Data analytics are increasingly defining our world. In some systems, that's wise and wonderful. In others, it laces catastrophic risk into a complex world that's more Calvinball than Moneyball.
I was just working on an essay with this premise, but sharing yours will have to do for now. It just goes to show that you can't predict what will be on someone else's mind.
This essay made me become a paid subscriber :) I like the puzzle vs mystery distinction. Taleb talks about Extremistan vs Mediocristan & I think the Santa Fe folks insist on the distinction between complicated vs complex.
Calvinball is the template that I've been unknowingly using when I talk about international relations with friends. I took an intro course in I.R., so I know next to nothing about the subject. However, I boil down all mystery moves countries make by saying that countries act in whatever way they think is in their best interest. Correct or not, I now have a shorthand term I can through out there to back up my presumed expertise. I fear these discussions will not end well. Thanks Brian for the essay.
I did go through Radical Uncertainty, which you recommended to me (thanks). I quite enjoyed it and it crystallized issues I have had about the predictive power of models for a long time. Even though my scientific speciality was numerical modelling, I came to appreciate that, in the famous George Box phrase, 'all models are wrong but some are useful'. I was annoyed though by the authors' insistence on narratives as ways to tackle mysteries, without defining clearly what that looks like. Narratives could be viewed as just so stories that we tell each other, as we are wont to do as human beings. To be fair, they do mention resilience and experimentation as strategies to ground these narratives with evidence. In the end, I much prefer the way you put forward resilience and experimentation as the key responses to mysteries. Stories are important, but as ways by which we work through the mystery and come up with hypotheses to experiment with and then also to explain the results to ourselves and others. We cannot underplay the role of science, even with mysteries, as it remains the best way I know to augment our flawed cognitive apparatus and steer it towards what works.
Yes, I think the fairest critique of Radical Uncertainty is that it doesn’t provide a lot of guidance for what to do when facing moments of radical uncertainty but you have to decide. That’s why I tried to cover some of that in Fluke. You can’t just say “Hey, look! Radical uncertainty!” when there’s a mystery illness killing someone. You still have to make a decision. That’s different from modelling what the world will look like in 2050, where it’s not necessary and often isn’t useful for anything other than a thought experiment.
I think one thing to touch on is the realm in between Calvin ball and moneyball. And that would be mathematical modeling. An attempt to take seemingly unpredictable systems and apply mathematical methodologies in order to recognize ordered structures within systems that previously weren't recognized. A classic example is mathematical oncology which is attempting to determine predictive structures within the development of transformed malignant cells. An international leader of this is Robert Gatenby and his team of mathematical oncologists at the Moffitt Cancer Center in Tampa, Florida who has developed new forms of cancer treatment called adaptive therapy that are predicated on predictive models of cellular responses from a data driven process. At extremely high leveld of mathematics that are way beyond me and developed by an entire team of mathematicians, not statisticians. I bring it up only to show that there is actually an in between area that represents systems that are yet to be recognized as puzzles and are still treated as mysteries. I agree completely about the distinctions that you make, but there are in fact some mysteries that are puzzles, and conversely there are puzzles that we don't recognize as being mysteries. This differentiation is critical in determining the most appropriate approaches to problems.
Yes, I agree with this. In my view, most medical issues are puzzles, not mysteries, even if we don’t understand them. Part of the reason for that is that human bodies are far more stable and interchangeable systems than, say, forecasting a political system in 80 years. Not understanding something doesn’t mean it isn’t understandable. Whereas true mysteries…are simply impossible to solve. No matter the data or computing power, we will never be able to forecast what the world will be like, with precision, ten years from now. (I wrote about this at greater length and with more nuance in the chapter on uncertainty and probabilities in Fluke).
I would respectfully disagree at least in part. I think that physiology, pathophysiology, and medicine represent a hybrid of puzzles and mysteries. To my mind, sentience, intelligence and thought are likely to remain mysteries. And some of the things we call puzzles are actually mysteries. I think there has been an oversimplification in medicine where dominant themes like genetics and receptors are viewed as simple systems that are puzzles when in fact they are so complex and intertwined that they are in fact mysteries. This explains why there are so many extraneous side effects to therapies that presumably arget single entities like protein receptors, but end up affecting multiple systems throughout the entire body in ways that we really don't understand. Perhaps they are just very complex puzzles, but I have my doubts.
One of the clear-cut examples is our fixation on genetics. As few people realize, genes only code for the proteins in our bodies and from the point of view of information theory, our genes contain exactly 1% of cellular information for any given cell. No one knows where the other 99% of information is stored, for all the non-protein elements like membranous structures of both the cell and organelles, cytoskeletal structures, cytoplasm, etc. There isn't even a theory of where that information is stored. I believe we think we know a lot more than we do and there is tremendous hubris in many of our therapeutic approaches. But I guess this is a limited digression to your whole point about moneyball, which I entirely agree with.
An affirmation of the idea “all models are wrong, but some are useful” and the appropriate corollary, “all analogies are wrong in their detail, but some are useful in their higher level application.”
Sometimes the “money balling” of data can also lead us to be “more rigorously imprecise”
Some experts believe that quantum computers (quantum simulation) will become powerful enough to model the most complex molecules in the human body and a mystery will eventually become a puzzle – perhaps. There are a 100 billion nerve cells in our brains and each can connect with more than 100 others so more theoretical connections than atoms in the universe – thankfully (most) human behaviour is always going to be a mystery. Best listen to the Blackbirds while you can.
I was just working on an essay with this premise, but sharing yours will have to do for now. It just goes to show that you can't predict what will be on someone else's mind.
This essay made me become a paid subscriber :) I like the puzzle vs mystery distinction. Taleb talks about Extremistan vs Mediocristan & I think the Santa Fe folks insist on the distinction between complicated vs complex.
Calvinball is the template that I've been unknowingly using when I talk about international relations with friends. I took an intro course in I.R., so I know next to nothing about the subject. However, I boil down all mystery moves countries make by saying that countries act in whatever way they think is in their best interest. Correct or not, I now have a shorthand term I can through out there to back up my presumed expertise. I fear these discussions will not end well. Thanks Brian for the essay.
I did go through Radical Uncertainty, which you recommended to me (thanks). I quite enjoyed it and it crystallized issues I have had about the predictive power of models for a long time. Even though my scientific speciality was numerical modelling, I came to appreciate that, in the famous George Box phrase, 'all models are wrong but some are useful'. I was annoyed though by the authors' insistence on narratives as ways to tackle mysteries, without defining clearly what that looks like. Narratives could be viewed as just so stories that we tell each other, as we are wont to do as human beings. To be fair, they do mention resilience and experimentation as strategies to ground these narratives with evidence. In the end, I much prefer the way you put forward resilience and experimentation as the key responses to mysteries. Stories are important, but as ways by which we work through the mystery and come up with hypotheses to experiment with and then also to explain the results to ourselves and others. We cannot underplay the role of science, even with mysteries, as it remains the best way I know to augment our flawed cognitive apparatus and steer it towards what works.
Yes, I think the fairest critique of Radical Uncertainty is that it doesn’t provide a lot of guidance for what to do when facing moments of radical uncertainty but you have to decide. That’s why I tried to cover some of that in Fluke. You can’t just say “Hey, look! Radical uncertainty!” when there’s a mystery illness killing someone. You still have to make a decision. That’s different from modelling what the world will look like in 2050, where it’s not necessary and often isn’t useful for anything other than a thought experiment.
I think one thing to touch on is the realm in between Calvin ball and moneyball. And that would be mathematical modeling. An attempt to take seemingly unpredictable systems and apply mathematical methodologies in order to recognize ordered structures within systems that previously weren't recognized. A classic example is mathematical oncology which is attempting to determine predictive structures within the development of transformed malignant cells. An international leader of this is Robert Gatenby and his team of mathematical oncologists at the Moffitt Cancer Center in Tampa, Florida who has developed new forms of cancer treatment called adaptive therapy that are predicated on predictive models of cellular responses from a data driven process. At extremely high leveld of mathematics that are way beyond me and developed by an entire team of mathematicians, not statisticians. I bring it up only to show that there is actually an in between area that represents systems that are yet to be recognized as puzzles and are still treated as mysteries. I agree completely about the distinctions that you make, but there are in fact some mysteries that are puzzles, and conversely there are puzzles that we don't recognize as being mysteries. This differentiation is critical in determining the most appropriate approaches to problems.
Yes, I agree with this. In my view, most medical issues are puzzles, not mysteries, even if we don’t understand them. Part of the reason for that is that human bodies are far more stable and interchangeable systems than, say, forecasting a political system in 80 years. Not understanding something doesn’t mean it isn’t understandable. Whereas true mysteries…are simply impossible to solve. No matter the data or computing power, we will never be able to forecast what the world will be like, with precision, ten years from now. (I wrote about this at greater length and with more nuance in the chapter on uncertainty and probabilities in Fluke).
I would respectfully disagree at least in part. I think that physiology, pathophysiology, and medicine represent a hybrid of puzzles and mysteries. To my mind, sentience, intelligence and thought are likely to remain mysteries. And some of the things we call puzzles are actually mysteries. I think there has been an oversimplification in medicine where dominant themes like genetics and receptors are viewed as simple systems that are puzzles when in fact they are so complex and intertwined that they are in fact mysteries. This explains why there are so many extraneous side effects to therapies that presumably arget single entities like protein receptors, but end up affecting multiple systems throughout the entire body in ways that we really don't understand. Perhaps they are just very complex puzzles, but I have my doubts.
I recently read a cool book entitled “Wicked Problems.“ Reading this stuff, reminds me of it.
One of the clear-cut examples is our fixation on genetics. As few people realize, genes only code for the proteins in our bodies and from the point of view of information theory, our genes contain exactly 1% of cellular information for any given cell. No one knows where the other 99% of information is stored, for all the non-protein elements like membranous structures of both the cell and organelles, cytoskeletal structures, cytoplasm, etc. There isn't even a theory of where that information is stored. I believe we think we know a lot more than we do and there is tremendous hubris in many of our therapeutic approaches. But I guess this is a limited digression to your whole point about moneyball, which I entirely agree with.
An affirmation of the idea “all models are wrong, but some are useful” and the appropriate corollary, “all analogies are wrong in their detail, but some are useful in their higher level application.”
Sometimes the “money balling” of data can also lead us to be “more rigorously imprecise”
There is a fascinating film about McNamara where he talks about some of the mistakes he made:
https://en.wikipedia.org/wiki/The_Fog_of_War
I strongly recommend it.
Some experts believe that quantum computers (quantum simulation) will become powerful enough to model the most complex molecules in the human body and a mystery will eventually become a puzzle – perhaps. There are a 100 billion nerve cells in our brains and each can connect with more than 100 others so more theoretical connections than atoms in the universe – thankfully (most) human behaviour is always going to be a mystery. Best listen to the Blackbirds while you can.
brian-excellent essay. thank you very much! keep going! we follow.
Well, the real existing socialism was also very much data driven… Do I have to say anything more.