In his 1939 discussion on Tinbergen's method in econometrics, Keynes comes back to his 1921 Treatise on Probability's question. Is statistics or econometrics a mere description of reality or is it a probable inference (prediction) and what are the assumptions implied passing from description to prediction. Keynes's judgement is that Tinbergen's method is fallacious and inappropriate to obtain a reasonable probable inference. Note that for Keynes, a reasonable probable inference does not mean a successful prevision. In a Treatise on Probability, Keynes quotes Herodotus, saying that if one make a decision foolishly and obtain something by mere luck (by mere success), this does not make the decision less foolish.
This article is food for thought. Starting from Keynes's question in his Treatise on Probability 1921, whether statistics provides a mere description of events or a probable inference (a prediction for the future) and whether one can logically (without making logical fallacies) pass from statistical description to probable inference (a prediction), it arrives at the modern use of statistics in AI, machine learning, Gpt algorithm and so on. In his Treatise on Probability, Keynes discusses mere enumeration, positive and negative analogy if they can justify probable inferences.
In a recent and very successful book, Noise: A Flaw in Human Judgment, by Daniel Kahneman (the 2002 Nobel laureate in economics), Olivier Sibony and Cass Sunstein, published in 2021 (Kahneman, Sibony, and Sunstein 2021), they suggest how to avoid bias and noise in crucial decisions in legal sentences, medical diagnoses, job selection, insurance evaluations, and various other situations. They focus on the randomness of individual impacts on people from judgments.. Avoiding flaws and errors in judgments is important to reach a “good” judgment—a remedy to the power of «poor» judgments. In the book, they raise the following question: “Could the wisdom of crowds offer a solution to this noise?” where the wisdom of the crowd is obtained by algorithms searching, by data mining, the most statistically recurrent among similar decisions. At the end of the book, they argue for the superiority, for example, of Google algorithms and artificial intelligence in job selection, in comparison to the biased and “noised” decisions and evaluations by human selectors. They show that decision-makers are very often too biased and “noised,” and that society needs a noise audit. Keynes would have answered the question raised arguing that if you know that x is John Maynard and if you know him, even partially, your own evaluation would be reasonable and probably right or good (for Keynes, a reasonable evaluation is neither true nor eventually successful): the probability of an unknown individual posting a letter undressed can be based on the statistics of the Post Office, but my expectation that I shall act thus, cannot be so determined (TP, 450) Keynes would also have suggested avoiding relying on the wisdom of the crowd—because, in his view, the crowd is as ignorant as you are. In the case of ignorance, you should ask for independent expert advice. In his view, independent judgments and unconventional decisions (i.e., not following rules, routines, or the crowd) are far better for the markets and for society as a whole. An opinion is also partly shared by the three authors in their conclusions. With this book on Noise, however, we are back to the old dichotomy, rules versus discretion, well-known to economists in monetary policy, e.g., the alternative between following rules and taking ad hoc decisions by the Central Bank. Monetarists like Milton Friedman and the orthodox Bundesbank argue for a rule-based monetary policy, while Keynesians would prefer ad hoc decisions, even if they might turn out to be poordecisions. If ad hoc decisions were reasonable and probably right and good, they would be the best decisions. The 2008–2013 great financial crisis and the 2020 Covid emergency have taught economists that, in crucial situations and dilemmas, monetary policy should be discretionary. Discretion ranges from Mario Draghi’s «whatever it takes» to the most unconventional monetary policies of injecting enormous amounts of liquidity into markets.
In a recent and very successful book, Noise: A Flaw in Human Judgment, by Daniel Kahneman (the 2002 Nobel laureate in economics), Olivier Sibony and Cass Sunstein, published in 2021 (Kahneman, Sibony, and Sunstein 2021), they suggest how to avoid bias and noise in crucial decisions in legal sentences, medical diagnoses, job selection, insurance evaluations, and various other situations. They focus on the randomness of individual impacts on people from judgments.17 Avoiding flaws and errors in judgments is important to reach a “good” judgment—a remedy to the power of «poor» judgments. In the book, they raise the following question: “Could the wisdom of crowds offer a solution to this noise?” where the wisdom of the crowd is obtained by algorithms searching, by data mining, the most statistically recurrent among similar decisions. At the end of the book, they argue for the superiority, for example, of Google algorithms and artificial intelligence in job selection, in comparison to the biased and “noised” decisions and evaluations by human selectors. They show that decision-makers are very often too biased and “noised,” and that society needs a noise audit. Keynes would have answered the question raised arguing that if you know that x is John Maynard and if you know him, even partially, your own evaluation would be reasonable and probably right or good (for Keynes, a reasonable evaluation is neither true nor eventually successful): the probability of an unknown individual posting a letter undressed can be based on the statistics of the Post Office, but my expectation that I shall act thus, cannot be so determined (TP, 450) Keynes would also have suggested avoiding relying on the wisdom of the crowd—because, in his view, the crowd is as ignorant as you are. In the case of ignorance, you should ask for independent expert advice. In his view, independent judgments and unconventional decisions (i.e., not following rules, routines, or the crowd) are far better for the markets and for society as a whole. An opinion is also partly shared by the three authors in their conclusions. With this book on Noise, however, we are back to the old dichotomy, rules versus discretion, well-known to economists in monetary policy, e.g., the alternative between following rules and taking ad hoc decisions by the Central Bank. Monetarists like Milton Friedman and the orthodox Bundesbank argue for a rule-based monetary policy, while Keynesians would prefer ad hoc decisions, even if they might turn out to be poor decisions. If ad hoc decisions were reasonable and probably right and good, they would be the best decisions. The 2008–2013 great financial crisis and the 2020 Covid emergency have taught economists that, in crucial situations and dilemmas, monetary policy should be discretionary. Discretion ranges from Mario Draghi’s «whatever it takes» to the most unconventional monetary policies of injecting enormous amounts of liquidity into markets.
https://www.linkedin.com/posts/anna-maria-carabelli-6386029b_goals-and-methods-in-statistics-machine-activity-7087912696058142721-SQNZ?utm_source=share&utm_medium=member_android
In his 1939 discussion on Tinbergen's method in econometrics, Keynes comes back to his 1921 Treatise on Probability's question. Is statistics or econometrics a mere description of reality or is it a probable inference (prediction) and what are the assumptions implied passing from description to prediction. Keynes's judgement is that Tinbergen's method is fallacious and inappropriate to obtain a reasonable probable inference. Note that for Keynes, a reasonable probable inference does not mean a successful prevision. In a Treatise on Probability, Keynes quotes Herodotus, saying that if one make a decision foolishly and obtain something by mere luck (by mere success), this does not make the decision less foolish.
This article is food for thought. Starting from Keynes's question in his Treatise on Probability 1921, whether statistics provides a mere description of events or a probable inference (a prediction for the future) and whether one can logically (without making logical fallacies) pass from statistical description to probable inference (a prediction), it arrives at the modern use of statistics in AI, machine learning, Gpt algorithm and so on. In his Treatise on Probability, Keynes discusses mere enumeration, positive and negative analogy if they can justify probable inferences.
On the current use of statistics in Ai.
In a recent and very successful book, Noise: A Flaw in Human Judgment, by Daniel Kahneman (the 2002 Nobel laureate in economics), Olivier Sibony and Cass Sunstein, published in 2021 (Kahneman, Sibony, and Sunstein 2021), they suggest how to avoid bias and noise in crucial decisions in legal sentences, medical diagnoses, job selection, insurance evaluations, and various other situations. They focus on the randomness of individual impacts on people from judgments.. Avoiding flaws and errors in judgments is important to reach a “good” judgment—a remedy to the power of «poor» judgments. In the book, they raise the following question: “Could the wisdom of crowds offer a solution to this noise?” where the wisdom of the crowd is obtained by algorithms searching, by data mining, the most statistically recurrent among similar decisions. At the end of the book, they argue for the superiority, for example, of Google algorithms and artificial intelligence in job selection, in comparison to the biased and “noised” decisions and evaluations by human selectors. They show that decision-makers are very often too biased and “noised,” and that society needs a noise audit. Keynes would have answered the question raised arguing that if you know that x is John Maynard and if you know him, even partially, your own evaluation would be reasonable and probably right or good (for Keynes, a reasonable evaluation is neither true nor eventually successful): the probability of an unknown individual posting a letter undressed can be based on the statistics of the Post Office, but my expectation that I shall act thus, cannot be so determined (TP, 450) Keynes would also have suggested avoiding relying on the wisdom of the crowd—because, in his view, the crowd is as ignorant as you are. In the case of ignorance, you should ask for independent expert advice. In his view, independent judgments and unconventional decisions (i.e., not following rules, routines, or the crowd) are far better for the markets and for society as a whole. An opinion is also partly shared by the three authors in their conclusions. With this book on Noise, however, we are back to the old dichotomy, rules versus discretion, well-known to economists in monetary policy, e.g., the alternative between following rules and taking ad hoc decisions by the Central Bank. Monetarists like Milton Friedman and the orthodox Bundesbank argue for a rule-based monetary policy, while Keynesians would prefer ad hoc decisions, even if they might turn out to be poordecisions. If ad hoc decisions were reasonable and probably right and good, they would be the best decisions. The 2008–2013 great financial crisis and the 2020 Covid emergency have taught economists that, in crucial situations and dilemmas, monetary policy should be discretionary. Discretion ranges from Mario Draghi’s «whatever it takes» to the most unconventional monetary policies of injecting enormous amounts of liquidity into markets.
In a recent and very successful book, Noise: A Flaw in Human Judgment, by Daniel Kahneman (the 2002 Nobel laureate in economics), Olivier Sibony and Cass Sunstein, published in 2021 (Kahneman, Sibony, and Sunstein 2021), they suggest how to avoid bias and noise in crucial decisions in legal sentences, medical diagnoses, job selection, insurance evaluations, and various other situations. They focus on the randomness of individual impacts on people from judgments.17 Avoiding flaws and errors in judgments is important to reach a “good” judgment—a remedy to the power of «poor» judgments. In the book, they raise the following question: “Could the wisdom of crowds offer a solution to this noise?” where the wisdom of the crowd is obtained by algorithms searching, by data mining, the most statistically recurrent among similar decisions. At the end of the book, they argue for the superiority, for example, of Google algorithms and artificial intelligence in job selection, in comparison to the biased and “noised” decisions and evaluations by human selectors. They show that decision-makers are very often too biased and “noised,” and that society needs a noise audit. Keynes would have answered the question raised arguing that if you know that x is John Maynard and if you know him, even partially, your own evaluation would be reasonable and probably right or good (for Keynes, a reasonable evaluation is neither true nor eventually successful): the probability of an unknown individual posting a letter undressed can be based on the statistics of the Post Office, but my expectation that I shall act thus, cannot be so determined (TP, 450) Keynes would also have suggested avoiding relying on the wisdom of the crowd—because, in his view, the crowd is as ignorant as you are. In the case of ignorance, you should ask for independent expert advice. In his view, independent judgments and unconventional decisions (i.e., not following rules, routines, or the crowd) are far better for the markets and for society as a whole. An opinion is also partly shared by the three authors in their conclusions. With this book on Noise, however, we are back to the old dichotomy, rules versus discretion, well-known to economists in monetary policy, e.g., the alternative between following rules and taking ad hoc decisions by the Central Bank. Monetarists like Milton Friedman and the orthodox Bundesbank argue for a rule-based monetary policy, while Keynesians would prefer ad hoc decisions, even if they might turn out to be poor decisions. If ad hoc decisions were reasonable and probably right and good, they would be the best decisions. The 2008–2013 great financial crisis and the 2020 Covid emergency have taught economists that, in crucial situations and dilemmas, monetary policy should be discretionary. Discretion ranges from Mario Draghi’s «whatever it takes» to the most unconventional monetary policies of injecting enormous amounts of liquidity into markets.