How to predict the future | Book Insights on The Signal and the Noise by Nate Silver
发布时间 2023-02-15 23:00:33 来源
摘要
*Our brains are machines that are always simplifying and approximating, focusing on some detail that captures our attention. Those details may be important, but we are only a brain within a vast universe, and our biases often lead us to bad predictions.
*Drawing on his groundbreaking forecasting work, Nate Silver surveys the field of prediction and discusses how to identify a meaningful signal in amongst a barrage of data.
*Only forecasting that arises from many points of data, both quantitative and qualitative, and that tries to take account of all biases, is likely to be correct.
*The only way to become better at forecasts, Silver says, is to make more of them.
*Anyone in the field of prediction has to be constantly taking in not just new points of data, but getting more and more rigorous about how they assess the data. Otherwise, you start to lose the signals for the noise.
Theme 1: Nate Silver and his Highly Successful Forecast Approach - 0:29
Theme 2: Poker, Baseball & Forecasting Things With Complex Causes and Inputs - 12:12
Theme 3: Examples Where Prediction Starts to Break Down - 22:25
Like what you hear? Be sure to like & subscribe to support this podcast! Also leave a comment and let us know your thoughts on the episode.
You can also get a free weekly email about the Book Insight of the week. Subscribe at memod.com/insights
Want quick save-able, share-able bullet points on this book? Check out the Memo: https://memod.com/MrBusiness/nate-silver-on-the-art-of-a-good-prediction-186
HEAR THE FULL INTERVIEWS MENTIONED IN TODAYS' EPISODE HERE:
YouTube. (2019). The Signal and the Noise - Nate Silver. [online] Available at: https://youtu.be/z4zhI9uLs4U?t=382.
Full Title: The Signal and the Noise: Why So Many Predictions Fail - But Some Don't
Year of Publication: 2012
Book Author: Nate Silver
To purchase the complete edition of this book click here: https://tinyurl.com/bdf3vevz
Book Insight Writer: Carl Dister
Editor: Tom Butler-Bowden
Producer: Gabe Mara
Production Manager: Karin Richey
Curator: Tom Butler-Bowden
Narrator: Kristi Burns
GPT-4正在为你翻译摘要中......
中英文字稿 ![](/static/new-window.png)
You're listening to Book Insights, brought to you by Memode, finding and simplifying the world's most powerful ideas to fit into your lifestyle. Each episode is a deep dive into a non-fiction bestseller that can change your life or make you think. In around 30 minutes, you'll learn all about a book that offers wisdom for your life, career, or business. So get ready to live and work smarter, better, and happier with Book Insights.
你正在聆听《书籍洞见》,由Memode带给你,它发掘并简化了世界上最强大的思想,以适应你的生活方式。每一期节目都是对一本非小说畅销书的深入探讨,可以改变你的生活或让你思考。在约30分钟内,你将学到一本为你的生活、职业或事业提供智慧的书。因此,准备好用《书籍洞见》来更聪明、更好、更快乐地生活和工作吧。
Johannes Gutenberg invented the printing press in 1440. Europe was producing only a few thousand books a year then. But a century later, around two million books were being produced annually and their cost plummeted. The dispersal of knowledge helped spark the industrial revolution and set the stage for the European enlightenment. But the printing revolution had other effects. Society became filled with inflammatory religious tracks that ramped up heresy. Sudo-scientific titles were popular, and errors in original texts were now multiplied by the thousands. A little knowledge became a dangerous thing. Pamphlets like Martin Luther's famous 95 Theses got printed over and over, and sewed the seeds for centuries of religious war. Masks did the mass printing of Bibles. On the plus side, thinkers like Galileo were able to spread their ideas beyond the reach of authority. The works of Shakespeare could be enjoyed by millions. And it's doubtful that the French and American revolutions would have happened without the fast dissemination of ideas that printing brought.
在1440年,约翰内斯·古腾堡发明了印刷机。当时欧洲每年只生产几千本书。但一个世纪后,每年约生产了两百万本书,它们的成本也急剧下降。知识的散布帮助点燃了工业革命,并为欧洲启蒙时代铺平了道路。但印刷革命产生了其他的影响。社会变得充满了煽动性的宗教刊物,加剧了异端邪说。伪科学的书名很受欢迎,原始文本中的错误现在被成千上万地繁殖。一点知识变成了一个危险的东西。像马丁·路德的著名的《九十五条论纲》这样的小册子被反复地印刷,为数百年的宗教战争埋下了伏笔。马斯克斯负责大量印刷圣经。好的一面是像伽利略这样的思想家能够将他们的思想传播到当局的掌控范围之外。莎士比亚的作品可以被数百万人欣赏。而且如果没有印刷带来的观念传播的快速,很难想象法国和美国的革命会发生。
The printing press also brought a huge amount of trivial information into the mind of the average person. There was now a ton more noise to sift through. Today's printing press is, of course, the internet. More data is produced each year than in the entirety of human history. Does this data automatically translate into greater wisdom and better decisions? Probably not. As Nate Silver argues in his 2012 bestseller, The Signal and the Noise, increases in information actually make it harder to identify what's important. Just because we have more data, it doesn't mean we can make better forecasts of where a terrorist attack may happen, or when an economy will go into recession. The more points of information there is, the more crucial it becomes to have filters and to be aware of our own biases. Humans are expert at seeing patterns where none exist.
印刷机还给人们的大脑带来了大量琐碎信息。现在有更多的噪音需要筛选。当然,现在的印刷机是互联网。每年产生的数据比人类历史上的总和还要多。这些数据是否自动转化为更大的智慧和更好的决策呢?可能不是。正如内特·西尔弗在他的2012年畅销书《信息和噪声》中所辩,信息的增加实际上使得辨别重要的东西更加困难。仅仅因为我们有更多的数据,并不能意味着我们可以更好地预测恐怖袭击可能发生的地点或经济何时会陷入衰退。信息越多,就越需要有过滤器和认识到我们自己的偏见。人类擅长在不存在任何东西的情况下看到模式。
The subtitle of The Signal and the Noise is, The Art and Science of Prediction. Successful prediction is an art because it requires human imagination and context. It's a science because it involves the objective use of statistics as much as possible. Why is prediction important? Because it's not just the realm of forecasters, we predict outcomes every day and make decisions based on them. This can be deciding on a second date or assessing the economy because it impacts on whether we take out a mortgage. There's such a thing as objective truth and we're all trying to get closer to it. Failure to see truth has big costs.
《信号与噪音》的副标题为《预测的艺术与科学》。成功的预测是一门艺术,因为需要人类的想象力和背景知识。它也是一门科学,因为需要尽可能客观地运用统计学方法。为什么预测很重要?因为它不仅仅是预报员的领域,我们每天都在预测事物的结果并根据它们做出决定。这可能是决定是否进行第二次约会,还是评估经济因为它影响我们是否要贷款买房。客观真理是存在的,我们都在努力接近它。如果看不到真相会带来巨大的代价。
The Signal and the Noise is a fascinating journey into the field of prediction making across several fields. In this book insight, we'll look more in-depth at Nate Silver and his highly successful forecasting approach. Second, see how his approach is applied in poker, baseball and politics. Third, learn about forecasting things with very complex causes and inputs such as housing crises and hurricanes. And finally, we'll discover how some areas are so complex that no methods are able to accurately predict outcomes. Examples include stock markets, earthquakes, climate change and terrorism.
《信号与噪音》是一次迷人的旅程,涉及到多个领域的预测制作。在这本书的深入洞察中,我们将更深入地了解Nate Silver及其高度成功的预测方法。其次,了解他的方法如何应用于扑克、棒球和政治。第三,学习如何预测诸如住房危机和飓风等输入因素非常复杂的事物。最后,我们将发现一些领域如此复杂,以至于没有方法能够准确地预测结果。例如股市、地震、气候变化和恐怖主义。
Silver was born in 1978 and grew up near Detroit. After completing an economics degree at the University of Chicago in 2000, he got a job with a countancy firm KPMG. Board by the work, he started developing a baseball prediction program while his boss wasn't looking. His PCODA or Player Empirical Comparison and Optimization Test Algorithm turned out to be an excellent system for identifying great players. He sold it to baseball prospectus, a baseball website and prediction company, and continued to manage PCODA for a few years. Then, from 2004 to 2005, Silver used his talents in a very different field. While still in his 20s, his skill at online poker netted him $400,000.
Silver 在1978年出生,并在底特律附近长大。2000年在芝加哥大学获得经济学学位后,他在会计公司KPMG找到了一份工作。由于工作毫无乐趣,他开始开发一个棒球预测程序,而老板并不知情。他的PCODA或球员经验比较与优化测试算法成为识别优秀球员的卓越系统。他将其卖给了棒球预测网站和公司棒球前景,并继续管理它几年。然后,在2004年至2005年期间,Silver 在一个非常不同的领域运用了他的才华。在还年轻的时候,他在在线扑克上获得了40万美元的成果。
Silver started getting more interested in politics and political forecasting and began building another prediction model. He gained fame when, in the 2008 US election, he famously called the result of 49 of 50 states. Actually, he didn't provide firm calls but probabilities. Still, the feat won him a place on Time Magazine's World's 100 Most Influential People List the following year. He did the same thing in the 2012 election and his website, 538.com, became legendary for its forecast and analysis. Though Silver did not predict Trump's victory in 2016, he did give him a much higher chance of victory than other pollsters and he was criticized for that forecast leading up to Election Day.
Silver越来越对政治和政治预测感兴趣,并开始建立另一个预测模型。2008年美国总统选举时,他以猜中了49个州的结果而名声鹊起。实际上,他没有做出确定的预测,而是给出了可能性。尽管如此,这一壮举在隔年让他获得了《时代》杂志评选的全球100位最具影响力的人物之一。他在2012年的选举中也做到了同样的事情,他的网站538.com因其预测和分析而成为了传奇。虽然Silver没有预测2016年川普的胜利,但他给出的胜利概率比其他民调专家要高得多,因此在选举日前他受到了批评。
So what Silver's secret? Actually, there is no magic to it. It's a combination of numbers crunching and taking into account tiny details for which you build up a picture of trends and likelihoods.
那么,Silver的秘密是什么呢?实际上,它并没有什么神奇的地方。它是数字计算和考虑微小细节的结合,在此基础上构建趋势和可能性的图片。
Before reviewing each area of prediction in the book, let's first summarize the first five aspects of Silver's method. The first thing you have to identify your biases. We are naturally overconfident in our knowledge. Silver admits that it's hard for any of us to recognize how much our relatively narrow range of experience can color our interpretation of the evidence.
在审阅本书中的每个预测领域之前,让我们先总结一下Silver方法的前五个方面。首先,你必须确定你的偏见。我们的知识天生自信。Silver承认任何人都难以认识到我们相对狭窄的经验范围如何影响我们对证据的解释。
Humans are awesome at pattern recognition but in our information age, this creates a big problem. Because the more data we have, the greater the likelihood that we'll read stories into things. Therefore, before we demand more of our data, Silver says, we need to demand more of ourselves.
人类在模式识别上非常出色,但在信息时代,这会带来很大的问题。由于我们拥有的数据越多,就越有可能在事物中看到故事。因此,Silver说,在我们要求更多数据之前,我们需要先要求自己更多。
The second aspect of better prediction is simply to get better at the odds. We face two main sources of uncertainty. One is the quality of the information we have. The other is our assessment of its meaning. A good clue that we don't know something, in other words, that we are biased, is when we think the probability of something happening is either 0% or 100%.
更好预测的第二个方面就是要更好地处理可能性。我们面临两个主要的不确定性来源。其中一个是我们所拥有的信息质量。另一个是我们对信息含义的评估。当我们认为某个事件发生的概率为0%或100%时,这表明我们并不了解事情的真相,也就是存在偏见。
There are a few absolute certainties, and the law of averages means that even things with very long odds routinely happen. Just because we know what's happened in the past and have data in the present, it isn't a guarantee that our odds will be accurate. Therefore, the better we understand how probability itself works, the more accurate our predictions can become.
有一些绝对的确定性,并且平均数法则意味着即使极低几率的事情也常常发生。仅仅因为我们知道过去发生了什么并且现在有数据,这并不保证我们的几率会准确。因此,我们越了解概率本身的工作原理,我们的预测就越准确。
Here is Silver himself speaking with the Royal Society in London. When I said, for example, Obama, say on September 1st, was a 70-30 favorite. A lot of people thought that he was sure to win. It's like, no, I meant, you know, Ramley is going to win 310 times if you could somehow randomize the world from that point forward. That's quite a lot. If you had woke up in the morning with a 3-10 chance of being stabbed, you wouldn't really want to take those chances. Take that quite seriously.
这里是Silver本人在伦敦皇家学会演讲。比如我说,奥巴马在9月1日时是70%的赢家,很多人认为他一定会赢。其实是这样的,如果你能从那一刻开始随机世界,那么Ramley赢的概率是310次。这已经非常高了。如果你醒来发现自己有3-10的可能被刺伤,你肯定不会愿意冒那个险。所以我们应该认真对待这个问题。
Silver's third approach of forecasting is to understand methods of handling uncertainty. Forecasters need a baseline of what they know and some certainty in this knowledge. This becomes the foundation of better prediction. Silver's chief inspiration is the 18th century English minister statistician and philosopher Thomas Bayes, of the famed Bayes theorem.
Silver 的第三种预测方法是理解处理不确定性的方法。预测者需要了解自己知道的一些基本原理,以及对这些知识有一定的把握。这是提高预测能力的基础。Silver 的主要灵感来源于18世纪的英国部长、统计学家和哲学家托马斯·贝叶斯,他提出了著名的贝叶斯定理。
Bayes believed that the best way to get to the truth of an idea involved admitting your assumptions, then changing your understanding appropriately when you receive new information. This approach, however, requires some understanding of the basic scientific structure behind the area being predicted or understood.
Bayes认为,了解一个想法的真相的最佳方法涉及承认自己的假设,然后在收到新信息时适当地改变自己的理解。然而,这种方法需要了解所预测或理解的领域背后的基本科学结构。
The problem with many expert predictions is that the expert's knowledge of their field comes with too many unarticulated assumptions or ideological baggage. It is only the experts who fully admit these and actually look at the data in front of them. Silver says, who can make accurate predictions.
许多专家预测的问题在于,他们对自己领域的知识包含太多未明确说明的假设或意识形态的包袱。只有那些完全承认这些假设并实际看待面前数据的专家才能进行准确的预测,就像Silver所说的那样。
The fourth aspect of forecasting is to test your predictions. Without this step, you can't learn and get more accurate. The prediction could be tested in chunks. For example, at first you may not be able to predict a hurricane, but you can start by predicting a change of temperature, then seeing if that change occurs. At the other extreme, there can be over-zealous testing.
预测的第四个方面是测试你的预测。如果没有这个步骤,你无法学习并更准确。可以将预测分块测试。例如,起初你可能无法预测飓风,但你可以先预测温度变化,然后看看这种变化是否发生。在另一个极端,可能会过度测试。
Silver points the finger at 20th century philosopher of science, Karl Popper. For Popper, no hypothesis was scientific unless you could falsify it. This strictness, however, tends to reduce imagination. The forecaster needs to entertain all possibilities, not just prove what is wrong. For instance, a popperian approach would never have led to a prediction of the Japanese attack on US ships at Pearl Harbor or 9-11.
Silver 指责 20 世纪科学哲学家卡尔·波普尔。对于波普尔来说,除非你能证明一个假设是错误的,否则它就不是科学的。然而,这种严苛性往往会削弱想象力。预测者需要考虑所有可能性,而不仅仅是证明什么是错的。例如,波普尔主义的方法永远不会导致预测日本对珍珠港的袭击或 9/11。
Finally, Silver says that the sensible forecaster realizes that their personal predictions or data are never the full story. To get closer to the truth, they must aggregate all available data. Aggregating multiple forecasts into a single one is a humble and effective technique. Silver never claims to be a forecasting savant or genius, and in fact, stresses repeatedly that good prediction is a shared enterprise.
最后,Silver表示明智的预测者意识到,他们个人的预测或数据永远不是全部,为更接近真相,他们必须整合所有可用数据。将多个预测聚合成一个是一种谦虚而有效的技术。Silver从未声称自己是预测天才,事实上,他反复强调好的预测是共同努力的结果。
Much of his political forecasting success has come from pulling together all available polls and making sense of them. Combined forecasts such as these, he says, nearly always outperform individual forecasts.
他成功的政治预测很大程度上是通过整合所有可获得的民意调查数据并理解它们。他说,像这样的综合预测几乎总是优于个人预测。
In this part, we began our exploration into statistician and writer Nate Silver's bestseller, The Signal and the Noise. We learned about Silver as a forecaster of baseball statistics before starting 538.com, a political and sports analysis website. We've also learned the five key aspects to Silver's forecasting method. Identify your biases, get better at the odds, understand methods of handling uncertainty, test your predictions, and aggregate all available data.
在这部分,我们开始探索统计学家和作家纳特·席尔弗的畅销书《信号与噪声》。我们了解到,席尔弗在创办政治和体育分析网站538.com之前,是一名棒球统计预测者。我们还学习到了席尔弗预测方法的五个关键方面。识别您的偏见,提高对赔率的认识,了解处理不确定性的方法,测试您的预测,并汇总所有可用的数据。
Next time, we'll learn how Silver applied these approaches in poker, baseball, and politics. Then we'll go into forecasting with complex causes and inputs. Enjoying this episode of Book Insights? If so, keep listening and learning. There's a collection of over 100 titles you can read or listen to now at memodeapp.com slash insights. That's m-e-m-o-d-a-p-p.com slash insights.
下一次,我们会学习Silver在扑克、棒球和政治中如何运用这些方法,然后我们会深入研究复杂原因和输入的预测。喜欢这一集的书籍洞察力吗?如果是这样,请继续听和学习。在memodeapp.com/insights上,现在有超过100本书可以阅读或听取。
In his bestselling book, The Signal and the Noise, statistician and writer Nate Silver noted our brains are machines that are always simplifying and approximating. Focusing on some detail that takes our eye. Those details may be important, but we are only a brain within a vast universe, and our biases often lead us to bad predictions.
在他的畅销书《信号与噪声》中,统计学家和作家内特·席尔瓦指出,我们的大脑是一台总是简化和近似的机器。我们会专注于一些吸引我们注意力的细节。这些细节可能很重要,但我们只是宇宙中微不足道的一部分,我们的偏见经常导致我们做出错误的预测。
Last time in this book insight, we learned about Silver's methods of forecasting. In this part, we'll look at how he applied them to particular areas. Then we'll discuss forecasting with complex systems.
上次在这本书中,我们了解了银的预测方法。在这一部分中,我们将看看他如何将它们应用于特定的领域。然后我们将讨论复杂系统的预测。
Poker
People mistakenly believe that poker is a psychological game. In fact, Silver says it's an incredibly mathematical game that depends on making probability judgments amid uncertainty. Coming up with probabilities involves humility, whereas most players are overconfident. Professional poker players, such as Tom Duan, agree with Silver's method. Duan stresses the importance of having in mind probabilities for each hand played, rather than going on a simple yes or no or gut feel. In general, Silver says that to win money at poker, it's best to be a big fish operating in a small pond. You don't have to be the very best. He advises spending time getting up the learning curve until you are at about 80% there. There is a law of diminishing returns whereby you spend a lot of time getting more accurate or advancing your skill, but not making additional money for the effort. In other words, it's worth being better than a good number of your competitors, but not all of them. Is Silver's insight here something we can all apply to our careers? Again, it's an example of how we are all making constant forecasts, little and big, about our chances of success, which lead to decisions on whether or not to commit resources.
很多人错误地认为扑克是一种心理游戏。事实上,Silver表示它是一种非常数学化的游戏,需要在不确定性中作出概率判断。制定概率需要谦卑,而大多数玩家都过于自信了。像Tom Duan这样的职业扑克玩家赞同Silver的方法。Duan强调必须铭记每手牌的概率,而不是简单的凭感觉决策。总的来说,Silver表示,要在扑克中赚钱,最好是做一个在小池塘里运营的大鱼。你不必是最好的。他建议花时间学习,直到你掌握了大约80%的技能。存在一定的边际效应,你花费大量的时间变得更加准确或提高技能,但不为此获得额外的收益。换句话说,与你的竞争对手相比,做得更好是值得的,但不必赢得所有人。Silver在这里提出的见解是我们都可以应用到职业生涯中的一个例子吗?再次强调,我们都在做出关于成功机会的持续预测,无论大小,这些都会导致是或否投入资源的决策。
Baseball
As mentioned earlier, it was the creation of his predictive baseball spreadsheet, Pacota, that originally got Silver noticed. He gives some idea of how the system worked. For example, even though the nation's baseball scouts did not rate Dustin Pradroya very highly, partly because he was only 5'7", Silver's system singled out the Red Sox second basement for success. The scouts tended to have a preset idea of potential greatness and often made selections based on gut feel. But as legendary baseball analyst Billy Bean, the subject of the film Moneyball pointed out, first impressions of one player can mean that you let a statistically good one through the cracks. You can easily miss the potential of players who don't fit the usual mold.
就像之前提到的那样,创造他的预测棒球电子表格Pacota,是最早让Silver引起注意的地方。他介绍了该系统是如何工作的。例如,尽管全国棒球侦察员没有很高地评价达斯汀·普拉德罗亚,部分原因是他只有5英尺7英寸的身高,Silver的系统却选择了红袜队的二垒手作为成功的对象。侦察员往往对潜在的伟大有着预设的想法,并经常基于直觉做出选择。但正如传奇棒球分析师比利·比恩所指出的那样,在一个球员的第一印象可能意味着你会错过一个统计上很好的球员。你很容易错过那些不符合通常模式的球员的潜力。
Silver's system took the qualitative measures of players into consideration, not just the usual quantitative ones. These include work, ability to concentrate and focus, a drive to compete, ability to humbly manage their stress, and being good learners. Pradroya had all these features in spades and it helped him achieve fame just as Pacota predicted, despite what the scouts had estimated. The last couple of decades have seen a battle between the nerdy statisticians of baseball and the scouts who are often former players. A good forecaster combines the numbers obsession of a baseball nerd with the whileness of a scout, Silver says. He recommends aggregating the analysis and predictions of each. It's a good example of his ability to look at the big picture, but also focus on fundamental details. Predictive innovators typically think very big and they think very small, Silver says.
Silver的系统考虑了球员的定性措施,而不仅仅是通常的定量措施。这些措施包括工作能力,专注力,竞争意识,谦虚地管理压力和成为良好的学习者。Pradroya拥有所有这些特点,并帮助他达到了名声,就像Pacota预测的那样,尽管球探估计出现了偏差。过去几十年,棒球运动中的书呆子统计学家和通常是前球员的球探之间进行了一场竞争。一个好的预测师结合了棒球书呆子的数字痴迷和球探的敏锐度,Silver说。他建议综合每个人的分析和预测。这是他能够看到大局,同时专注于基本细节的能力的很好例子。预测性创新者通常是非常具有远见和细节思维的,Silver说。
Politics
Every country has television shows where a host brings on experts to hash out the latest developments in politics. These shows make for good TV. But as political scientists, Philip Tetlock demonstrated in his 2005 book, Expert Political Judgment. Predictions made by these talking heads have a wouful record. In fact, no better than flipping coins. Part of the problem is that the experts tend to be part of political camps so they cannot objectively assess information. Some even actively make bold predictions just to get noticed in the media. Tetlock found that the only people who were better than average in political forecasting tended to have no ideological position. They were able to gather up many little ideas and take into account uncertainties in their predictions.
每个国家都有电视节目,主持人会邀请专家来探讨最新政治发展。这些节目非常受欢迎。但是,像菲利普·特特洛克在他2005年的书《专家政治判断》中所表明的,这些交谈的专家所做的预测记录相当糟糕。事实上,不比抛硬币好多少。问题的一部分是专家们往往是政治派别的成员,因此他们无法客观地评估信息。一些人甚至积极进行大胆的预测,只是为了在媒体上引起注意。特特洛克发现,在政治预测方面,只有那些没有意识形态立场的人比平均水平更好。他们能够聚集许多小想法,并考虑到他们的预测中的不确定性。
This kind of forecast or silver notes is more like the Fox of the ancient Greek fable, who knows a lot of little things. In contrast to the hedgehog, who knows one big thing. Here is silver speaking with the Royal Society.
这种预测银笺更像古希腊寓言中那只狐狸,它知道许多小事情。相比之下,刺猬只知道一件大事情。现在,这里的银正在与皇家学会交谈。
The Fox personality type tends to be scrappy and scrounged around for different ideas. They're not hidey ideological. They don't have one big theory. The hedgehog would say that I have a big theory. The theory that explains everything, capital E, is a very hedgehog mentality. If you have a pristine theory or what you think is pristine theory, then you have absolute certainty. But in fact, the people who demonstrate that overconfidence actually do a worse job of making predictions.
狐狸型人格倾向于狡猾和四处搜寻不同的想法。他们并不是隐藏意识形态者。他们没有一个大的理论。刺猬会说我有一个大理论。解释一切的理论,大E,是非常刺猬的思维方式。如果你有一个纯净的理论或你认为是纯净的理论,那么你就有绝对的确定性。但实际上,表现出这种过度自信的人做出的预测更糟糕。
The Foxy forecasters are more successful because they are capable of making connections between diverse bits of data to arrive at a measured judgment. This bottom-up approach to prediction is a lot more trustworthy than most political prediction, which is simply the projection of the mind of the so-called expert.
狡猾的预测者之所以更成功,是因为他们能够将不同的数据点联系起来,得出有慎重的判断。这种自下而上的预测方法比大多数政治预测更值得信赖,因为后者不过是所谓专家思想的投影。
Housing crises and the economy Silver reflects on the 2006-2007 housing bubble that preceded the financial crisis. He notes, we ignore the risks that are the hardest to measure, even when they post the greatest threats to our well-being. In other words, it's tempting to simply focus on what we prefer to happen, or that we could at least understand happening.
《住房危机和经济》的作者Silver回顾了2006-2007年的房地产泡沫,那是金融危机之前的一段时期。他指出,我们往往忽视最难以度量的风险,即使它们对我们的福祉构成最大的威胁。换句话说,我们很容易只关注我们喜欢发生的事情,或者至少是我们能理解的事情。
Oftentimes, multiple risks can occur that seem very independent from one another. Various signals taken separately don't seem too worrisome, but when considered as a whole and in retrospect, the outcome is not surprising. The upward movement in house prices in the mid-2000s may not have seemed that dramatic at the time. Yet as part of a 100-year graph of house prices adjusted for inflation, it's clear that there had been a sudden ramping up of values. It should have sent up warning flags that something unrealistic was happening. Samultaneously, home lending had become irresponsible. Defaults were increasing at a pace, and the rating agencies were incorrectly labeling risky subprime mortgage securities as triple-A rated.
很多时候,可能会出现多个风险,这些风险看起来似乎相互独立。单独看各种信号似乎不太令人担忧,但当考虑到整体和回顾时,其结果并不令人意外。20世纪中期房价的上涨可能当时并不显得那么引人注目,但在考虑通货膨胀调整后的房价100年的趋势图中,显然价值已经急剧上升。这应该引起警惕,表明某些不现实的事情正在发生。同时,房屋贷款已经变得不负责任。违约速度在增加,而评级机构错误地将高风险的次级抵押贷款证券评为三A级。
Join the dots on these things, and it's clear the housing boom was a house of cards. But the fact that the economy seemed to be very strong acted like a veil to these deep problems. The group of prominent Federal Reserve economists was saying that there was a one-in-five hundred chance of a recession. They were so taken in by rising asset prices that they were blind to the various signals. Silver attributes some of these negligence to simple fear and greed, which overwhelmed healthy skepticism and judgment.
连接这些事物的点,就可以清楚地看到房地产繁荣是一张纸牌楼房。但经济似乎非常强劲的事实,像一层面纱掩盖了这些深层次的问题。著名的联邦储备经济学家团队声称,出现经济衰退的可能性是500分之一。他们被不断飙升的资产价格迷惑,对各种信号变得视而不见。 Silver将这些疏忽归于简单的恐惧和贪欲,这些良好的怀疑和判断力被淹没了。
A lot of the time, people fail to make accurate predictions because they don't want to see where current conditions are leading.
很多时候,人们无法做出准确的预测,因为他们不愿意看清当前情况所导致的结果。
Hurricanes. Silver is happy to say that the prediction of weather has made dramatic strides. Chaos starts to kick in when prediction is attempted over periods longer than a week, but day-to-day forecasting has become pretty reliable. Case in point? We are 30 times less likely to be struck by lightning today than in 1940, thanks to improvements in forecasting. And in the even more frightening case of hurricanes, we now tend to get at least a couple of days notice before one hits land. Plus, we now know within 100 miles of exactly where it will hit.
Silver很高兴地表示,天气预报取得了巨大的进展。如果预测时间超过一周,混乱就开始蔓延,但日常预报已经相当可靠了。一个例子?由于预测技术的提高,今天我们被雷击的可能性比1940年减少了30倍。即使是更可怕的飓风,现在我们至少能在它袭击陆地前预先得到几天的通知。此外,我们现在能够在100英里内精确地预测它将击中的区域。
When it came to Hurricane Katrina, which he discusses at length, the true disaster was not in forecasting, but in responding. Silver attributes the success of hurricane prediction and weather prediction to three federal agencies, the National Center for Atmospheric Research, the National Oceanic and Atmospheric Administration, and the National Weather Service.
关于飓风卡特里娜,他详细讨论了这个话题,真正的灾难不在于预测,而是在于应对。Silver 将飓风预测和天气预测的成功归功于三个联邦机构:国家大气研究中心、国家海洋和大气管理局以及国家天气局。
The NWS costs each citizen only $3 per year. Such great prediction at low cost is the gold standard for forecasters, Silver says, and should be our model. In general, Silver argues that politics, personal glory, and economic gain are the three sins of forecasting. They have stopped the science and art of prediction from advancing to where it should be.
美国国家气象局每年只需向每个公民收取3美元。银表示,如此低廉的代价而能做出如此出色的预测,是预报员标准的黄金,应该成为我们的典范。总的来说,银认为政治、个人荣誉和经济利益是预测的三大罪孽。它们阻碍了预测科学和艺术的进步,使其无法达到应有的高度。
In this part, we continued our dive into Nate Silver's The Signal and the Noise. We looked into three areas where Silver applied his methods of forecasting. Poker, baseball, and politics. In poker, knowledge of probability and humility serve better than understanding of psychology. In baseball, Silver suggests a combination of money-ball statistics and gut-level shrewdness. In politics, the smartest forecasters take in knowledge from all sides of the political spectrum instead of clinging to one dogma.
在这一部分中,我们继续深入了解Nate Silver的《信号与噪声》。我们探讨了Silver应用他的预测方法的三个领域。扑克、棒球和政治。在扑克中,概率知识和谦虚比心理学的理解更有效。在棒球中,Silver建议通过资金统计和直觉机智的结合来进行。在政治中,最聪明的预测者应该从政治谱系的各个方面了解知识,而不是坚持一种教条。
We also looked into Silver's methods approaching complex inputs and causes. In the housing market, clear minds were clouded by fear and greed rather than looking at rising defaults and falsely labeled risky subprime mortgage securities.
我们还研究了Silver的处理复杂输入和原因的方法。在房地产市场中,人们的清晰思维被恐惧和贪婪所掩盖,而未能注意到默认风险上升和错误标记的高风险次贷抵押证券。
When it comes to weather forecasting, we're more advanced than ever thanks to three federal agencies, the National Center for Atmospheric Research, the National Oceanic and Atmospheric Administration, and the National Weather Service. We simply need to improve our preparedness and responses to disasters.
说起天气预报,得益于三个联邦机构——国家大气研究中心,国家海洋和大气管理局和国家气象服务,我们比以往任何时候都更先进。我们只需要提高应对灾害的准备和响应能力。
Next time, we'll conclude our book insight on the signal and the noise by looking at examples where predictions start to break down. We'll wrap up with a consideration of the book's legacy. Enjoying this episode of book insights? If so, keep listening and learning. There's a collection of over a hundred titles you can read or listen to now at memodeapp.com slash insights. That's memodeapp.com slash insights.
下一次,我们将通过研究预测开始崩溃的例子来结束我们对《信号与噪声》这本书的洞见。最后,我们将考虑这本书的遗产。喜欢这一集书籍洞见吗?如果是这样,请继续听和学习。现在在 memodeapp.com/insights 有一百多个可以阅读或听取的书籍标题收集。
According to statistician and writer Nate Silver, it's human nature to view the admission of uncertainty as weakness, but it's often the most boastful predictions that tend to be wrong. In his best-selling book, The Signal and the Noise, Silver notes that only forecasting which arises from many data points and tries to take account of all biases is likely to be correct. Here is Silver being interviewed by the Royal Society. We are our own constraint, basically. The idea is to build a computer program and solve everything. If you put it in the code, the computer will replicate that silly instruction and will reproduce that bug millions of times over. We should be humble about our own ability to perceive the world, especially complex matters like forecasting earthquakes or the economy or what have you.
根据统计学家和作家Nate Silver所言,把不确定性作为弱点看待是人性的本能,但往往夸张的预测往往是错误的。在他畅销书《信号与噪音》中,Silver指出,只有从许多数据点出发并试图考虑所有偏见的预测可能是正确的。这里是Silver接受皇家学会采访的视频。我们自己是限制自己的因素,基本上。思路是构建一款计算机程序并解决所有问题。如果你将其放入代码,计算机将复制那个愚蠢的指令并会多次重复该错误。我们应该对自己感知世界的能力保持谦虚,尤其是像预测地震或经济等复杂事情。
In this part, we're concluding our book insight into The Signal and the Noise by looking at examples where predictions start to break down. We'll review everything we learned, then look at the wider implications of Silver's work.
在这一部分中,我们将总结我们对《信号与噪音》一书的洞察,着眼于预测开始出现问题的例子。我们将回顾所学内容,然后探讨Silver的工作对社会的更广泛影响。
Stock markets and the economy. The stock market is a classic, complex system, subject to chaos. Despite what people may claim, no one is really capable of predicting movements in the market. Silver makes the point that even if insider information was available to traders, their predictions would be stronger, but not perfect. There are just too many variables happening in real time, and each new piece of information, even if small, changes the picture. The same is true for the economy as a whole. Economists want to be seen as accurate in their forecasting, so they give very specific forecasts on employment, the rate of growth, etc. But at the risk of appearing vague, it's better to present a range of likely probabilities. Economists have historically been way too confident in their forecasts. They're wrong by a significant degree, both by underestimating actual growth in the economy, or by discounting risk to the economy. Economists frequently don't even know a country is in recession. Jan Hatsius, Goldman Sachs chief economist, told Silver, nobody has a clue. It's hugely difficult to forecast the business cycle. Understanding an organism as complex as the economy is very hard.
股市与经济。股市是一个经典的、复杂的系统,容易受到混乱的影响。尽管人们可能会声明,但没有人真正能够预测市场的波动。Silver指出,即使内部消息对交易者来说是可得到的,他们的预测也会更强,但不是完美的。现实中有太多的变数,每一个新的信息,即使很小,都会改变这个局面。整个经济也是如此。经济学家希望自己的预测看起来准确,因此他们会对就业、增长率等进行非常具体的预测。但是在显得含糊的风险下,最好呈现一系列可能性的概率。经济学家历来对他们的预测非常自信。他们在低估经济实际增长或忽视经济风险方面都是错误的。经济学家经常甚至不知道一个国家正在经历经济衰退。高盛首席经济学家Jan Hatsius告诉Silver,没有人有头绪。预测商业周期非常困难。理解一个如此复杂的有机体——经济——是非常困难的。
Earthquakes. In contrast to the weather, the field of earthquake prediction does not fare very well. In 2009, an earthquake in Laquia, Italy, caused 300 deaths. The town had a history of earthquakes, but after a time complacency set in. Why? The span of a human life is so much shorter than geological timeframes, people simply become less worried as time marches on. The sensors measuring the Earth's crustal movements put out millions of readings. It's easy to mistake this noise for a credible signal of an earthquake. Like weather systems and the stock market, earthquakes sit in the realm of great complexity. Silver praises the US geographical survey for coming straight out and stating that they cannot predict the date and time of an earthquake. There is a relationship over long periods of time between the magnitude and the frequency of occurrence of historical earthquakes. But largely, earthquakes are still unpredictable.
地震。与天气不同,地震预测领域表现不佳。2009年,意大利拉奎亚市的一次地震造成了300人死亡。该镇历史上曾发生过地震,但随着时间的推移,人们变得越来越漠不关心。为什么?人类寿命要比地质时期短得多,随着时间的推移,人们简单地变得不那么担心了。测量地壳运动的传感器输出了数百万个读数。很容易将这种噪音误认为是地震的可信信号。像天气系统和股票市场一样,地震处于极其复杂的领域。Silver赞扬美国地理调查局直截了当地声明他们无法预测地震的日期和时间。历史地震的震级和频率之间存在着长时间的关系。但在很大程度上,地震仍然是不可预测的。
Climate change. One of the most crucial predictions facing the planet relates to the future of climate change. Bias is definitely play an important role here. As outlined earlier, it's important to have a baseline in order to incrementally work towards the truth. The International Panel on Climate Change, or IPCC, tries to provide this foundation with the body of facts. First, the greenhouse effect is a solid scientific fact and is what keeps the Earth's temperature in check. Second, emissions from human activities increase greenhouse gases. Third, as water vapor goes up, temperatures tend to rise. But the big question is, what are the outcomes of such facts and to what extent? The wrong way to see climate change, Silver says, is as a free for all, where stakeholders get to cherry pick facts to suit their argument. This is a case where wrong predictions are extremely dangerous. Assessing climate is pretty complex compared to weather forecasting. Factors which come into play include El Niño effects, Sunspot activity, volcanic eruptions, levels of CO2 which cause warming, and levels of sulfur which cool the atmosphere. In short, there is no way to avoid the layers of uncertainty in climate change forecasting. The only thing we can do is keep feeding back new data into our models to get more accurate.
气候变化是一个非常关键的预测,它关乎整个地球的未来。偏见在这里起着重要作用。正如之前提到的那样,为了逐步接近真相,有一个基线是很重要的。国际气候变化专门委员会(IPCC)试图提供基础事实,第一,温室效应是一个坚实的科学事实,它使地球温度保持平衡。第二,人类活动的排放增加了温室气体。第三,随着水蒸气的增加,温度往往会上升。但是重要的问题是,这些事实将导致什么样的结果以及影响范围是多大?错看气候变化的方式是一种无序的选择事实以适应自己论点的方式。这是一种非常危险的错误预测。气候评估相对于天气预报来说相当复杂。涉及到的因素包括厄尔尼诺现象、太阳黑子活动、火山爆发、引起变暖的CO2水平以及冷却大气层的硫水平。简而言之,在气候变化预测中无法避免层层不确定性。我们唯一能做的就是不断将新数据反馈到我们的模型中,以获得更精确的预测。
Terrorism. When it comes to predicting acts of terrorism, it's not so much about seeking physical or statistical signals, but rather understanding and tracking terrorist intentions. With all the noisy information on tips and conflicting leads on potential terrorist attacks, it's no wonder that conspiracy theories abound. Conspiracy theories, Silver says, are an irresistible method of labor saving. They're a hack for those faced with overwhelming complexity but who have an inability to sift through the facts. He says, a conspiracy theory might be thought of as the laziest form of signal analysis. Silver interviewed former US Secretary of Defense during the 9-11 attacks, Donald Rumsfeld. Rumsfeld recounted the history of terrorism, starting with Pearl Harbor, and recalled that even back then there were early signs. The biases clouded everything. The US was expecting some type of sabotage or subterfuge from Japan, but not a huge planned act of war. Similarly, with the 9-11 attacks, the problem was not data or information, but the ability to see the forest for the trees and understand the true scope of the terrorist intentions. There is a deep human bias towards avoiding what is unfamiliar, improbable, and unknown. More often than not, big shocks involve a failure in our imagination. We didn't see it coming for any number of reasons. We didn't want to, or we didn't have the intellectual scope to do so.
在预测恐怖主义行为方面,不是寻找物理或统计信号,而是理解和跟踪恐怖主义者的意图。在关于潜在恐怖袭击的众多提示和矛盾线索的信息杂音中,阴谋论遍地开花也就不足为奇了。阴谋论是一种耗时、耗力的方法。对于那些面对大量信息但无法分辨事实的人来说,它们是一种便利的方法。一种阴谋论可以被认为是最懒惰的信号分析形式。Silver采访了美国前国防部长唐纳德·拉姆斯菲尔德,讲述了恐怖主义的历史,从珍珠港开始,回忆起早期的迹象。偏见影响了一切。美国当时预期日本会进行某种形式的破坏或欺骗,但不会发动一次巨大的战争行动。同样的,对于9/11袭击,问题不在于数据或信息,而在于识别出真正的恐怖主义意图的能力。人们有一种根深蒂固的偏见,对不熟悉、不可能和未知的事物避而不见。往往是我们的想象力出现了问题导致了重大震惊的发生。我们没能预见到任何一种原因,可能是我们不想预见,或者是我们的智力和眼界有限。
A quick recap. We first took a look at the author, Nate Silver, and his highly successful forecasting approach. It involves being aware of your biases, getting better at understanding the probabilities for any event, getting smarter when faced with situations of uncertainty, always testing your predictions, and using the power of aggregation of results. Second, we saw how Silver has applied this approach with success in poker, baseball, and politics. We then learned about the art and science of forecasting when it involves very complex causes and inputs, such as for housing bubbles and hurricanes. Finally, we learned that some areas are so complex that no methods are able to accurately predict outcomes. Examples include stock markets, earthquakes, climate change, and terrorism. With over a thousand carefully cited references and interviews with the most respected names in prediction, it's hard to argue against the ideas in the signal and the noise. That said, after Silver published it, philosopher and author of the Black Swan, Nassim Taleb, publicly challenged Silver's work on technical reasons. He argued that some of the data or points of information that Silver was taking into account shouldn't even be considered, and that it's easy to impute meaning from them after the fact. For Taleb, the only prediction that matters is the one made just before an event, which takes in all available information. In 2016, Silver's failure to predict the outcome of the US election showed that people claiming to be accurate forecasters might just get lucky some of the time. Taleb may be right, but if anyone is able to make predictions that are better than even odds, it's people like Nate Silver. Despite recent controversies, the signal and the noise is an authentic, transparent look at the prediction industry from a practicing expert in the field.
快速回顾一下。我们首先看了作者Nate Silver和他高度成功的预测方法。这涉及到意识到您的偏见,更好地理解任何事件的概率,面对不确定情况时变得更聪明,始终测试您的预测并利用聚合结果的力量。其次,我们看到Silver如何成功地将这种方法应用于扑克,棒球和政治方面。然后,我们了解了在涉及非常复杂的原因和输入(例如房地产泡沫和飓风)的预测艺术和科学。最后,我们了解到有些领域如此复杂,以至于没有任何方法能准确预测结果。例子包括股市,地震,气候变化和恐怖主义。凭借超过一千个仔细引用的参考文献和与预测领域中最受尊重的人物的访谈,很难反驳《信号与噪音》中的思想。话虽如此,自Silver发表该书后,哲学家和《黑天鹅》一书的作者纳西姆·塔勒布公开挑战了Silver的技术论证。他认为,Silver考虑的一些数据或信息点甚至不应该被考虑,只有在事件发生之前考虑到所有可用信息才能进行预测。对于塔勒布来说,唯一有意义的预测是在事件发生前做出的预测。虽然Taleb可能是正确的,但如果有人能够做出比赔率更好的预测,那就是像Nate Silver这样的人。尽管最近有争议,《信号与噪音》是在预测领域的一位实践专家的真实、透明的观察。
We can all borrow something of Silver's approach in making predictions in our own lives, but we must proceed with caution. Although we're in the age of big data, Silver says there's a risk that we become too starry-eyed about what it can do for us. He says, there is no reason to conclude that the affairs of men are becoming more predictable. The opposite may well be true.
我们可以从Silver在预测方面的方法中借鉴一些东西,应用到自己的生活中,但必须小心谨慎。尽管我们处于大数据时代,Silver表示存在一种风险,即我们可能对它所能为我们提供的东西过于迷恋。他说,没有理由断定人们的事务变得更加可预测。相反,可能恰恰相反。
The same science is that uncover the laws of nature are making the organization of society more complex. He says the book isn't so much about what we know, but the gap between what we think we know and what we actually do. Closing that gap means adopting a Bayesian approach to prediction under uncertainty.
同样的科学揭示自然规律,使社会组织更加复杂。他说这本书不仅仅关于我们所知道的,更关乎我们认为自己知道什么与实际掌握的知识之间的鸿沟。缩小这个鸿沟意味着采用贝叶斯方法来预测不确定性下的情况。
This means being a bit more humble and thinking in probabilities given our many strong biases. It's human nature to view the admission of uncertainty as weakness, but as we've learned, it's often the most prideful predictions that tend to be wrong. Only forecasting that arises from many points of data, both quantitative and qualitative, and that tries to take account of all biases, has any chance of being correct.
这意味着要更加谦虚,考虑到我们很多的强烈偏见,算出可能性。将不确定性作为一种脆弱的表现是人类的本性,但我们已经学到,往往是最自大的预测最容易错。只有从许多数据点来临,既定量又定性,并试图考虑所有的偏见,才有可能正确地预测。
According to Silver, the only way to become better at forecasts is to make more of them. It's the only way you can test your hypothesis of how the world works. This doesn't mean refining your models so that they become more elegant, but doing constant real world tests, in the same way that companies like Google and Facebook do A-B testing. Silver says, forget models.
Silver认为,成为更好的预测者的唯一方法就是做更多的预测。这是您测试关于世界运作方式假设的唯一方法。这并不意味着您需要改进您的模型,使其变得更加优雅,而是需要进行持续的真实世界测试,就像Google和Facebook等公司进行的A-B测试一样。Silver说,要忘记模型。
Constant testing of our own hypothesis is the only way we ever get closer to truth. Thank you for listening to Book Insights. Check out the rest of our content at memo.com. Please keep in mind that the information provided in or through our Book Insights episodes is for educational and informational purposes only.
不断测试我们自己的假设是接近真理的唯一方式。谢谢您收听 Book Insights。请在 memo.com 上查看我们其余的内容。请记住,在我们的 Book Insights 集数中提供的信息仅用于教育和信息目的。
It's not intended to be a substitute for advice given by qualified professionals and should not be relied upon to disregard or delay seeking professional advice.
这并不是要替代来自有资格专业人士的建议,也不应该依赖它来忽视或延误寻求专业建议。