I.H.60.Warren Buffett.Wrong Incentives

错误影响人的行为方式。

1、《2015-05-02 Berkshire Hathaway Annual Meeting》

31. Lessons on hidden incentives from Teledyne’s Henry Singleton

来自Teledyne的亨利·辛格尔顿的隐性激励课程

WARREN BUFFETT: OK. Station 7.

沃伦·巴菲特：好的。第七站。

AUDIENCE MEMBER: Hi. My name is Dan Hutner from Vermont.

观众成员：嗨。我叫丹·哈特纳，来自佛蒙特州。

I was wondering if you could talk about Henry Singleton’s Teledyne, and whether you learned lessons from that, used it as a model, and what you think about how it ultimately unwound, and how you might want Berkshire to continue differently?

我想知道你是否可以谈谈亨利·辛格尔顿的Teledyne，以及你是否从中学到了经验教训，是否将其作为模型，以及你对它最终解散的看法，以及你希望伯克希尔以不同方式继续发展的想法？

WARREN BUFFETT: Yeah. That’s a very good question. And Charlie knew Henry Singleton. I knew a lot about him. I mean, I studied him very carefully, but Charlie knew him personally, as well as studying him.

沃伦·巴菲特：是的。这是个很好的问题。查理认识亨利·辛顿。我对他了解很多。我是说，我非常仔细地研究了他，但查理不仅研究了他，还与他有过个人交往。

So I’m going to let Charlie answer that. But there’s a lot to be learned from both what Singleton did in his operating years and then what happened subsequently.

所以我让查理来回答这个问题。但从辛格尔顿在他运营期间所做的事情以及随后发生的事情中，有很多可以学习的地方。

CHARLIE MUNGER: Henry Singleton was very interesting. He was a lot smarter than either Warren or I.

查理·芒格：亨利·辛顿非常有趣。他比沃伦和我都聪明得多。

Henry was the kind of guy that always got 800 on every test and left early. And he could play chess blindfolded, at just below the grand master level, when he was an old man.

亨利是那种每次考试都能得到 800 分并且提前离开的人。作为一个老人，他能在盲棋下接近大师级别的水平。

That said, I watched him invest, and I watched Warren invest, and Warren did a lot better. He just worked at it.

话虽如此，我看着他投资，也看着沃伦投资，沃伦的表现要好得多。他只是更加努力。

Henry was thinking about inertial guidance, and Warren was thinking about securities. And the extra work enabled Warren to get by with his horrible deficit of IQ, compared to Henry. (Laughter)

亨利在考虑惯性导航，而沃伦在考虑证券。额外的工作使沃伦弥补了他与亨利相比的智商劣势。（笑声）

And the interesting thing —

有趣的是——

WARREN BUFFETT: But let’s not quantify it. (Laughter)

沃伦·巴菲特：但我们不要量化它。（笑声）

CHARLIE MUNGER: No. The interesting thing about Singleton is he had very clever incentives on all the key executives, and they were tough, and they were important, and they were meaningful.

查理·芒格：不。有趣的是，辛格尔顿对所有关键高管设定了非常聪明的激励措施，这些激励措施很严格、很重要，并且意义重大。

And in the end, he had three different Defense Departments that got into scandals.

最后，他有三个不同的国防部卷入了丑闻。

He wasn’t doing anything wrong. He wasn’t trying to hurt the Defense Department on purpose. But the incentives got so strong, and the culture of performance got so strong, that people actually — it went too far — in dealing with the government, Teledyne did.

他并没有做错什么。他并不是故意想要伤害国防部。但激励措施变得如此强烈，绩效文化也变得如此强烈，以至于人们实际上——这走得太远了——在与政府打交道时，泰雷德公司确实如此。

And so, we haven’t had any trouble like that, that I know of. Can you think of any, Warren?

所以，我知道我们没有遇到过这样的麻烦。你能想到有什么吗，沃伦？

WARREN BUFFETT: No. And Charlie and I, we really believe in the power of incentives. And there’s these hidden incentives that we try to avoid.

沃伦·巴菲特：没有。查理和我确实相信激励的力量。而且我们试图避免这些隐藏的激励。

One — we have seen, both of us, more than once, really decent people misbehave because they felt that there was a loyalty to their CEO to present certain numbers — to deliver certain numbers — because the CEO went out and made a lot of forecasts about what the company would earn.

首先——我们都曾多次看到，一些真正正直的人因为觉得对他们的CEO有忠诚，要呈现某些数字——交付某些数字——而出现不当行为，因为CEO出去做了很多关于公司收入的预测。

And if you — if you go and say — if I were to say that Berkshire’s going to earn X per share next year, and we have a bunch of executives in the insurance business that set loss reserves and do all kinds of things, or companies in other areas that can load up channels at the end of quarters, at the end of years, I’ve seen a lot of misbehavior that actually doesn’t profit anybody financially, but it’s been done merely because they don’t want to make the CEO look bad, in terms of his forecast. Or he’s done it, because he doesn’t want to look —

如果你——如果你说——如果我说伯克希尔明年每股收益将达到X，我们在保险业务中有一群高管负责设定损失准备金和做各种事情，或者其他领域的公司可以在季度末、年末囤积渠道，我见过很多不当行为，实际上没有给任何人带来经济利益，但只是因为他们不想让CEO在预测方面显得很糟糕。或者他这样做，因为他不想显得——

When they get their ego involved, people do things that they shouldn’t do.

当人们把自我牵扯进来时，他们会做一些不该做的事情。

So we try to eliminate incentives that would cause people to misbehave, not only for financial rewards, but for, you know, ego satisfaction.

所以我们努力消除那些会导致人们不当行为的激励，不仅是为了经济奖励，还有为了自我满足。

I think that’s probably pretty unusual to even be considering that in the business, but we’ve seen enough, so we do consider it.

我认为在这个行业考虑这个可能是相当不寻常的，但我们已经看到了足够的情况，所以我们确实考虑它。

CHARLIE MUNGER: I might also report that at the end, Henry wanted to sell his business to Berkshire for stock, so he was very smart right to the very end.

查理·芒格：我还想报告一下，在最后，亨利想将他的业务以股票出售给伯克希尔，所以他一直非常聪明。

WARREN BUFFETT: We had a case at National — it’s interesting.

沃伦·巴菲特：我们在National公司有一个案例——这很有趣。

You really have to understand — should understand — human behavior, if you’re going to run a business, because when National Indemnity — we’re going back to the late 1960s —

你真的必须理解——应该理解——人类行为，如果你打算经营一家公司，因为当国家赔偿公司——我们回到 20 世纪 60 年代末——

Jack Ringwalt was a marvelous man, and he ran it, and he had another marvelous man who worked for him, his tennis partner, and that fellow was in charge of claims.

杰克·林格沃特是一个了不起的人，他负责管理一切，还有另一个了不起的人为他工作，他的网球搭档，那位先生负责索赔。

And when the claims man would come in to Jack and say, “I just received a claim for $25,000” or something, for some long-haul truck or something, Jack would say — Jack — it was just his personality.

当理赔人员来找杰克说，“我刚收到一个25000美元的索赔”或其他类似的事情，杰克会说——杰克——这只是他的个性。

He would start berating the fellow and say, “How could you do this to me?” and “These claims are killing me,” and all of that, and he was joking.

他会开始责备那个家伙，说，“你怎么能这样对我？”以及“这些索赔正在毁了我，”所有这些，他是在开玩笑。

But the fellow he was joking with couldn’t take it, really, and he started hiding claims. And he just didn’t — he stuck them in a drawer.

但那个他开玩笑的家伙真的不能接受，他开始隐藏索赔。他只是——把它们放在一个抽屉里。

And that caused us to not only misreport fairly minor figures, but it also caused us to misinform our reinsurers, because they had an interest in the size of claims.

这导致我们不仅错误报告了一些相对较小的数字，还误导了我们的再保险公司，因为他们对索赔的规模有兴趣。

And the fellow that was hiding the claims had no financial interest in doing it at all, but he just didn’t like to walk into the office and have Jack kid him about the fact that he was failing him.

那个隐藏索赔的人根本没有任何经济利益，但他就是不想走进办公室，让杰克嘲笑他失败的事实。

And you really have to be very careful in the messages you send as a CEO. And if you tell your — if you tell your managers you never want to disappoint Wall Street, and you want to report X per share, you may find that they start fudging figures to protect your predictions.

作为首席执行官，您在发送的信息中必须非常小心。如果您告诉您的经理，您永远不想让华尔街失望，并且希望报告每股 X 的收益，您可能会发现他们开始篡改数据以保护您的预测。

And we try to avoid all that kind of behavior at Berkshire. We’ve just seen too much trouble with it. (Applause)

我们尽量避免在伯克希尔出现这种行为。我们已经见识到太多麻烦了。（掌声）

2、《2017-05-08 张一鸣-如何应对公司变大之后的管理挑战》

第三，弱化层级跟title。

我们鼓励年轻人多提想法。我第一次担任CEO是26岁，我相信我们公司26岁的人有很好的实践经验，受过很好的教育，只要给他们好的Context，他们也能做出好的决策。为了避免形式感给基层节点带来压制，我们弱化层级，首先是不允许这种称呼——“老大”、“某某总”、“老师”，这种称呼一旦出来之后，很多想法就不能涌现出来了。他们可能会倾向于先听听“老师”有什么意见，自己不能先说出来。我们没有title带来的日常可见的待遇区别，比如什么样的人配备什么样的电脑，什么样的人配备什么样的办公桌，这样也会带来层级感，也会影响不同的同事发表意见。

3、《2023-03-27 Ilya Sutskever.Why Next-Token Prediction Could Surpass Human Intelligence》

Ilya Sutskever

伊利亚·苏茨克维尔

I think dedicated training will get us there. More and more improvements to the base models will get us there. But fundamentally I also don't feel like they're that bad at multi-step reasoning. I actually think that they are bad at mental multi-step reasoning when they are not allowed to think out loud. But when they are allowed to think out loud, they're quite good. And I expect this to improve significantly, both with better models and with special training.

我认为专门的训练可以实现这一目标。对基础模型的不断改进也会帮助我们达到目标。但从根本上说，我并不认为它们在多步推理方面表现得那么差。我实际上认为，它们在不被允许“出声思考”时才表现得不好。但当它们被允许“出声思考”时，表现相当不错。我预计，随着模型的改进和专门训练，这方面会有显著提升。

I.H.60.Warren Buffett.Wrong Incentives

I.H.60.Warren Buffett.Wrong Incentives

1、《2015-05-02 Berkshire Hathaway Annual Meeting》

2、《2017-05-08 张一鸣-如何应对公司变大之后的管理挑战》

3、《2023-03-27 Ilya Sutskever.Why Next-Token Prediction Could Surpass Human Intelligence》

Article Comments Update

热门标签

Related Articles

1988-04-11 The Inside Story of Warren Buffett (Fortune, 1988)

Warren Buffett Quotes

1994-03-01 Warren Buffett's Letters to Berkshire Shareholders

2023-05-03 WARREN BUFFETT GUIDE TO INVESTING-FROM CNBC

1965-07-09 Warren Buffett's Letters to Limited Partners