What are the characteristics of those who grow up fast?

Hello, I am Chen Yang, the project leader of "Guo Dongbai's Architecture Class".

In the last class, Mr. Dongbai and Mr. Hualiang exchanged views and opinions on issues such as career planning and career choice models. So let's continue this class and see what else they talked about.

Moderator: Guo Dongbai, CTO of Guazi Used Car Group Guest: Chen Hualiang, VP of Daily Fresh Technology, former chairman of Meituan Algorithm Committee

The full video of the conversation follows:

!video[https://static001.geekbang.org/resource/image/47/f8/47c09de35405886be6f4565f028d65f8.jpg](https://media001.geekbang.org/customerTrans/7e27d07d27d407ebcc195a0e78395f55/2431b7fd-17fea301c2d-0000-0000-01d-dbacd.mp4, https://media001.geekbang.org/967297549fc2456e920eb17b6c6a085d/9e8d372b4d0847e883b865c1e4c4995f-902eb7938d3b8775087bc430c7d9991a-sd.m3u8)

In addition, I also compiled a version of the essence of the text based on the video, and some of the dialogues have been deleted. The dialogue is a process in which the two teachers inspire each other and collide with each other, and it is difficult to present the whole picture in words. So if you have the learning conditions, I recommend you to watch the full video.

Essence of text

Dong Bai: Let’s go back to a relatively universal topic. In fact, you also mentioned just now that you recruited a lot of people from Baidu, Meituan, and Ali to Daily Youxian as architects. In this process, what kind of people do you think are suitable for working hard in the industrial Internet?

Hua Liang: That’s right. I just said that the industrial Internet is equal to the operation plus the consumer Internet. In fact, I have always emphasized this formula. Because operation is a very critical, even the most critical element in the industrial Internet. Therefore, if those of us who work on the consumer Internet want to enter the industrial Internet, I think the first requirement is to have a deep understanding of operations.

Because in essence, we use technology to reconstruct all aspects of operation to make the entire operation more efficient and intelligent. Therefore, the term we talk about in terms of talents is called "first principles" . For example, I work in technology, and my technical level is very good. But I still have to go to the site to see how those businesses are running offline.

Still taking supply and demand as an example, you need to design an algorithm for supply and demand. In order for the algorithm to be well applied to this business, it is necessary to have a deep understanding of the offline operation logic of supply and demand.

For example, in the procurement process, how did you purchase it? Is there any possibility that something could go wrong? Where might something go wrong? We have to go to the first scene to find out. So this is the first suggestion. To contribute our technical capabilities and create value in the industry, we must first of all have the first principle.

The second is some more general requirements, which I call evolutionary power . Evolutionary power essentially means that each of us should continue to grow.

Just like me, as a technical person, if I go to the industry, I need a strong evolutionary force, because there are many things to learn. We need to consult experts in the industry, learn the history and evolution of the industry, and improve our ability and level. Finally, combine our understanding of the industry with technical understanding, and then make higher-quality decisions.

The third is the technology itself , we still have to pursue some of the latest technologies. Why do we bring in senior technical experts from big companies like Ali and Meituan? In essence, we need to graft the latest technology in the industry to the fresh food industry to maximize value. Rather than saying that I am using some relatively backward technologies to empower this industry, the value it brings is not big enough.

Dong Bai: The first principle you mentioned, from my understanding, is more about the ability to think. Thinking ability, evolutionary ability, and technical depth, assuming that you have all three, you can basically enjoy it in any industry.

Hua Liang: I'm also curious. In the used car industry, how do you choose suitable talents?

Dong Bai: In my opinion, the first thing needed is courage . Many people are infinitely afraid before entering an industry that has not been proven . The two of us just said that when changing industries, we will build some parameters to calculate. Like this model of an unknown industry, the calculated results may be relatively uncertain. So before entering this industry, this person must have a character that likes to embrace uncertainty.

The advantage of entering a track with great uncertainty is that there are fewer people coming in. If the bet is successful, the ROI is very large; but if the bet is not successful, there are many risks. So courage often comes first.

The second one is actually similar to your thinking, which is thinking ability . In a completely uncertain track, everyone is guessing. Everyone will talk about the cost relationship, where the profit will be, and where the next growth point will be. At this time, your personal thinking ability and judgment ability are very important. You need to clearly restore the thinking path of others , and see where there is a scale advantage.

Just like Rashomon, everyone is blah blah blah, saying that they are right and what their direction is. But we must find something that can be locked in the end. Only at that point will there be an explosion of technology.

Because in the industrial Internet, the explosion of technology is definitely not universal, only in the field of e-commerce. So the third ability is the ability to get results . In my opinion, if it is not a leader or a decision maker such as a CTO, it is necessary to see whether this person can turn these points into reality. This is very critical.

You'll find a lot of people with amazing resumes. But when that point is really given, only a few people can change that point from 0 to 1. Such people are very hard to find.

Hua Liang: I'm still a little curious, because you have a lot of experience. Do you think there are any differences in the selection of talents between Chinese and American companies?

Dongbai: The difference is quite big. In my opinion, the difference lies mainly in the survival principles that companies believe in . Most American companies believe in technology and believe in in-depth iterations on a track. You can see that Google has dug extremely deep in a vertical field, and then earned huge value. The same is true for Amazon. It only adds cloud to e-commerce, and cloud is just a by-product of e-commerce. Unlike many domestic companies, as long as they occupy a monopoly position, they can't wait to spread it out.

Therefore, foreign companies pay special attention to the absolute barrier advantages of talents . For example, two people are doing graphics, or are experts in Video Processing. One of the people's algorithmic ability is obviously better than the other, that's no problem, the company is looking for him, and the cost is not important at all. why? This is an Internet product logic. In the end, the cost is actually a fixed cost, which can always be ignored.

And the goal of American companies is basically to dominate the world. In this case, once a company finds an excellent talent, it can continue to increase the size. In other words, what American companies find are basically one-dimensional talents , who have taken a fancy to his ability in a certain aspect.

But Chinese companies often value multi-dimensional talents . For example, this person not only needs to have architectural capabilities, but also needs to have certain algorithmic capabilities, and at the same time must have certain leadership skills. Because domestic Internet companies still have a very large voice . Why do you have such a big voice? There is too much domestic supply, so big factories have become a very attractive option, even the first option, if they are willing to embrace uncertainty.

From the perspective of the enterprise, a person who possesses all three abilities is a high-potential talent. In fact, it is also this selection method that leads to the fact that Chinese talents are multi-dimensional talents, but these talents often get more rewards and opportunities.

Hua Liang: Let me ask one more question. We may face some technical practitioners who hope that they can become a senior technical expert in technology. So is it better to go to a foreign company for training first, or is it better to grow in a Chinese company first? What is your personal suggestion?

Dong Bai: If your goal is the technical route, not the kind of entrepreneurship you just mentioned, then I feel that it will develop faster if you go to a big factory. But there is a prerequisite for going to a big factory, that is, you must be able to win in a big factory. What does it mean to be able to win ?

Some people entered big factories reluctantly. For example, after interviewing ten big factories, they only got one offer in the end. After entering, it happened to be at the bottom of the big factories, basically between 3.25 and 3.5 every year. Then what you get is actually a tail opportunity, and it can't even be called an opportunity. It's a bit like a labor-type job . In this case, you actually have no development in the big factory.

If you are looking for a job, you have offers from 10 major companies in your hand. At this time, you can choose. You will see which big factory has given me what kind of opportunities, and then choose the one with more room for development, or the one you like more. In this case, I think it will be good for you to enter a big factory and grow.

If you can win in a large factory, the knowledge and resources you gain will be better than those in a small factory. But the premise is to win . If you do some very tail-end work in a big factory, it will not work.

But relatively speaking, I think if you can’t win in a big factory, it’s not bad to win in a medium factory. "It's better to be a chicken head than a phoenix tail", I feel that such a judgment is better for career development.

Hua Liang: Our cognition is quite consistent. I personally also suggest that if you want to go deeper in technology, it is the same whether you go to a large factory or an overseas company, and you need to continuously improve your competitiveness in the market.

Dong Bai: Students often ask me in private, should I change jobs, and which track should I change to? What do you think about this question?

Hua Liang: I think so. Although overall, the economic pressure is indeed relatively large. The Internet industry, in particular, may have passed its golden age. But I think this industry as a whole still has relatively large potential and space. Especially for technicians. As I said just now, the opportunity for the consumer Internet may be relatively small, but there are still plenty of opportunities for the industrial Internet . There is still a lot of work to be done in terms of industrial digitalization and intelligent transformation.

Including the country is now also proposed that we want to realize the digitalization of the entire society, in essence, use technology to change society and the world. I personally agree with the profession of technology very much. I call technology " a tool that changes the world at zero cost ".

Think about it, what resources do we have that can change the world? Money, connections, 99.99% of people don't have enough resources to change the world. So relatively speaking, technology is a tool with zero cost. Whether it is a student or a person who has just entered the workplace, as long as he is willing to delve into technology, learn the latest technology in the industry, and then apply these technologies to the industry or consumer Internet, it will definitely bring great benefits to the society. great value.

Come back and talk about whether there will be any problems with employment, I think there is no problem at all. Because technology can be applied in all aspects. Although some directions may be relatively saturated, we can use them in other directions, so I don’t think you need to worry too much.

The main thing is to continuously improve our technical level and make ourselves a top expert in the industry. Then everyone will definitely have opportunities, and there will definitely be corresponding positions to use their heat.

Dong Bai: You have revealed a concept called "skills do not overwhelm your body". My concept may not be quite similar to yours. It may be related to my work experience in the United States. In fact, what I care more about is the scarcity of skills of a technician.

I hope to learn one of the rarest skills, and the market demand for this skill has been magnified to the extreme. For example, deep learning, if you learned this skill 5-10 years ago, then you are right. It's equivalent to choosing an ability that was scarce at the time, and digging deep into this ability. Then the market demand for this ability is suddenly magnified, and it is very valuable! So I actually believe in a life-gambling logic , and I like to gamble more.

My philosophy is that if you don't gamble, you will be a mediocre person no matter what you do, so it's better to gamble. Of course you have to be smart enough, or run fast while betting.

So I think you must bet on one at the beginning, at least you think it may be more promising technology, and then amplify your own scarcity in it. Otherwise, it would be like everyone else. I don't know what you think?

Hua Liang: I can interpret your idea in another way, because we don't have any fundamental contradictions.

I just said that we need to learn technology in depth, which is equivalent to becoming a top expert in this industry, at least the first-class level in this industry. In this way, you will have the opportunity to exert your heat.

But the technology will change. For example, we used LR or more traditional machine learning in the past, and we played very well. Now it is suddenly said that new technologies are coming, and deep learning is coming. So are you missing out on this opportunity? I don't think so either. Because there is actually a characteristic in it, which is the evolutionary force I mentioned just now. It cannot be said that I learned LR before, but now that deep learning comes, I will stop learning it. We must keep up with technological changes.

Technical brothers often complain that technology changes too fast. Three months ago, I used LR, but now I start to use DNN for deep learning. Is it changing too fast? Compared with other industries, the changes must be relatively fast.

The technology in some industries may not change for 10, 20, or even 100 years. But in the field of Internet technology or IT technology, there will really be a wave of major technological changes in a few years. So for individuals, you are required to have a certain evolutionary power and have enough ability to adapt .

Because the underlying principles of technology are still the same, for example, you know LR, and you are at the top. Is it difficult to learn deep learning at this time? I don't think it's difficult, it's really not difficult, and that's how I came here. You may spend a few months applying the original principles of LR to deep learning, and soon you will be able to acquire various skills in deep learning and become an expert in this field.

So as long as there is enough evolutionary power, technology can be transferred.

Dong Bai: Speaking of which, I want to ask you a question. In the fresh food industry, is there a particularly relevant technology stack?

Hua Liang: In terms of pure technology, I don't think there is anything new. From the underlying database to components, and then to the processing of big data, the underlying technical framework of this message stack has not changed much. So this is also a good thing for everyone, and the technology you have learned can also be applied to the industry. Essentially everyone is the same.

And I believe that these technologies are common in most fields, but the upper-level business technologies will be different. So I divide technology into basic technology and business technology .

While we use basic technologies to solve business problems, we also need to build the technical architecture of the business. This is where it's different. So how do we acquire this ability?

I think it is still necessary to have a deep understanding of the operation of the industry , and then combine it with technology to form a business technical architecture. Specifically, let me expand a little bit. What are the characteristics of the business technology architecture? I can give a few examples. The first is to say that there must be first principles when building a technical architecture .

I still emphasize the first, because the business itself is very complicated, so when you build it, you need to understand the entire process of the business, such as our 300 links.

Then there is still a network relationship between these links, so the business itself is also a complex network structure. You need to have a better understanding of it before you can build this business architecture.

The second is that businesses are diverse, so our business architecture needs to be able to support diverse changes.

In actual business, the business architecture will be more complicated. As I said just now, we have built a system to manage picking. When managing picking, you have to consider all aspects. For example, what should I do if the stock is out of stock? What should I do if I find that the goods are damaged? Also, it was good, what should I do if I accidentally broke it when I picked it up again?

In the actual business operation process, each link has a lot of possibilities, which I call diversity . Then we need to consider the diversity of business when building the business architecture, so that my business architecture can catch the changes of the business and adapt to the needs of the entire business.

The third is the bottom line principle . In the actual operation process, we went to the site and did a lot of interviews, forming a deeper understanding of the business. Then the business I did is also very good to use, and it can be turned around.

But there will always be some contingencies, so in this case, we need to be able to keep this business going. In other words, when an accident occurs, we need to have a mechanism to keep the business going.

Let’s take an example of supply and demand. For example, during the epidemic, there will be an extreme situation, that is, a certain demand is very strong, and the goods will explode.

For example, I could only sell 5 watermelons, but suddenly the epidemic came, and I could sell 500, so what should I do? There needs to be some back-and-forth mechanisms to ensure that the business can continue.

To sum up, the basic technical architecture is basically reused, and there is no major innovation. More innovation lies in the business architecture . Businesses are different, so how do we make a targeted business architecture to better meet the needs of business development. This is my understanding.

Dong Bai: There is a problem, I have not found a good solution. There are many attempts in our industry, but they are actually ineffective, that is, invalid packaging. It is very related to the large search space just mentioned, which is equivalent to "burn after use". It went online quickly, and then found that it was not easy to use, and quickly went offline after snapping.

Just like consuming the Internet, I always say how much my QP has reached, with high concurrency and high availability. But it is actually stable, not moving for many years.

This is actually a wide-ranging search. Once it doesn't work well, throw it away immediately. In fact, no one in the world can do it. Have you ever thought about this question?

Hua Liang: Yes, I think everyone will have the same experience. This is the need of industrial innovation. It means that the industry needs to do a lot of innovation, and if this innovation is written in hard code, the cost will be very high. I am going to recruit a large number of engineers to support the innovation of the industry, and the cost is extremely high. General industrial innovation cannot support such costs.

How do we innovate at a lower cost? To answer this question, our experience is like this. One is called MVP. In my industry, whether it is operations or business, what business do I want to do? I want R&D to be supported by a system.

Sorry, this system is very expensive, don't ask me to support it, run it yourself first. You can use relatively simple solutions, even very crude ones. Maybe an Excel table plus an MIS system information management system will allow you to run this business very painfully.

Very low cost, preferably a few man-days, or a dozen man-days, dozens of man-days, is fine. Make the business work and let it prove its worth. In the end, the business can be said, you see, I ran it manually, and with a little technical support, it proves that this thing can run smoothly, so the value in the future will definitely be great. At this time, we will enter the iteration of technology and really start to do technical empowerment.

In this way, we can reduce our trial and error costs and reduce the cost of research and development, which is also a requirement for business.

Second, of course we also have requirements for ourselves, that is, how can my system architecture be forward-looking. So one of my very important requirements for architects is that you have a forward-looking view of the architecture design of this system. Stability and simplicity are the requirements of Baseline. The more important requirement is that you have forward-looking capabilities and whether your architecture design can meet the needs of future business changes.

I think it is very important whether the needs of future business innovation are reflected in the design.

I hope that our technical architecture has a certain degree of forward-looking, and then allows the business to do some low-cost customization on it, and support some innovative attempts, thereby reducing the cost of innovation.

This is the requirement for ourselves. This requirement is called low-code in a more fashionable way, and we will also make some attempts in this regard. The built platform is as low-code as possible, and then supports this business, and this thing can be run through some simple configurations. In this way, the cost of innovation will be reduced.

Dong Bai: You see that you keep talking about the word " evolution ". I like the theory of evolution very much, and I have read it many times. I found that the business form and the final technology are actually restricted by the environment. The general environment we both live in is very similar, that is, the environment of the industrial Internet .

If subdivided, there will definitely be differences. But you will find that under the big environment of the industrial Internet, some features are still beginning to emerge. The low-code you just mentioned, including this configuration, we have a lot of Workflow and low-cost digital capabilities.

Changing the subject, let's talk about this growth in a broader sense. For example, how should you grow before you become an architect? Like school enrollment, is there any feature that is very good? It means that when you see this young man, you feel that you will become a great weapon in the future.

Hua Liang: Everyone's growth path is definitely different. Some people do have great potential and grow very fast. We had previously recruited a school to recruit students, and his growth was very fast, basically he was promoted every year. Why can these people grow up so fast? My personal summary is this.

First of all, he needs to have a good business understanding . Specifically, you have to know what industry you are engaged in, and what kind of formats this industry has. Then what does your company do in this industry, and what does your project serve.

These things should be clearly understood and understood. In this way, when you use technology to solve business problems, it will be relatively more accurate, at least you will not make things wrong.

The second one I still call it technical ability . We take the technical route as a whole, which is our housekeeping skill. Then a person who grows fast is usually very strong technically, and he will have his own set of learning methods. For example, he will communicate with more senior people and collect more information. Also look at some of the latest technical directions, and quickly improve your technical capabilities to the industry-leading level.

Having said that, I personally have an experience that I would like to share with you. How can I let myself quickly master the best technology? In fact, GitHub is a very good tool with a lot of treasures on it.

Because in addition to learning some theoretical knowledge, we also need to learn some practical knowledge. Then GitHub is a very good place for actual combat, and there are a large number of star projects in it. We can read the code of these star projects to understand how senior architects structure the code.

The third one I think is the power of evolution . This is the point I have been emphasizing. A person with evolutionary power usually doesn't stay in his comfort zone all the time, he will reflect on whether he has made progress. Before going to bed every night, ask yourself: Have I gained something new today? Did I improve today?

This progress can be very small. But as long as you can make a little progress every day, it will be a very considerable progress in a year. He will continue to find himself growing and evolving. Such a person has great potential. His growth characteristics will also be very obvious, far beyond the average friend and colleague.

Dong Bai: I really resonate with what you said. I have observed that many students will enter a contract performance mode at work , and technology is particularly easy to enter this mode, because our technology is at the end of all demand chains. What is the performance model? For example, I go to work on Monday morning to see if there is any demand. The demand came and I was busy until Friday night. Some people even 996, busy until Saturday night. The logic is "because I'm busy, I grow."

This logic is actually very wrong, equating "busy" with value and growth . Is there really growth? not necessarily. But you will find that those excellent people will never accept this logic. Whether it is an individual or a company, once it enters this state, it is very scary.

Hua Liang: Let me also share my growth experience. A few things have been mentioned just now, that is, I personally have a very big step.

Everyone will understand that sometimes growth is a quantitative change leading to a qualitative change. After accumulating enough in certain stages, there will be a relatively large step in the next stage. I actually went through a similar process.

When I first joined Baidu, I was also a newcomer to the workplace, and I was still a rookie. Although I learned some things in school, it is still very different from the actual application.

But we had a good opportunity at that time to do technical iteration . For example, I made a technical solution upgrade, and then applied it online. What value will this program bring? We have to prove this value first, and then we can have full traffic.

Then we had an email mechanism called Launch View at the time. There will be a lot of content in it, including what I did, what technical solution I used, how I experimented, and what the result was.

I learned a lot of valuable information from here. I will read every Launch View to see how other people practice, how they think, and what technical solutions they use to solve any problems they encounter.

At the beginning, it was tiring to read, and some of the deeper technologies were hard to understand. Then after watching too much, I suddenly understood, I understood all these things they did. At this time, I found that I would have a great improvement in technical cognition.

The second experience, I also look at GitHub's star projects. By watching this kind of actual combat project, I will have a more three-dimensional understanding of how to do the architecture and help me better solve practical problems.

So what I want to emphasize is that one is that we need to learn theoretical knowledge. Buy a variety of books, learn if else, learn functions, classes, and architecture. I call this theoretical knowledge cross-sectional knowledge .

But at the same time, I think it is necessary to combine longitudinal knowledge , that is, practical knowledge. If we have the opportunity, we can learn the actual projects in the company and learn the actual projects in GitHub.

The combination of these two kinds of knowledge will greatly improve our technical capabilities.

Dong Bai: I especially agree with what you said. I share two stories. In the first story, Amazon's code is open to all employees. After I joined Amazon, it took almost half a year to read the codes of more than 20 people in the team at that time, as well as the codes of all their relying parties. It just so happened that my position at the time was quite special. I was a data architect, so I didn't write code, just read code, which was OK.

Looking at the back, in fact, the details of the code are seen less and less, and the API is seen more and more. I think API is more representative of an architect's layout thinking.

After reading some APIs, I really feel that "there are hills and valleys in my chest". It's a bit like we used to read literary works, or even listen to other people's words, but at first we couldn't understand what they said. When you can hear and see the brilliance, your level will come. But it can't be distinguished before that, because our feedback function is not formed yet.

So I think looking at the code or looking at the API is the process of forming dopamine, which is very exciting. I guess you should feel this way.

The second thing is that big companies still have some valuable content. Amazon has a thing called COE. We were open to all employees at the time, and we could see all historical P0 faults, and there seemed to be 400. I saw the replay of these P0 failures from beginning to end, which is the biggest treasure I got from Amazon.

The third thing is also quite interesting. Ali has an operation university, and my study ranking at the operation university was the first. Many people ask, how can technology increase business understanding? I think that after reading all the operation strategies, you will understand the business. Optical code, business understanding can never be learned.

So my conclusion is similar to yours, that is, theoretical knowledge needs to be learned. However, this is the basis for talking about war, and what really sublimates our capabilities is some actual combat information and content.

I am very happy, I can gain something every time I chat with Hualiang. In fact, talking to someone who thinks very deeply can indeed learn a lot. This is also one of my learning methods. Find someone I like or respect from the bottom of my heart, and then chat with him often, so that I can get the knowledge I want. And this kind of communication can also help each other. This mutual exchange of friendship is actually a very good way of learning.

Then I will ask Teacher Hualiang to share his advice again.

Hua Liang: I am very glad to have the opportunity to communicate with Dongbai. I feel the same way, we must communicate more with each other. Because everyone has different knowledge and experience, through communication we can learn from each other and learn more.

I would also like to share some of my personal experiences here. Just some thoughts, I hope it can help you.

First, I hope everyone must do career planning. It sounds like correct nonsense, but it is indeed a very important experience for me. Although our career development will be affected by many factors, the general direction should be determined from the very beginning.

For example, whether we are going to climb the Himalayas or go on a long voyage to the ocean, the general direction must be determined, and then we will make adjustments according to the actual situation. For example, should I take a small boat first and then change to a larger ship to go to the ocean, or go directly to the larger ship to go to the ocean.

The second is to believe in the power of technology. Technology is a tool to change the world at zero cost, and everyone must make good use of this tool. As long as you can concentrate on learning technology and become an expert or senior architect in this field, you can change the world and exert your personal value in various fields.

And the third one is the evolutionary force that I have been emphasizing. Because the world is changing all the time, society is changing, and technology is also changing, so everyone must have better evolutionary power to keep up with the changes in society.

I think the last point is very important. Especially the students at school, or brothers and sisters who are new to the workplace, I hope everyone will not rush to start a business. Your current resources are very limited. If you start a business, the chances of winning are probably not high. We shouldn't waste our precious time on things that don't have a high chance of winning.

So at the beginning, we still have to enter some big factories, continue to improve our capabilities, and accumulate more resources. If you really want to start a business in the future, you can wait until you have accumulated a certain amount of resources, then your winning rate can be increased dozens of times.

This is some of my experience sharing today, I hope it will be helpful to everyone!

Article source: Geek Time " Guo Dongbai's Architecture Class "