aside and didn’t come back to it for a few months. It was no problem to pick it up and carry on. It’s semi-cookbook structure lends it to that.
If you liked Freakonomics, or any book of that sort, this is that with more depth and breadth.
The seven-model examination of the causes of wealth/income inequality is extremely enlightening and worth the price of the book. It takes only 11 pages. It will give you a very different perspective and is worth reading, even if you have to get the book from the library or read it at your local bookstore. Starts on page 343! Not one mention of politics.
Displaying 1 - 30 of 93 reviews