2016 年前后,做自动驾驶的团队几乎都会在车顶装一个旋转的激光雷达(lidar)。那东西像个桶,几万美元一台,是当时公认的必要配置。没了它,测距就只能靠摄像头去猜,而猜错的代价是撞车。
马斯克把方案推到桌上:拆掉它。只用摄像头,八个摄像头围车身一圈,剩下的交给深度学习和海量数据。
从 DARPA 无人车挑战赛里走出来的那批机器人专家,听到这个方案基本都在摇头。他们太清楚这道题有多难,也太清楚激光雷达为什么存在。激光一打,距离就是精确的毫米级数字;摄像头能拿到的是像素,得靠视差、靠语义、靠模型去推断,误差天然存在。
莱克斯·弗里德曼(Lex Fridman)在乔·罗根(Joe Rogan)的节目上聊到这段,语气很平。他采访过马斯克两次,真正想讲清楚的是这个人脑子里那套东西怎么运转。「他很牛」这种话,说出来等于没说。
他坐在 CEO 的椅子上,干的是总工程师的活
弗里德曼采访过好几位 CEO,包括谷歌前掌门埃里克·施密特(Eric Schmidt)。那些人的思维重心在商业上,聊市场份额、组织架构、竞争格局、人才流动,路径清晰,也很专业。
马斯克不一样。他会把一个工程技术问题一路往下推,推到物理层面才停。
弗里德曼说,跟这种人聊天是一种享受,因为他自己是个工程师,能听懂对方在推到哪一层。火箭的推重比、电池的能量密度、正极材料的成本、神经信号的带宽,这些东西马斯克能在同一个下午里连着聊,中间不需要切换语言。
这个评价听起来像恭维,其实是个技术判断。绝大部分公司的最高决策者,距离「第一性原理」这个词只有一层修辞的距离,落到具体参数上就断了。他们知道这个词,也会在年会上用它,但你问一句「那具体到我们这个零件的良率是多少」,对话就接不下去。
马斯克那套东西能接到具体参数上。这是弗里德曼反复强调的区别。
第一性原理在四个战场上分别长什么样
SpaceX 那边,火箭的基本问题是材料、燃料、结构、控制。铝锂合金、不锈钢、液氧甲烷的价格摆在那儿,一枚火箭的物料成本就存在一个下限。传统航天工业报出来的价格是这个下限的几十倍,中间那几十倍就是可以被攻击的对象。可回收复用是同一个逻辑的延伸,把一次性消耗品变成资产,把每次发射的边际成本一层层往下压。
Neuralink 那边,基本问题是神经元用什么信号通信,电极能做到多细,植入过程能有多小损伤,读取和写入的带宽怎么提上去。这些问题没有现成答案,但都有物理边界,也就可以被拆开算。
特斯拉的电池那边,主流观点认为电池包成本降不下来,因为正极材料、钴、镍、电芯良率这些约束摆在那儿。马斯克的做法是把整个成本结构拆成原材料、加工、良率、工厂折旧几块,一块一块往下压,同时把工厂本身当成产品来设计,用产线的物理布局去换效率。
自动驾驶那边,主流方案是激光雷达加高精地图加规则引擎。马斯克的做法是摄像头加神经网络加数据闭环,把全部赌注压在一个会自己学习的系统上。
这四个战场有个共同点:每一步都在问同一个问题,这条约束是真正的物理约束,还是行业习惯。弗里德曼转述马斯克那种态度:
别再说这件事不可能了。它为什么不可能?它真的不可能吗?(莱克斯·弗里德曼转述马斯克的口头习惯)
大多数从第一性原理出发的问题,拆到最后会发现,其实没那么不可能。真正难的是后半段:既然理论上能做,那要用什么基础设施把它做出来,怎么让系统从海量数据里学,怎么把学到的东西部署出去,怎么让成本降到普通人买得起。
第一性原理这四个字在网上被用滥了,很多人以为它是一种自信姿态,一种说话方式。弗里德曼描述的那个版本朴素得多。它是一连串具体的追问,追问到最后逼出一份工程清单。清单出来之后,剩下的是苦活,是几百个人的几年时间。
一辆买回家之后还会变聪明的车
弗里德曼认为,特斯拉做过最有革命性的事,很多人都没注意到。
OTA,空中软件更新。你买的那台车,隔一段时间会收到一次更新,跟手机一样。
这件事听起来琐碎。放在汽车工业一百年的历史里,它彻底改变了车的定义。在此之前,车出厂是什么样,开到报废就是什么样。你付完钱那一刻,它的能力就到顶了,之后只会随磨损而衰减。汽车行业从来没把车当成一个可以持续学习的系统来看待,连这个念头都没有过。
一旦车能联网更新,事情的性质就变了。车上跑的模型可以从整个车队收集的数据里学,学完之后几周内推给所有用户。今天这批车的表现,比上个月好一点。下个月再好一点。这种复利效应在硬件行业里从来没出现过。
弗里德曼说,先别管 Autopilot,别管那些宣传。单是「软件可以被更新」这一条,就已经足够革命。而当时几乎没有别的车厂在做这件事。
这里有个容易被忽略的门槛:一个学习系统要成立,必须同时具备三条腿。采集靠车队规模,路上得有几百万台车在跑,才能覆盖足够多的罕见场景;训练靠算力,得有一堆芯片日夜不停地算;部署靠 OTA,学到的成果得能推回每一台车上。
少了任何一条,学习就停在实验室里。传统车厂三条腿都不全,所以它们即便做出了原型,也没法把它变成产品。
推理芯片自己造,训练集群也自己造
弗里德曼提了一个常被混淆的技术区分。
车上的那块硬件,无论哪个版本,做的是推理。模型已经训练好了,它只是接受摄像头的原始输入,输出驾驶决策。这部分硬件是特斯拉自己设计的,从零开始。一家车厂去设计芯片,这个动作本身就极少见,因为芯片行业的门槛摆在那儿,流片一次就是几千万美元。
训练是另一回事。训练是学习的那一端,需要把海量数据灌进去,反复调整几亿几十亿个参数。这一端特别吃算力,也特别吃电力。
行业里的通行做法是买。谷歌有 TPU,全称张量处理器(Tensor Processing Unit),专门为神经网络的训练做过优化,砍掉了不需要的功能,换来更高的能效。英伟达有 GPU,也就是图形处理器,原本是给游戏玩家渲染画面用的,后来大家发现它并行计算的结构特别适合训练神经网络,于是整个机器学习行业都堆在这上面。
一家正常的车厂如果要搞这些东西,会去找谷歌合作,或者去买英伟达的卡。这省事,也省钱,还不用背芯片团队的人力成本和组织风险。
特斯拉的做法是自己造。他们管那套专用训练硬件叫 Dojo。弗里德曼说这个名字听着像开玩笑,「他这个人就爱玩梗」,但东西是认真的:一家车企在自己的机房里搭一台自研超算。
从零开始的思维方式,在这里体现得最完整。买现成的,你永远被供应商的路线图绑住,你的模型得迁就别人的芯片,别人换代你就得跟着重写。自己造,你可以为一个非常具体的工作负载做优化,代价是你得先把自己变成一家芯片公司,还得把芯片公司塞进一家车厂里管。
关于硬核工作,他没打算道歉
弗里德曼说,他从马斯克身上学到的另外两件事,跟代码和芯片关系不大。
第一件是硬核的工作强度。
这个时代的公共话语里,休息是一种美德。所有人都在劝他注意平衡,注意睡眠,累了就停下来,回来之后会有新想法。这些话都对,他自己也承认睡眠很重要,还说自己经常熬夜,作息乱得一塌糊涂,干过不少只靠专注撑着的疯事。
他补了一句:
睡眠很重要,但它只是让工作成为可能的那个东西。大多数人不告诉我的是,苦干比这些更重要,热情比这些更重要。它们应该排在最前面,然后睡眠才去支撑它。(莱克斯·弗里德曼)
这个顺序听起来刺耳。放到工程领域里,它其实很朴素:一个复杂的系统,你要把它一块一块啃下来,靠的是连续几个月的沉浸,反复试错,把同一个问题在脑子里泡到发软。中途切走两个月,回来等于重来,因为那些没写下来的上下文全丢了。
马斯克在招人这件事上也不打算包装。他不会说「来我们这儿吧,环境友好,零食免费,工作生活平衡」。他说的是:你会做你这辈子最狠的一摊活,处理你人生里最重要的问题。
弗里德曼说,他喜欢这种直白,因为它把苦干放在了该在的位置上。那些把「平衡」当成招聘话术的公司,往往在说的是另一件事:我们这儿没什么真正重要的事,所以你可以慢一点。听清楚了反而省时间。
18 小时工作制是冲刺,不是生活方式
罗根在这里插了一句,说出了很多人的顾虑。
弗里德曼经常想一个问题:能不能既要那种工作强度,又有一个家。他很想当父亲,又担心一天干十八个小时的日子会让他没办法给孩子需要的时间。
他自己给了答案。十八小时一天的状态,人生里总有一段时间要经历,可能几个月,可能几年,但不会是永远。那是冲刺。冲完之后把系统、团队、流程都搭稳,然后才可以往后退一步。
问题出在很多人退不下来。这一点他讲得很清楚:像马斯克这样的人,很难找到对等的替代者。公司是他想出来的,技术路线是他定的,他要真的退到后面,得有一个跟他水平相当的人接手。而那些人,早就各自在忙自己的事了。
罗根补了一句同样的意思:很多人一辈子都在冲刺,冲到最后也没停下来过,然后身体先停了。
这段话里藏着一个可以在自己身上用的判断:你得知道自己在冲什么,以及冲完之后要留下什么。冲完只留下一个疲惫的身体和一堆没整理的经验,那这次冲刺的收益就折损了大半。
麦克风打开之前,他是另一个人
罗根有亲身经验。他请马斯克上过节目,聊了几个小时,整个过程都很费劲。
弗里德曼也发现同一个现象。录制之前,马斯克非常放松,爱笑,爱开玩笑,甚至会主动抱人。麦克风一开,整个人就收进去了,回答问题像在解题,一句话从最省字的角度说完就停。
罗根说,那天他是靠喝酒才把马斯克从壳里拽出来的。喝了几杯之后,那个放松的人回来了,话也多了。
弗里德曼听完笑着说,他应该也这么干。
这个小细节有意思的地方在于,一个人在公开场合的表达密度,跟他的思维密度经常是反着的。脑子里同时跑着一百条线程的人,开口时反而会挑最省的表达,因为把话说完整需要先做一次压缩,而压缩是额外的工作量,还要承担被断章取义的风险。少说,是成本最低的防守。
濒临破产、失去孩子,同时还能当个好爸爸
弗里德曼说,翻马斯克的传记会发现一件事:他一直在高压下工作。特斯拉和 SpaceX 都数次走到破产边缘。2008 年那阵,两家公司同时缺钱,一家差点卖给谷歌,另一家靠最后一笔融资续命,圣诞节前后账上几乎见底。他还失去过一个孩子。
这些事拼在一起,让弗里德曼最触动的,是他同时做到了几件互相挤压的事。公司在崩,家里在痛,他还能当一个不错的父亲。
这一点对他有私人意义。他一直在掂量自己的时间表:那种强度之下,还能不能有家庭,能不能真的给孩子足够的时间。
答案落在一个灰色地带。冲刺期你确实给不了,那就别指望在那个阶段给,也不用自责到把整件事否掉。真正的问题是,你什么时候结束冲刺,以及你有没有为结束做准备。
这种脑子为什么换不掉
回到那个接班人问题。弗里德曼认为这里有个结构性困境。
一家公司的核心资产如果是创始人的思维方式,那这家公司就没有真正的第二人选。他既是 CEO,也是所有技术路线的作者。他要退,接手的人得能对他的判断做出同等质量的挑战,不能只会执行。
而这样的人,市面上几乎没有。有这种能力的人,都在做自己的事,没人愿意来给别人的路线图做副手。
弗里德曼说,这件事让他有点失望。这种思维方式太稀有了,稀有到不正常。罗根用了个半开玩笑的说法:人的心智在一条光谱上,马斯克明显在光谱的另一端。
他接着形容马斯克那种注意力分布:对工程问题、交通问题、污染问题、全球网络问题有一种本能的吸引。星链(Starlink)要把互联网铺到地面基站到不了的地方,无聊公司(The Boring Company)在挖隧道解决地面拥堵,这些东西在他脑子里同时运转,而且还都在往前推。
脑子里有一条奔涌的河
弗里德曼说,他跟马斯克聊完之后最深的印象,是这个人的想法多到装不住。
他用的比喻是:像一条奔流的河,他伸手去抓,一次只能捧起一捧水。
这件事有一个童年的版本。马斯克说,他小时候以为所有人都这样思考,脑子里同时转着这么多东西是常态。长大之后他才发现不是,于是他一度怀疑自己是不是疯了。
弗里德曼说,他能对上这个感受。他自己说话也有困难,因为话说到一半,脑子里会岔出十几个方向,每个方向都想深挖下去。他说,做研究的人大概都懂这种感觉,解题的快感就在岔路上。
让他意外的是,一个 CEO 居然还保留着这种状态。管理者的训练通常是把发散收敛成决策,把岔路砍掉,把不确定的部分外包给下属,最后只留一个「做还是不做」。砍得越干净,越像合格的管理者。马斯克没砍,他一直停在发散那一侧,然后靠工程师团队去收敛。
2006 年那条推文
罗根提到一条旧推文。马斯克在 2006 年发过一段话,大意是把接下来的计划列了出来:造一台高性能电动跑车,用赚到的钱造一台买得起的电动车,再用赚到的钱造更便宜的车,同时提供太阳能发电。
十几年过去,那几件事基本都做完了。
这段话值得琢磨的地方,是公开承诺的作用。一个人把路线图钉在墙上,等于给自己套了一个长期约束。做得慢会被嘲笑,做不成会被记账。反过来,这个约束也让所有想参与的人知道要往哪儿走,让投资人知道钱花在什么位置,让工程师知道自己拧的螺丝属于哪个环节。
大部分人不会这么做,因为公开承诺的代价立刻兑现,收益要等很多年。愿意吃这个亏的人,通常是真的打算把它做完。
恨他的人,和爱他的人
罗根说,喜欢马斯克的人很多,讨厌他的也有一大群。
弗里德曼的回答很直接:做成这么多事的人,有哪个是被所有人喜欢的?他举了乔布斯的例子。
他有些做技术的朋友,至今不用苹果产品,原因就是受不了乔布斯这个人。他们认识在乔布斯手下干过的工程师,听到的版本是:被吼,被羞辱,被要求做到不可能的事。做不到就走人,走得很难看。
罗根说,他希望世界能好一点。弗里德曼说,等这些人不在了,大家就只记得伟大了。
这句话有点凉。它的意思是,我们对一个人的评价,往往要等他退场之后才能分层:作品归作品,人品归人品。活着的时候,这两件事搅在一起,谁都看不清,也没人有耐心去看清。
迟到一年两年,本来就是计划的一部分
弗里德曼说,马斯克身上有一个特质,他自己也有,而这个特质特别容易惹人烦。
先立计划,公开承诺,然后交付时间晚一年,晚两年,晚三年。
他给了这套做法的辩护:做极难的事情,超期几乎是必然。而且「过度承诺」本身是必要的工具。你得让身边所有工程师相信这件事一年能做出来,他们才会拼到那个程度,然后才有可能在两年内真的做出来。承诺本身就是推进力的一部分。
还有一层:他自己得真的相信。一个连自己都不信的时间表,传不出那种能量,团队一听就知道是行政任务,拼劲会立刻掉一半。
这个逻辑成立,代价也真实。它消耗的是公众耐心,而且每一次延期都会被人记账,成为「他不靠谱」的证据。弗里德曼说,他不确定那些因为延期而愤怒的人是真的被冒犯,还是本来就看不惯他,正好找到了一个说得出口的理由。
把雾件变成现实
罗根提到 Roadster。有人按全价下了单,二十五万美元,买一台当时还不存在的车。按软件行业的说法,这叫雾件(vaporware)。
弗里德曼的回答是:把雾件变成现实,这件事本身就非常了不起。只要马斯克不死,那台 Roadster 就会造出来。
罗根补了一句:他要是死了,赶紧跑。
话题转到股票。特斯拉是历史上被做空最多的股票之一,而它一直在涨,这件事把很多人搞糊涂了。弗里德曼说他看不懂那些做空的人。
他对股市的态度更狠。他觉得那是赌博,而且是最无聊的一种赌博。他说自己已经把整天聊股票的人从生活里删掉了,因为那些人不会带来任何有意思的想法。
他补了一个例外:真正的投资,把钱投给值得做的事,是对社会很重要的服务。他反感的是另一类人,对真实工程师正在解的物理问题指指点点,用财报和股价去评判一条他们没搞懂的技术路线。这些人听到他的说法会不爽,他觉得正常。
火星名单上都是些什么人
罗根提到 Cybertruck,说那台车荒唐得可爱,外形像从科幻片里直接开出来的。弗里德曼说,只要马斯克活得够久,火星上一定会有人。
至于第一批去的是谁,罗根表示怀疑。他的怀疑有道理。愿意单程去火星的人,跟能在一颗没有空气、没有磁场、重力只有地球三分之一的星球上建起一座城市的人,重合度不一定高。
节目到这里没再展开。留下的问题很清楚:把一件事推到物理极限是一回事,把一群人组织起来在另一个星球上活下去是另一回事。前者靠第一性原理,后者靠组织、制度和运气,还要靠几代人不发疯。
整期节目里,弗里德曼反复回到同一个观察:马斯克身上最稀缺的东西,跟勇气和钱都无关。他把「不可能」当成一个待验证的假设,先记下来,再拆开算,算完就动手。这个习惯在工程师身上偶尔能见到,在一个管着几万人、几家公司的 CEO 身上几乎见不到。
如果想从这段对话里拿走点什么,下面几条可以直接用。
- 挑一个你所在行业里被当作常识的配置,把它写下来,然后逐条追问:这条约束的依据是物理定律、成本结构,还是别人一直这么做。追问到你自己答不上来为止,那里通常就是可以动手的地方。
- 把学习闭环做成三条腿:数据怎么采、模型怎么训、结果怎么推给全部用户。任何一条断了,你的改进就永远停在演示阶段。优先解决部署那一条,它最不性感,也最常被忽略。
- 给团队一个你自己真信的时间表。目标可以说得比现实激进,但你必须能说清楚为什么信。做不到这一点,激进的时间表只会变成笑话,第二次就没人跟了。
- 把高强度当冲刺来管理,别把它过成生活方式。冲之前明确这次要搭好什么,冲完之后必须有退出的节点。没有退出节点的冲刺,最后拼的是身体。
- 在公开场合把路线图钉下来。写清楚做哪几件事、按什么顺序、为什么是这个顺序。这会让你难受好几年,也会让对的人主动找上你。
- 想清楚你要找的是能接替你位置的人,还是能跟你一起解谜的人。前者几乎找不到,后者可以培养,而且更值得花时间。
- 少花时间评价别人怎么做技术,多花时间读一手细节,比如一份芯片的架构说明、一块电池的拆解报告。判断力只能从细节里长出来,从评论区里长不出来。
0:00the Joe Rogan experience also on the podcast just like you got just to talk to you all mosque meet him talk to him in person and realize that there's you know there's people in this world they can make the impossible happen you interviewed him as well yeah twice yeah tell me what does experience like for you so you know there was it's quite
0:24incredible in the sense that he is a legit engineer and designer which is like a pleasure for me I've talked to a few CEOs talked to Eric Schmidt just the CEOs and they're a little bit more business oriented Elon is really really focused on the fundament like the first principles to like the physics level of the problems that are being solved whether that's SpaceX with the fundamentals of rocky reusable rockets and and you know going into deep space and colonizing Mars whether that's a neuro-link sure the getting to the core the fundamentals of what it's like to have a computer communicate with the human brain and with Tesla on the battery side sort of
1:03saying he he throw away a lot of the conventional thinking about what's required to build first of all an appealing car electric car but also one that's has a long range that's something I don't know as much about but that on the AI side just I mean he boldly said from scratch we can build the system ourselves in a matter of months now a couple of years that's able to drive autonomously and most people would laugh at that idea that the most roboticist said no from the DARPA challenges most of them know how hard this problem is he said no no we're going to we're not only gonna throw a lidar which is this laser based sensor we're gonna say cameras
1:44only and we're gonna use deep learning machine learning which is learning based system so it's a system that learns from scratch and we're going to teach it to drive from eight cameras and so on so just talking to somebody like that was the the not the fact that he thinks like that I think it's just fun to talk to people like that I don't meet them off and let's say I don't know stop this [ __ ] of thinking that this task is impossible let's say why is it impossible is it really impossible will you find out when you start to think about most problems from first principles is that it's not actually impossible and then you have to think okay so how do we make it happen how do
2:27we create an infrastructure that allows you to learn from huge amounts of data so one of the most revolutionary things that Tesla is doing and hopefully other car companies will be doing is the over-the-air software updates just like the update that you got the fact that just like on your phone you can get updates overtime means you can have a learning system a machine learning based system they can learn and then deploy the thing you learned over time and do that weekly that sounds like maybe trivial but it's come nobody else is doing it and it's completely revolutionary so cars once you buy them they don't learn most cars Nestle learns that that's a huge thing
3:08forget about s autopilot all the stuff just the fact that you can update the software I think there's a revolutionary idea and then they're also doing everything else from scratch this is this first principles type of thinking the hardware so the the hardware in your car I don't know when you got the Tesla but it should be hardware version two but that hardware performs what's called inference so it's already trained it's already learned its thing and it's just taking in the raw sensory input and making decisions okay they built that hardware themselves from scratch again ballsy move now they're building what they're calling again he's such a troll
3:49but they're calling dojo is the the name of the the specialized hardware for training than your own though so training the models what training is is the learning side of it so they're building their own like supercomputer at Google has a TPU to improve the training tape you what's that stand for tensor processing unit it's the same thing as the more general Nvidia has graphics processing unit GPUs that all the Nerds all the people like me have been using on for machine learning to train you all networks it's what most also gamers use the video games right but they have their this nice quality that you can train huge neural networks on them okay TPU is
4:31is a specialized hardware for training neural networks GPUs allow you to play video games and train your own networks TPU is clean some stuff up to make it more efficient energy efficient more efficient for the kinds of computation y'all in networks need Google has them a bunch of other companies have them you know most most car companies would be like okay let me partner with somebody else to you from Google to use their TP use or use invidious GPUs Tesla's building you from scratch so that kind of from scratch thinking is is incredible and the other two things I really listen
5:11where the like about musk is the hard work we live in a culture like so many people like I often don't sleep I do crazy [ __ ] in terms of just focus stay up night sometimes and now often people recommend to me the you know balance is really important and taking a break is important you know that you've innate yourself you return to it with fresh ideas all those things are true sleep is important yet the people on the podcast tell you how important sleep is but what most people don't don't advise me is hard work is more important passion is more important than all of those things like that should come first and then sleep empowers it rest and powers at
5:54rejuvenation powers especially in the engineering disciplines hard work is everything and he's sort of unapologetically about that it's not like a come come to us come work with us it'll be a friendly environment with free snacks it's like you're gonna work the hardest you've ever worked on whether you agree with him or not on the most important problems of your life ok I like that kind of thinking because it emphasizes the hard work the the other part it was in terms of meeting him in person I don't know if you got to interact with that off because when he was on mic with you he was very he was very kind of it was hard to bring it out of him yeah in person before that
6:40was very jovial and friendly and Huggy he's great yeah and then once he got on the microphone I was like oh this is heavy lifting yeah bring this out of him so then we started drinking drinking and then oh yeah it helps a lot and then once the drinking you know then then I got to see who he is yeah I shouldn't I should have done that but no drinking yeah no I mean the the thing that's really interesting is he's gone if you look at his biography like the kind of stress he's been under in terms of he's been at the brink of losing his companies several times yes and he you
7:17know he lost a child and he's just he's a well that's the other thing inspired me as it's that he can be a good dad while running so many companies because like I was often wonder about you know the kind of hours I pull on what I'm doing can I have a family can i really give them because I'd love to be a father and can I have a family can I be a good part like it's very very very very difficult if you work in 18 hours a day yes to give your kids at the time that they need but it's possible not not 18 oh there's always I believe there's in life day months maybe years that you have to do the 18 hours a day but not always it's time for right do the Sprint
7:58Sprint's yeah and then establish everything and then sit back but the problem with a lot of guys like him is first of all it's very difficult to find a replacement for the way he thinks right so if he's if he's a CEO of these companies and he's the one who's the mastermind behind all these things and then he wants to step back finding a commensurate replacement is insanely difficult because most people who would be a potential replacement are already off doing their own [ __ ] and there's not many people like him I know that's the interesting that's a actually the disappointing thing to me is that he's a his kind of thinking is a rarity I'm not yes I'm not sure why that is
8:39exactly well he's I joke around about it but it I think there's a there's a spectrum of evolution and his mind is clearly way more advanced in my mind there's there's something going on in his mind in terms of his attraction to engineering issues solutions to global problems solutions to traffic problems pollution problems all the other things that he's the Internet he's trying to put he's trying to give the world Internet I mean he's got all these things going simultaneously and one of the things that I got out of him when I was talking to him was that he almost has a hard time containing these ideas that are just pouring out of his head like a raging river like he's trying to
9:27catch and fulls of water and there's raging river of ideas that's going through his head you know and when he described his childhood that he thought that everybody was like that and then as he got older he you know thought he was insane yeah I didn't I've I can relate to that I've I have I'm trying to learn how to talk but I have trouble talking because it's like a million ideas running in my head I get anything you say I'll immediately start there's these like weird tangent they go off and I want to start thinking about them is that true with a lot of people in your line of work and I think so I think that's kind of puzzle solving like that's what comfort is I'm just
10:04surprised that a CEO is able to continue being that kind of puzzle solver did you see that tweet that he made about his plans like he put a tweet up and I think it was 2006 and then it's he's essentially done all those things he's done all those things now the thing is most people I did so a lot of people love you know musk but there's a quite a large community of people that don't love him so much well that's always the case and I don't know I don't anybody great I don't I don't know if that's always the case when is it not the case who does who accomplishes as many things as that guy does where everybody loves them it's a difficult I mean I'm not a historian I
10:48could say Steve Jobs a terrible example so many people hated that guy so many people hated that guy I have personal friends that are involved in technology that wouldn't use Apple products cuz he's such a [ __ ] sure they didn't want to have anything to do with him they knew people that were engineers under him they said it was horrible and mean and just required so much would scream at people and insult them and you know he had these ideas in his head that he needed to get done and if you couldn't work the hours that you needed to do what he wanted to accomplish he would treat you like [ __ ] you know you're right and I just wish the world was better I think because I like with
11:27all people like that like we Steve Jobs and with Elon Musk when he dies people will always yeah you'll remember the greatness right yeah so that's that's how it seems to work it's just sad that you can't celebrate that currently but I do think there's one particular aspect of his personality that I also share that pisses people off really bad which is like you said he had a plan but he's laid down that plan he's promising things and then he keeps being like a year or two or three late right and that really I don't know if it actually angers people or if people that already don't like you use that as a thing to say why they don't like you but it's
12:06certainly a thing that people say a lot yeah and but I think that's an essential element of doing extremely difficult things is over-promising and trying to over-deliver that's the whole point right is to to say to make all the engineers around you believe that is doable in a year that's essential to do it in two years that's like that that kind of and truly believing it is it seems to be essential didn't he have people pay full price for that roadster like you got on a list a time yeah yes so you paid a quarter of a million dollars for a car that's essentially vaporware yeah but the thing so I don't know there's a little bunch of financial people to get like mad at ya know yeah
12:49like there's investors you know it's like it's I think it's the most shorted stock in history yeah so that keeps kicking ass I don't confuse is the [ __ ] out of people Oh to me the stock market is the most boring thing ever in people it's it's a it's a gambling yes and so you trying to say you're an expert in investing in the stock market that I blocked I removed those people from my life because they don't say anything any interesting ideas I said it but the you know when you're doing legitimate investment yes that's that's a really important services society but if you're commenting on the fundamentals of engineering problems that real engineers are trying to solve that's not
13:32interesting to me the so it that kind of stuff upsets I think financial folks but the beautiful thing is when you have people by vaporware and you bring that vapor word to reality that's the amazing thing yeah that he will definitely bring that roadster to reality if he doesn't die that roadster will happen yeah if he dies bail out now same with him that's the same with that insane cyber truck yeah there's our truck is [ __ ] awesome I'm so ridiculous if he lives long enough you better believe there's humans being put on Mars whether it's him or he gets everybody else see that one I'm skeptical of just the type of people that are gonna want to go [Applause]
资源与出处
- 视频在 YouTube 收看
- 档案silicon-valley.tech 档案页本页视频与文稿数据来自 Elon Musk Archive
- 精读本篇精读 Markdown 版