CHAPTER 6
SYSTEMIZING OUR IDEA MERITOCRACY:
The more I did the research on people, the clearer it became that there are different types of
people and that, by and large, the same types of people in the same types of
circumstances are going to produce the same types of results. Said differently,
by knowing what someone is like we can have a pretty good idea of what we can
expect from them. So I was more motivated than ever to continue gathering lots
of data on what people are like to build pointillist pictures of them to help
us match people to responsibilities well. Doing this in an evidence-based way
would enhance the idea-meritocratic process of aligning people’s
responsibilities with their merits.
While this all seemed soclear and commonsensical to me, it was much harder to achieve in practice.About a year into my transition, I saw that many new managers (and some older
ones) still couldn’t see the patterns of people’s behaviors through time (in
other words, they couldn’t connect the dots between what people are like and
the outcomes they produce). Their reluctance to probe hard to get at what
people are like was making things more difficult.But then I had a breakthrough, which grew out of an observation that the challenges we were having with making management decisions didn’t exist in our investment decision making. I realized that, by using big data analytics and other algorithms, ouR computers could connect those dots more efficiently than any of us could, just
as they had helped us make connections in the markets. These systems also
didn’t have personal biases and emotional barriers to overcome, so those being
analyzed couldn’t be offended by the data-driven conclusions the computers were
coming up with. In fact, they could look at the data and algorithms, assess
them for themselves, and suggest changes if they wanted. We were like
scientists trying to develop tests and algorithms for analyzing ourselves
objectively.
On November 10, 2012, I shared my thoughts with the Management Committee in an email. Its subject line was “The Path Out: Systemizing Good Management”:
It is now clear to me that the main difference behind why the investment management part
of Bridgewater is likely to continue to do well and most of the other parts of Bridgewater are unlikely to do as well (if we don’t change how we are operating) is that the decision-making processes for investment management have been so systemized that it’s hard for people to screw them up (because they are largely following the systems’ instructions) while the other areas of
Bridgewater are much more dependent on the quality of the people and their decision making.
Think about that. Imagine how Bridgewater’s investment decision making would work if it
operated the same as Bridgewater’s management decision making (i.e., dependent
on the people we hired and how they collectively made decisions in their own
ways). It would be a mess.
The way the investment decision-making process works is that a small group of investment
managers who created these systems see the systems’ conclusions and the
reasoning of the systems while we make our own conclusions and explore our
reasoning on our own. . . . The machine does most of the work and we interact with it in a quality way. . . . [And] we are not dependent on much more faulty people.
Think about how different management is. While we have principles, we don’t have
decision-making systems. In other words, I believe that the investment decision-making process is effective because the investment principles have been put into decision rules that make
decisions that people then follow while the management decision-making process
is less effective because the management principles have not been put into
decision rules that people can follow to make management decisions.
It doesn’t have to be that way. Having built the investment systems (with the help of
others) and knowing about both investment decision making and management
decision making, I am confident that it can be the same. The only questions are
whether it can happen fast enough and what will happen in the meantime.
I am working with Greg (and others) to develop these management systems in the same way I
worked with Greg and others (Bob, etc.) on the investment systems. You are
seeing this happen via the development of the Baseball Cards, Dot Collector,
Pain Button, testing, job specing, etc. Because I have a limited time to do
this, we need to move fast. At the same time we will have to fight the battles in
the trenches, with hand-to-hand combat, to clean out those who are incapable
and bring in or promote those who are excellent.
One of the great things about algorithmic decision making is that it focuses people on cause-effect
relationships and, in that way, helps foster a real idea meritocracy. When
everyone can see the criteria the algorithms use and have a hand in developing
them, they can all agree that the system is fair and trust the computer to look
at the evidence, make the right assessments about people, and assign them the
right authorities. The algorithms are essentially principles in action on a
continuous basis.
While our managementsystem has a long way to go before it is as well automated as our investmentsystem, the tools it has made possible, especially the “Dot Collector” (an appthat gathers information about people in real time described in detail in theWork Principles), have already made an incredible differencein the way we work.
All these tools reinforce good habits and good thinking. The good habits come from thinking
repeatedly in a principled way, like learning to speak a language. The good
thinking comes from exploring the reasoning behind the principles.
The ultimate goal of all this was to help the people I cared about be more successful without me, which was becoming increasingly pressing as life’s milestones continued to remind me
of my stage in life. For example, I became a grandfather with the birth of
Christopher Dalio on May 31, 2013. And in the summer of 2013, I had a serious
health scare that turned out to be nothing but reminded me of my mortality. At
the same time, I still loved playing the markets, which I plan to do until I
die, making me even more eager to speed the transition from the second to the
third phase of my life.
译文：
随着研究加深，愈加明显显示存在不同类型的人，并在很大程度上，同样类型的人处于同样类型圈子最终也将制造同样类型的结果。换种说法，如果知道某人看起来像是我们能得到一个相当不错的主意-我们期待获得，那么我们就能得到。所以我有更多的动机继续在能帮助我们匹配责任方面更好的人画像而收集大量数据。基于证据基础的方法将强化主意优化的过程—将人们的优点与责任匹配。
所有这一切对我来说似乎是很明显和常识一般，然而在实践中却非常困难。大约在过度期一年的时间，我看到很多新管理人（也有一些老经理人）任然不能通过时间看清人们的行为模式（换句话说，他们不能把人们的表象和实际产出联系起来）。他们固有的缺点使得他们很难明白表象却使事情更加困难。
但是不久我就取得了一次突破，这次突破来自于一次观察，那些我们让他们做管理决策的挑战者们做出了出乎意外的决策，我意识到借助大数据分析和其他算法，我们的计算机系统可以更加有效的连接这些点，比如他们能帮助我们在市场中做出链接。这些系统还没有个人偏见和情绪障碍要客服，所以那些即成的分析不能由后续的数据驱动的计算机得出。实际上，他们能自己发现数据和算法，自己评估数据和算法，并按需建议变化。我们就像科学家尝试开发测试和算法去更加客观的分析我们自己。2012年11月10日，我将关于管理委员会的想法在邮件中分享。主题是“出路是：系统化优秀管理”。
为什么桥水基金的投资管理部门会继续运作良好，而其他部门则不那么好（如果我们不改变我们运作的方式）是因为投资管理部门的决策过程已经如此的系统化以至于人们很难把事情搞糟（因为决策过程绝大部分遵循系统指引）而其他部门则依赖于人的素质和他们的个人决策，现在对我来说已经很清晰了，影藏在这之后的主要区别存在于哪里。
试想桥水基金的投资决策制定方式会起作用，如果桥水管理决策制定（即依赖于我们雇佣的人和他们集体决策的方式）与投资决策一样会怎样？可能是一团糟。
投资决策系统的工作原理是一小群创造这些系统的投资管理人看系统的结论并推理系统的原因，同时我们得出自己的结论并探讨我们自身的理由。。。机器做了大部分工作，而我们和机器以高质量的方式互动。。并且我们并没有比那些犯错的人更多的依靠。
想想管理之间是如何的不同。而我们有原则，我们没有决策制定系统。
另一方面，我相信投资决策制定过程是高效的因为投资原则已经纳入了决策规则当中，这使得人们只要接受决策就好了，而管理决策制定过程少一些效率因为管理原则尚未纳入那些人们可以遵循做出管理决定的决策规则。
本不必那样。既然已经建立了投资系统（借助其他的帮助）并了解投资决策制定和管理决策制定两者，我非常有信心两者可以是一样的。唯一的问题是能否及时的产生和同时会发生什么。
我和乔治一起工作（还有其他人）研发这些管理系统-用我们在投资系统曾经用过的同样方式。你可以发现这在棒球卡系统、点收集、痛苦按钮，测试，工作短波等的研发过程中也出现过。因为我只有有限的时间做这些，我们需要快一些。同时我们还要在战壕里继续战斗，徒手战斗，清理出那些不称职的，并引进或者提升那些优秀的人。
关于算法决定决策制定最伟大之一是执着于因果关系，通过哪种方式帮助培养一种真实的精英主义理念。当所有人都能明白算法使用的标准和插手研发算法的标准。人们都认为系统是公平的，并且只要看看证据就会信任计算机，关于人们总能做出正确的评估，恰当的分配权限。基于连续的基础算法是基础原则。
当然我们的管理系统还有很长的路要走才能和我们的投资系统一样优秀，已有的工具使这一切成为可能，尤其是“点收集器”（一款app能实时收集人们的信息-在工作原则中的细节），已经与我们工作的方式有了不可置信的分别。
所有这些工具都加强了好的习惯和优良的思考能力。好的习惯来自于原则性的反复思考，就像学习一门语言。好的思考来自于探究原则背后的因素。
所有这一切的终极目标是去帮助那些我关心的人取得更大的成功，在没有我的情况下，这些都越来越迫使我就像生命中的那些里程碑在持续的提醒我每个阶段。举例，我在2013年5月31日随着克里斯托弗的诞生成为爷爷。就在2013年夏天，我遭受了一次严重的健康危机，虽然最后证明没事，但还是提醒我死亡的威胁。同时我任然热爱市场博弈，而我计划是博弈到死的，也使我更加渴望去加速从我的生命的第二阶段过度到第三阶段。
