任正非:鼓励“胡说八道”,自由研讨科学和技术问题

colabug免费资源网站导航
华为心声社区发布了《任总与系统工程领域科学家、专家会谈纪要》。在会谈中任正非表示办“黄大年茶思屋”的宗旨,就是希望大家就科学和技术问题来“胡说八道”,只要不涉及社会科学问题。每个科学家、专家来华为,就讲你们最擅长那一点,哪怕华为公司所有人听不懂也不要紧,因为它留在人的心中,敲开他的脑瓜,他就不封闭了。

如果你们说要接近我们公司的研究,我们感谢,但不是我们最需要的。我们是干具体活的,目光比较短,最需要的是我们看不见的那一小部分,我们要看“山外的青山,楼外的楼”。你们看得远,你们讲的东西正好是我们不明白的,就只要告诉我们。我年轻时曾听计算机控制系统的科学家讲课,他说:“计算机用于公共汽车的票务管理。”我听懵了,计算机是算数字的,怎么用于票务管理?当时留在我心里一个问号,后来我才知道原来是真有的。所以,我们要建立十年、百年对社会事物的假设,假设我们开始要做这些事,我们引进来的科学家做的也就是十年、二十年以后的底座,他们也不参加“杀猪”,我们只要求一部分拿着“手术刀”参加“杀猪”的战斗。

任正非:鼓励“胡说八道”,自由研讨科学和技术问题

以下为全文:

一、系统工程不仅是理论、方法和实践,更是开放的思想和哲学。

我们要利用系统工程的思想,把公司内的“围墙”炸开,摧毁各种“土围子”,打开思想的禁锢,形成开放的思维,慢慢消化吸收,不断积累组织的能力,更好的面向未来。现在实行干部任期制,就是不断降低部门“围墙”的措施。集中少数人搞系统工程来解决一个问题是没有意义的,只有千万人的思想都走向系统工程,有系统工程的概念,才能真正的发挥价值。我们公司的每个人都要学会建模,复盘-建模、再复盘-建模,做完一件事,要总结哪些做得好,哪些做得不好,建立一个初步的模型;然后用这个模型去做第二件事,有偏差再回来复盘修正,反复复盘,就建成一个小小的模型;我们几个团队打通,几个模型在一起再次碰撞,用理论来指导实践。模型变大、变系统,减少不必要的能耗就提升了竞争力。以前我们实践不总结、理论没有实践,两个脱节,效率低、成本高。

为什么要把系统工程提到这个高度?我们逐渐在通信领域世界领先,比如在光网络上已经深入到“根”里,我们现在主要骨干传输是用400G,远距两千多公里传输不需要电中继,一根光纤里面可以传80个波的400G,就是32T;另外,已经在5G中应用6G的一部分技术。在算力网络、存储网络、计算网络上,各方面的协同需求越来越突出,更需要系统工程。大算力时代已经到来,我们正在转弯,转到这个大的算力网络上来,我们要理解并参与到这个时代的改变。在这个大算力时代,我们是5G+云+AI聚合起来,发挥云、管、端、芯的综合竞争力,我们一定要用科学的方法去拆除“小农经济”思想下的“围墙”,通过系统工程的思想,让大家放弃“小农”意识,面向未来。

二、系统工程是科学方法,即使一颗螺丝钉也有系统工程。

几千年人类的繁衍传承本身就是系统工程,整个地球也是一个自然系统工程。自然而然就是缺少竞争力的,要优化自然,改善自然,才有竞争力。

第一,系统工程不能搞垄断,如果就几个专家能搞,只集中在小部分人,是会落后的。我们要在系统工程上面有个“灯”,谁都可以搞系统工程。“北斗计划”就是瞄准未来数十亿台服务器一个操作系统,从顶上往下做架构。指引“登山”的攻关的参考路径,从三万台服务器到几十万台服务器……,上亿台服务器一个操作系统,允许专家务虚“胡说八道”,他们只是给从下往上攻的“登山部队”一个路径参考。系统工程的“高山”谁都可以去攀登,谁都贡献一点,自然就形成了系统工程。每个口、每个人都要去理解系统工程,在他自己这个地方怎么应用。

第二,工业互联网是系统工程,要做好工业互联网中的云底座。每个行业都有自己的应用平台,比如煤矿、公路、铁路、机场……,它们都具有很强的个性和行业特征,这些都是系统工程,需要大家一起来做,一起做贡献。这些努力就帮助各行各业逐步走向工业互联网。

工业首先要走向机械化,机械化走向电气化,电气化走向信息化,信息化走向智能化,这时候才有机会把中国的制造业转变成工业互联网,德国目前走在了我们的前面。华为在这个过程中,重要的是做出自己的贡献,例如光传感、光传输、5G连接,以及云底座等。

三、科学就是无尽的前沿,不要急功近利。

第一,国家重视理论建设,就要防止急功近利。对于一些短期内不能创造价值的纯理论科学家,要理解他们,让他们有体面的生活,这样国家才有未来。我认为,理论科学就是我们国家未来的“水塔”。黄河、长江是母亲河,母亲的乳汁在哪?在青藏高原,母亲的乳汁就是喜马拉雅山的雪水,黄河、长江全靠西藏的雪水流下来,西藏下雪多,就意味着国家明年要丰收。“爬喜马拉雅山”的贡献是很大的,下游的“农场”、“牧场”要分一些粮食给“爬雪山”的科学家们,不能让他们在“喜马拉雅山”上饿死。

我们公司“爬雪山”的人每年要花掉十几亿美元,合作部支持了很多钱给高校做前沿研究,其实对我们是没有商业价值的,因为理论研究出来后是对全人类完全开放、共享的。只是我们有十几万工程师,消化能力特别强,就有可能优先消化这些理论。比如,土耳其Arikan教授十多年前曾发表了一篇数学论文,发表两个月后,我们发现了这篇论文,就根据他的理论做成了5G Polar码。所以,理论科学可以脱离实际,不一定要和实际相结合,也不一定要解决实际问题。

第二,尊重知识分子,尊重知识分子的创造发明,允许他们犯错误,思想本来就是一个闪亮的明灯。比如,软件是个复杂系统,不仅要在理论上“捅破天”,更重要是“扎到根”,“扎到根”比“捅破天”难。我们能解决的还只是信息网络这一块,软件的难度无法想象,我们要想办法通过3-5年或更长时间逐步解决。

第三,重视基础教育,允许一些差异化的教育。目前的教育体制可能还停留在工业革命时代,理论建设不够,不适合信息时代,希望在教育上能够真正重视,允许一些“怪异”的学生成长。我们公司的天才少年,最重要是看他的破题能力,而不是他的综合素质。

四、“山外青山楼外楼”,鼓励“胡说八道”,自由研讨科学和技术问题。

我们办“黄大年茶思屋”的宗旨,就是希望大家就科学和技术问题来“胡说八道”,只要不涉及社会科学问题。每个科学家、专家来华为,就讲你们最擅长那一点,哪怕华为公司所有人听不懂也不要紧,因为它留在人的心中,敲开他的脑瓜,他就不封闭了。如果你们说要接近我们公司的研究,我们感谢,但不是我们最需要的。我们是干具体活的,目光比较短,最需要的是我们看不见的那一小部分,我们要看“山外的青山,楼外的楼”。你们看得远,你们讲的东西正好是我们不明白的,就只要告诉我们。我年轻时曾听计算机控制系统的科学家讲课,他说:“计算机用于公共汽车的票务管理。”我听懵了,计算机是算数字的,怎么用于票务管理?当时留在我心里一个问号,后来我才知道原来是真有的。所以,我们要建立十年、百年对社会事物的假设,假设我们开始要做这些事,我们引进来的科学家做的也就是十年、二十年以后的底座,他们也不参加“杀猪”,我们只要求一部分拿着“手术刀”参加“杀猪”的战斗。

“黄大年茶思屋”是一个开放的交流平台,做好内外沟通,将来可以打通国内高校和国外高校的联接。你们科学家、专家不需要自己去提供算力,用我们云平台,包括AI的算法、存储等,都可以提供给你们。你们把精力集中到前端,后端平台我们就能支持你们。

If you say you want to get close to our company's research, we appreciate it, but it's not what we need most. We do specific work, the vision is relatively short, what we need most is the small part that we can not see, we want to see "the green hill outside the mountain, the building outside the building". You can see far, and what you are talking about happens to be something we don't understand, so just tell us. When I was young, I listened to a scientist in computer control systems, who said, "computers are used in bus ticket management." I am confused. The computer is numeric. How can it be used in ticket management? There was a question mark in my heart at that time, and only later did I know that it was real. Therefore, we have to establish assumptions about social things for ten or a hundred years, and suppose we start to do these things, and the scientists we bring in will do the base ten or twenty years later, and they will not take part in "killing pigs." we only ask some of them to take part in the battle of "killing pigs" with scalpels.

The following is the full text:

First, systems engineering is not only theory, method and practice, but also open thought and philosophy.

We should use the idea of systems engineering to blow up the "walls" in the company, destroy all kinds of "soil walls", open the confinement of thinking, form an open mind, slowly digest and absorb, constantly accumulate the ability of the organization, and better face the future. At present, the implementation of the cadre tenure system is a measure to constantly lower the "fence" of departments. It is meaningless to concentrate a small number of people to engage in systems engineering to solve a problem. Only when the thoughts of thousands of people move towards systems engineering, and with the concept of systems engineering, can the value be brought into full play. Everyone in our company should learn modeling, review-modeling, review-modeling, finish one thing, summarize what is done well and what is not, and build a preliminary model; then use this model to do the second thing, if there is a deviation, come back to review and correct it, review repeatedly, and build a small model; several of our teams communicate, and several models collide again, using theory to guide practice. The model becomes larger, the system changes, and the competitiveness is enhanced by reducing unnecessary energy consumption. In the past, we did not summarize the practice, the theory did not practice, the two were disjointed, the efficiency was low and the cost was high.

Why should systems engineering be raised to this height? We are gradually leading the world in the field of communications. For example, we have gone deep into the "root" in the optical network. Now our main backbone transmission is 400 gigabytes, and long-distance transmission over more than 2,000 kilometers does not require electrical relay. 400g, which can transmit 80 waves in an optical fiber, is 32T. In addition, part of the 6G technology has been applied in 5G. In computing network, storage network and computing network, the demand for cooperation in all aspects is becoming more and more prominent, and system engineering is more and more needed. The era of big computing has come, and we are turning to this big computing network, and we need to understand and participate in the changes of this era. In this era of big computing power, we are 5G + cloud + AI together to give full play to the comprehensive competitiveness of cloud, management, end, and core. we must use scientific methods to remove the "fence" under the thought of "small-scale peasant economy" and, through the idea of system engineering, let everyone give up the consciousness of "small farmers" and face the future.

二、系统工程是科学方法,即使一颗螺丝钉也有系统工程。

The reproduction and inheritance of mankind for thousands of years is itself a systematic project, and the whole earth is also a natural systematic engineering. Naturally, it is lack of competitiveness. It is necessary to optimize and improve nature in order to be competitive.

First, systems engineering cannot be monopolized. If only a few experts can do it and concentrate on a small number of people, it will lag behind. We should have a "lamp" in systems engineering, and anyone can engage in systems engineering. The "Beidou Project" aims to build an operating system for billions of servers in the future and build it from top to bottom. The reference path to guide the key problems of "mountaineering", from 30,000 servers to hundreds of thousands of servers. Hundreds of millions of servers have an operating system that allows experts to treat "nonsense". They are just a path reference for the "mountaineering troops" who attack from the bottom up. Anyone can climb the "high mountain" of systems engineering, and everyone can contribute a little, which naturally forms a system engineering. Everyone has to understand systems engineering and how to apply it in his own place.

Second, the industrial Internet is a systematic project, and it is necessary to do a good job in the cloud base of the industrial Internet. Each industry has its own application platform, such as coal mines, roads, railways, airports. They all have strong personality and industry characteristics, these are systems engineering, we need to do it together and make contributions together. These efforts help various industries gradually move towards the industrial Internet.

工业首先要走向机械化,机械化走向电气化,电气化走向信息化,信息化走向智能化,这时候才有机会把中国的制造业转变成工业互联网,德国目前走在了我们的前面。华为在这个过程中,重要的是做出自己的贡献,例如光传感、光传输、5G连接,以及云底座等。

Third, science is the endless frontier, so don't rush for quick success and instant benefits.

First, if the state attaches importance to theoretical construction, it is necessary to prevent the rush for quick success and instant benefits. For some pure theoretical scientists who cannot create value in a short period of time, we should understand them and let them have a decent life, so that the country can have a future. In my opinion, theoretical science is the "water tower" of our country's future. The Yellow River and the Yangtze River are mother rivers. Where is the mother's milk? On the Qinghai-Tibet Plateau, my mother's milk is the snow water of the Himalayas, and the Yellow River and the Yangtze River all depend on the snow water of Tibet, which means that the country will have a bumper harvest next year. The contribution of "climbing the Himalayas" is very great. The "farms" and "pastures" downstream should distribute some food to the scientists who "climb the snow-capped mountains", and they should not be allowed to starve to death on the "Himalayas."

People who "climb snow mountains" in our company spend more than US $1 billion every year, and the Ministry of Cooperation supports a lot of money to do cutting-edge research for colleges and universities. in fact, it has no commercial value to us, because the theoretical research is completely open and shared to all mankind. It is just that we have more than a hundred thousand engineers and have a very strong digestion ability, so it is possible to give priority to digesting these theories. For example, Professor Arikan of Turkey published a mathematics paper more than ten years ago. Two months after it was published, we found this paper and made 5G Polar codes according to his theory. Therefore, theoretical science can be divorced from practice, and it is not necessary to combine it with practice, nor does it have to solve practical problems.

Second, we should respect intellectuals, respect their creations and inventions, and allow them to make mistakes. Thinking is a shining light. For example, software is a complex system, not only to "pierce the sky" in theory, but also to "take root". It is more difficult to "take root" than to "pierce the sky". All we can solve is the information network, the difficulty of the software is unimaginable, and we have to find a way to solve it gradually over a period of 3-5 years or more.

Third, attach importance to basic education and allow some differentiated education. The current education system may still be in the era of the Industrial Revolution, and the theoretical construction is not enough and is not suitable for the information age. I hope that we can really attach importance to education and allow some "weird" students to grow up. The most important thing for a young genius in our company is his problem-solving ability, not his overall quality.

Fourth, "there are Castle Peak buildings outside the mountains", encourage "nonsense" and freely discuss scientific and technological issues.

Our purpose of running the "Huang Danian Tea House" is to hope that everyone will talk nonsense on scientific and technological issues, as long as they do not involve social science issues. Every scientist and expert comes to Huawei to talk about what you are good at best. it doesn't matter if everyone in Huawei doesn't understand, because it stays in people's hearts and knocks open his head, and he won't close it. If you say you want to get close to our company's research, we appreciate it, but it's not what we need most. We do specific work, the vision is relatively short, what we need most is the small part that we can not see, we want to see "the green hill outside the mountain, the building outside the building". You can see far, and what you are talking about happens to be something we don't understand, so just tell us. When I was young, I listened to a scientist in computer control systems, who said, "computers are used in bus ticket management." I am confused. The computer is numeric. How can it be used in ticket management? There was a question mark in my heart at that time, and only later did I know that it was real. Therefore, we have to establish assumptions about social things for ten or a hundred years, and suppose we start to do these things, and the scientists we bring in will do the base ten or twenty years later, and they will not take part in "killing pigs." we only ask some of them to take part in the battle of "killing pigs" with scalpels.

"Huang Danian Tea House" is an open communication platform. With good internal and external communication, we can connect domestic and foreign universities in the future. You scientists, experts do not need to provide their own computing power, using our cloud platform, including AI algorithms, storage, and so on, can provide you. If you focus on the front end, we can support you on the back end platform.

colabug导航在线提交网站

相关文章

colabug私人网站导航

暂无评论

暂无评论...