Meeting Diversified Reality And Multi Model Thinking Era

Zheng Lei / Wen

Danny Roderick, a famous economist, has put forward a view that for practical problems, it is best to use several small models at the same time and know which models to use in what circumstances.

This view applies to any other practical problems as well, because the world we live in is complex and diverse. A model may be too simplistic and ignores other important factors and perspectives. However, too complex models can not be solved by human beings. Even if they are processed by high-speed computers, most of them can not be analyzed and explained.

The fact that mastering information can bring a high economic and social status has not changed yet, but the premise must be the ability to make use of massive information. From information to knowledge, a processing process is needed before rising to intelligence, that is, the ability to recognize and apply relevant knowledge, including the selection of correct knowledge and models.

With the progress of technology, with the help of artificial intelligence and big data technology, we seem to be able to make some judgments and predictions with deep learning methods, such as the "Alpha dog" who can defeat the highest human level player, and the robot who is good at examination. However, this method can not generate new knowledge, and its algorithm itself is only relying on data processing to find out the mode, but not the way. Know how to interpret the results.

Models can make our thinking more effective. In other words, we humans have to go back to the thinking model. We should master several small models according to their work and life needs. There is a book that can really satisfy this purpose, which is Scott's model of thinking. This book is based on a very hot online course. The author originally wanted to name "32 models that make you a genius". As of this book, the number of online courses has exceeded 1 million. This book has also been revised and perfected many times, and the thirtieth draft is published. Its value is not difficult to see.

The thinking model. Data map

Scott page divides the model into three categories: simplification of the world, mathematical probability analogy, and artificial exploratory model. "Let's take a look at it." When our thinking is consistent with a number of logically consistent and validated frameworks, we are more likely to make wise choices. This is the "grille" built by many knowledgeable Munger, who has worked with Buffett for many years. The diversity of models will lead to more accurate conclusions, similar to what we call "three cobbler over Zhu Geliang". Of course, there is a premise that everyone in the group has their own views and can not express themselves in the same way. The 32 models introduced in this book come from many disciplines and lead us to think in a logical way.

When I was teaching economics graduate students, I found that students had two main defects. One was lack of problem oriented consciousness, the other was that they did not know how to choose and use models. The resonance in these two aspects has led to the lack of academic research ability. To face a real problem, we need information, knowledge and wisdom to get the correct answer. And from processing information to getting answers, we must use suitable models.

Choosing and using models is not an easy task. An astonishing view comes from the philosopher Wittgenstein that "all models are wrong." The correct understanding of this sentence is that all models are simplified and are established only under certain conditions. But simplicity is not the same as simplification, and we can use multiple models at the same time to achieve the intersection of multiple possible situations, so as to overcome the narrowness of the single model. Furthermore, the model should be chosen according to the conditions of use. The model is special, and there are few models that are well suited to various situations and problems.

Scott page describes the use of the model as REDCAPE, which is the 7 major feature. I think that the most worthy of discussion is reasoning, interpretation and prediction. If a model can't help us do these three things, it will not be an appropriate model. French mathematician and astronomer Leville used Newton's law to explain the orbit of Uranus. He thought it could be explained only when there was another big planet on the periphery. He passed the advice to the Berlin Observatory, and it turned out Neptune in his predicted position in 5 days. An example of the important role of prediction is that in June 1, 2009, the plane crashed into the the Atlantic incident. In the absence of all kinds of detection technologies, the French national statistical analysis bureau used probability model to analyze ocean currents, and identified a small area most likely to be sunken by the plane. It took only a week to find the wreckage.

Finally, the exchange purpose of the model is emphasized. People attach great importance to the positive role of education in social progress, but underestimate the role of communication. To a large extent, economic growth depends on the transmissible nature of knowledge (usually relying on models). There is ample evidence that the transmission of ideas can contribute more to economic growth than education does. Back to the model introduced in this book, I think readers need to learn and master normal distribution, power-law distribution, network model, entropy, path dependence, random walk, signal model and learning model, not more than ten most commonly used models.

?