Focus on the most valuable things! ——Sharing the work experience of the chief scientist of Amazon Cloud Technology

This article was first account 161e8f360b8851 in May 2021 by , Chief Scientist of Amazon Cloud Technology at 161e8f360b884e

Five years ago today I flew to Seattle for an interview with Amazon. After the interview, I took a red-eye flight to Boston overnight to attend my wife's doctoral defense at MIT. Halfway through the defense, the phone rang, and the other party said congratulations on passing the interview and wanted to talk about salary. I said that in fact, I will meet your family, give it directly, and hang up first.

The next day after the defense, he and his wife went to the Municipal Council to register their marriage. When I was queuing at the entrance of the oath hall, the boss called and was very excited to say that you can do this and that when you come. I said it was fine, but I had to get married first. The boss was stunned for a moment, said congratulations, and continued on. I had to interrupt: Gotta go first, it's our turn to go in and take the oath.

Five years passed in the blink of an eye. From the outside, the biggest change is that there are two more babies. But the biggest change comes from cognition, which is the change of outlook on life, world outlook and values.

When I graduated from my doctorate, I wrote about my experience - "Dr. Five Years"

（https://zhuanlan.zhihu.com/p/25099638）

Many students commented that they were very encouraged. Now I would like to share with you the same experience and insights from the past five years of work. Rather, it is a failure lesson, because every point is a lesson learned after paying the tuition. Hope these can also be helpful and inspiring to everyone.

The value of things is the value to society

When you study, you have clear goals, such as test scores, further schools, or getting a good job. The biggest difference after work is that you have so many goals to pursue.

The change this brings is that you need to decide what to do now, what to do later, and what not to do.

Prioritization should be based on the value of things. I now assess the value of a thing is its value to society, in formulas it is

Number of beneficiaries x per capita time x unit time value difference

The number of people who can benefit from an event and the per capita time benefiting from it are attributes of the event itself. The third depends on how well you do it, that is, you do better than others do similar things, so that the beneficiaries benefit more from you than from others.

This formula can be used for a variety of different things, and we'll be using it over and over again. Here are a few examples. For example, great products are generally of great value. Take WeChat as an example, it is a mobile communication software for billions of mobile phone users, each of whom uses it for several hours a day, so the first two items of its value are very large. Because the user experience of WeChat is very good, the value it brings to users is poorer than the user experience of other alternatives. So WeChat is a very valuable product.

For a small example, say you take someone to write a paper. The number of people affected by a paper is the size of the research field; the time of impact is the time someone else spends doing research related to your work, which may be a few months in a lifetime; the difference in value is the contribution of your research on the previous work. . In this way, you need to do hot field and cross-generation papers to get high value. But we know that a paper generally does not contribute much, and only a few people will read it, so it is basically worthless. Why do people still actively "water"?

Here we have to take a closer look at two values: First, you are familiar with a new field or new method through this research, which has learning value for you personally. Second, you bring people to do research, which can improve their ability to think of methods, do experiments, and write papers, which is of great value to them. So even a well-known researcher's name will appear on many papers that are the same as the novice's work.

Here's another smaller example. I've spent more time with my baby than work for the past four years. For quite a while, I felt that my career was delayed by my baby. Until later, I used this formula to calculate: Although the number of people is only two, the benefit time is quite high, more than 50 hours a week. And there is only one father. There is a huge difference between children with me and without me (self-perception), so the last two items of the value formula are very high. In addition, having a baby is also valuable to me personally, including how to understand others who have completely different ways of thinking, and always fight against my urge to get angry, and finally achieve a state of Buddha nature. That's how I feel at ease.

Serve the society and finally serve oneself

The value formula above emphasizes the contribution of things to others. Before using it, my value formula was more focused on itself. For example, I often use the fun level of a thing, or the technical content in it to prioritize. The problem is that although I enjoy the process of doing things, the sense of accomplishment afterward is not high. Kind of like the emptiness after playing the game. Because after doing it, I often find that although this is cool, it is useless, and not many people pay attention. The reason is that the things that have direct high value to the individual are not necessarily of great value to others. It is very likely that this matter itself is only meaningful to a small group of people, and it may be that each person benefits for a short time, or it is actually a repetitive wheel, and there are already similar substitutes on the market.

If you optimize the value to society, you get a delayed reward for yourself. This reward includes the higher level of inner satisfaction you bring when you know that doing it is useful to others, as well as the gifts that others give when they benefit from you (like you, or your boss gives you a promotion and a raise). Of course, the two do not necessarily appear at the same time. Many times the value you create is not necessarily noticed by others (decades of maintaining open source toolkits that everyone is accustomed to), and sometimes people will exaggerate your contributions. You should actively seek affirmation from others, which will give you more resources to do bigger things. But you should care more about inner satisfaction, because it is more controllable and not easy to be misled by others. It's more that it will give you inner motivation to do things better, which is the foundation on which you can continue to grow.

technology ultimately serves the product

Students majoring in technology who have just entered the company usually continue to do technology. When I first graduated, I thought going into a big company was about doing technology and being one of the best technologists in the world. And don't make products, because if I make products, why don't I start my own business, the money I make is still my own. The second half is fine, but the first ignores the fact that technology ultimately serves the product. Although the expected time for technology to enter products varies from company to company, it is usually between six months and five years. Expectations are that there are only a handful of companies older than five years, and most of them are dead. So even if you're in the company's research department, you should know the company's expectations for when the technology will land. Otherwise, when the time comes, companies will cut back on investments in technologies that don't live up to expectations. The worst case is that you are on the news: the director of the research institute of a certain company leaves, and the members of the department go their separate ways.

So what kind of technology can enter the product? Usually you have a rough idea based on the company's existing products. The next thing you need to know is what the main value of the product is (applying the formula from earlier). Then you need to think about the value of the technology you make to this product. If your technology can improve the core function of the product, even a little bit, you will get resources to implement the technology. For example, improving WeChat's video compression technology, Toutiao's recommendation algorithm, and Apple's shell materials. Conversely, if the backbone is not seized, such as WeChat skins and Toutiao web version acceleration, the Apple operating system is compatible with other hardware. Even if you can do a lot better than the existing technology, the product team may not be motivated to help you, and even tell you not to do it in the first place.

So whether you are doing technology in the product team or in the company's research institute, you should have an understanding of the value of the product. For example, deeply understand the mantra of product managers: market, rigid needs, pain points, high frequency. At the same time, you should also know the value of the technology in your hand to the product, and use it to guide your planning of the technology route.

Soldiers who don't want to be generals are not good soldiers

People's satisfaction comes from comparison, whether it is comparing with others or comparing with one's own past. This desire drives you to pursue something of greater value. That means you need more resources to be bigger and stronger. At the very least you need a team. You may be the manager, leader, or both of the team.

Maybe you prefer to do tech alone, at least that's what I thought at first. But as your abilities grow and others expect more of your responsibilities, you inevitably have to lead a team. Otherwise you have to look elsewhere for fulfillment. Instead of being pushed into this position, it's better to be prepared from the beginning.

There are plenty of career books to refer to. My own experience is simple: leaders lead the way and need to have a good vision. Managers are logistics officers who make teams perform efficiently. These two points are explained below.

Look three years later

The most important thing for a leader is to find the right direction when exploring uncharted territory with the team. That is to say, to ensure that the products or technologies you make are valuable. Because it takes time to do something, you have to anticipate its value in the future. If the judgment is not accurate, everyone has worked hard for a long time, and found that the effect is average after the completion, then the morale of the team will be low. All kinds of problems will follow.

When you think about the future value of a thing, the time should not be too short or too long. Three years is more appropriate. Assuming that you want to continue in the current direction, you need to consider whether the user base and usage time you focus on will change in three years. More is a good thing, constant means you can't do much, but if there is less, you have to think about turning. You also need to be wary of the emergence of new technologies, which are likely to completely overturn old technologies (deep learning, smartphones, electric cars) in just a few years. Analyzing those failed cases, the parties actually perceived the new technology early on, but underestimated its power. They only see the inadequacies of the new technology compared to the existing mature technology, and then apply the development speed of the mature technology to the new technology, underestimating the height that the new technology can reach three years later, and the degree to which users like the new and dislike the old.

If you want to take a new direction, then you may no longer have the advantage of technology accumulation, that is, the value difference you bring to users may not be obvious or even lower than others. Then you need to find a good track (investment mantra), often disruptive new technology, and the changes that come with it in products and users. Only on a fast-changing track will it be easier for new entrants to overtake corners by more accurately predicting future values. That is, a hero in troubled times.

Good vision requires a long-term training. You need to constantly think deeply and gain your own unique point of view, instead of relying on the opinions of everyone in your circle of friends. So every now and then you need to let go of what you are doing and give yourself time to think deeply. For example, every now and then I would go to the Bay Trail near my house for a few hours, thinking about it as I walked.

core of management is to treat people with

If you have a clear team goal, and a high-quality team, efficient execution is a natural thing. So managers have three core things: recruiting people, retaining strong members, and helping those who fall behind. The best way to recruit people is to recruit people who are better than you. In addition, every recruit is more powerful than half of the same level, which can ensure that the team can continue to improve the quality of the team when it expands.

Members with outstanding abilities are welcome everywhere. One of your tasks is to keep them on the team for as long as possible (though eventually they will leave). One way is to put yourself in their place and imagine how you want your leader to treat you. For example, my own hope is to keep doing things with greater value (sense of achievement) and learn new things from it (personal improvement). The boss was supportive when I was in trouble (often). Others can be converted into the current treatment, such as how much time you can do things you don’t like (do not agree with the value of a thing, but you can’t convince others not to do it), the commute to get off work is crowded, and the cafeteria doesn’t have Chinese food. Therefore, the general direction is to create a relaxed environment, to launch new and valuable projects every year, and to strive for treatment for everyone as much as possible.

For team members whose performance is not ideal, you need to point out problems and give suggestions frequently. If there is no improvement for a period of time, you need to discuss whether the current project is not suitable. If there is still no progress, it can only help them change groups, or ask them to leave. Again, you need to put yourself in the other person's shoes and understand what kind of help and respect you want. Most of the time, it's not that they are not good enough, it's just that you are not suitable. A happy breakup can make ex-members find more suitable positions faster (thus preventing them from sending you blades).

! Focus!

Some people say that a common cause of death for startups is blind expansion after certain achievements. This applies everywhere. Whether you are alone, leading a team, or leading a company, resources are always limited. Focusing resources on the most valuable things ensures success. For example, Apple has hundreds of thousands of people, but it is very cautious about the expansion of product lines. This ensures that each product has enough resources to subvert the market. In the early days you may be able to cast a wide net and catch more fish, but once the value of things becomes clear, we need to gradually focus our resources. Because it is best to do several similar things at the same time, it is better to only do one thing to the extreme and be the best on the market. This way you can always get a positive value difference. One first is better than ten second, and third usually doesn't last long. The same principle applies to life, socializing, and learning.

As long as you put in the effort, shortcomings can become advantages

In the past, every time a new feature of MXNet was released, Zhihu students complained: go back and write the document first. We all know that programmers don't like to write documentation. Since I was a child, both Chinese and English were hovering at the pass line, and I was even more conflicted. When I was 17 years old, I made up my mind to write documents. I spent all my time left to do technology on it. Finally, I wrote the textbook "Learning Deep Learning by Hand" with you, which is now used by nearly 200 institutions around the world. Universities use as textbooks. So, your shortcomings can become your opportunities. As long as you face it head on, work hard and keep improving, your weaknesses will become your strengths.

strengths and avoid weaknesses

All the gifts of fate have secretly marked prices. MXNet itself is neglected when I focus on documentation. The failure to organize a lot of resources to continuously improve its performance and ease of use caused it to fail to achieve the top two. In terms of value, "Hands-on Deep Learning" and MXNet are similar in the number of users and user value, but users spend more time using the deep learning framework than reading textbooks, so MXNet is more valuable. It is difficult to say that it is worthwhile to spend a lot of effort to win a battle in a field that is not good at, but lose a greater value in an area of strength. Therefore, when it comes to improving strengths and making up for weaknesses, it must be judged on the basis of value, not face.

communication overhead in the distributed system is the

When one machine has insufficient computing power, we use multiple machines to work together to complete the task. Although distributed systems are the infrastructure of many Internet companies, it is still difficult to improve performance. Because the actual computing power of each machine will be different, it will strike from time to time. Moreover, when one machine completes its own small tasks, it often needs to wait for the results of other machine tasks, resulting in frequent data communication and waiting. So everyone knows that the key to improving performance is to reduce communication overhead.

When you need a large team to work together to do one thing, the communication overhead is also a big one, and it is more difficult to optimize than a distributed system, because the differences and instability of people are greater than that of machines. Different people have different abilities and different efficiency. Everyone's division of labor is different, resulting in different ways of doing things, and even different optimization goals. Most people only care about their own affairs and do not want or want to worry about other people's affairs. If you can't effectively screw everyone together, it's just a mess of sand, and you can't do great things. If you are new to the workplace, the most critical thing you need to realize is that you need to set aside enough time and energy to communicate. Don't complain that it's a system problem in your company, it's inherent in large teams.

Promotion

I was promoted to a relatively high position because of luck, so I had the opportunity to participate in the company's promotion evaluation from senior engineer, scientist to director. Although different companies, positions and levels bring differences, in general, whether a person can be promoted successfully depends on whether the value of the biggest project she does to the company meets the requirements of the position. There are three main points here: First, the value of the project to the company. It means that the target group and the value difference are what the company cares about, not what you personally or society care about. The value here is usually how much money is made to the company, or how much money it might make in 3-5 years. The second is to see that your biggest project should be "grade" enough. Accumulating a lot of projects, it may not work to look at the hard work without looking at the credit. The third is your contribution to the project, such as how much you are responsible for, whether you have contributed code, team coordination, publicity, planning, or applying for resources. A common misconception is that working with people reduces my contribution. If you don't cooperate well with your collaborators, so that 1+1 is much less than 2, then your contribution is indeed reduced. But if the value of the project is increased through cooperation, then your contribution will not be less. Especially if the project value is up a notch, all the better.

An often-overlooked "unspoken rule" of promotion is influence. As the position increases, so does the company's expectation of your influence. From influencing a small team, including formulating technical routes, helping players get started, answering questions, and even helping others to complete their work, to influencing the next group (manager), the next department (senior manager), and the next group (director), Finally to influence the strategy of the entire company (Vice President).

Unless you're a natural leader, you have to work hard to cultivate your own influence. Simply put, you are actively helping others while taking care of your own affairs. When others trust you, consult you, and are willing to work with you, then you have influence over them. You may feel that helping others will delay your life. But from a company perspective, that dedication needs to be encouraged and rewarded. Plus, the trust you earn from it gives you fame and connections that can be useful in the long run.

raise

Salaries are relatively transparent in large companies, and the salaries for each level can often be found online. Salary fluctuates within a level, and generally has a minimum and maximum value. 80% is around the median, with 10% at each end, those who have just risen to this level and those who are about to reach the next level. It can be simply thought that with the improvement of ability, your salary will gradually jump from the minimum value to the maximum value, and then be promoted to the corresponding range of the next level.

The median pay for different grades is usually a series of equal ratios, not equal differences. For example, someone one level higher than you may earn half as much as you, but someone three levels higher is not 150% more than you, but 238% more. In this model, you need to optimize the height you can reach in five years, or ten years. So when comparing offers, you don’t care too much about the digital price difference between them, but about the development after you go (you said that I will buy more coins now, maybe I will have freedom of wealth immediately, that is also a ideas).

Summary: Focus on the Most Valuable Things

If the insights of these five years were refined into one sentence, it would be a very bland sentence: focus on the most valuable things. First, you need to have a clear understanding of the value. Then, for a thing, it is not only to recognize the current value, but also to predict the future value. Secondly, when you have a grasp of the future through continuous rapid trial and error, you need to gradually focus the resources you can call on the most valuable thing, and do it as well as you can.

If you can do a few things that are of great value in your life, it will be worth it.

While writing this article, I was shocked to hear that Mr. Yuan Longping passed away. I would like to commemorate his great life with this article: focusing on hybrid rice, creating one of the greatest values in human history. My role model.