2020 年,联盟号飞船还在飞。
这个型号的设计工作启动于 20 世纪 50 年代末。如果把这个消息告诉当年主持设计的总设计师科罗廖夫(Sergei Korolev),他大概会觉得你在开玩笑。马斯克在 Satellite 2020 大会上说,科罗廖夫"估计在坟墓里翻身"。
11 年前,马斯克在同一个场合被人问"搞这些到底图什么"。他回答:用美国造的火箭、从美国造的发射场,把人类重新送回太空,最终让人类永久定居其他星球。
11 年后,载人龙飞船(Crew Dragon)停在卡纳维拉尔角,准备第一次载人飞行。距离 SpaceX 成立,过去了 18 年。
主持人开玩笑:18 年,孩子都该上大学了。
马斯克接了一句:还没送走呢,但确实是很长一段路。
然后他话锋一转,说出了这场访谈里最重要的一句话:
"现在最让我担心的事情是,除非我们大幅提高创新速度,否则月球基地和火星城市都没戏。"(马斯克)
载人龙飞船值得庆祝,但要小心局部最优
马斯克先交代了龙飞船的进度。它已经去过空间站又回来了,之后做了海量测试,把各种边界情况挨个推了一遍。"测试数量多到荒谬",他专门向 NASA 和 SpaceX 的龙飞船工程师团队致意。
紧接着他给这个成就降温。
龙飞船本质上只是一个近地轨道运输工具。它能送几个人上去,成本依然很高。技术上可以让它绕月飞行,但飞船太小,没多少人愿意这么干。
他用了那个词:局部最优(local maximum)。
航天飞机就在局部最优里困了很多年。联盟号也是。
"说句实话,联盟号为什么还在飞?科罗廖夫估计在坟墓里翻身。它设计于 50 年代。如果你告诉科罗廖夫和其他人,2020 年我们还在飞联盟号,他们会说:这太疯狂了。"(马斯克)
马斯克承认联盟号是可靠的飞行器,但"该往前走了"。
这里藏着一个判断标准:完成一次载人飞行值得庆祝,但要警惕把阶段性成果当成终点。航天飞机当年也是这么一步步被困住的,飞得越久,越难承认该换架构了。
只有一件事真正重要:完全且快速可复用的火箭
观众投票选出的最热门问题:拓展人类在太空的存在,最大的挑战是什么?
马斯克的答案没有铺垫:只有一件事重要,完全且快速地可复用火箭。
而且必须够大。
"就像你不会用一堆带舷外机的小船横渡太平洋。要运真正的货物、走真正的距离,就得用大船。所以我们需要一个相当大的、完全可快速复用的火箭。没有这个,我们哪儿也去不了。"(马斯克)
猎鹰9和龙飞船的技术架构已经接近渐近线。意思是再改也不会有本质提升了。搞一个"Block 6"版本没有意义。这两款是运营中的成熟产品,新的技术开发全部压在星舰(Starship)上。
星舰的复用指标:着陆后一小时内再次发射,零名义工作。定期维护可以有,小毛病可以处理,但常规情况下唯一需要更换的消耗品是推进剂。就像商用飞机,你唯一预期会定期更换的是燃油。
节奏目标是一天飞三次。
大部分时间花在等轨道。除非在赤道发射,飞船得绕几圈才能让地面轨迹重新经过着陆场。绕三到四圈,然后一小时完成其他所有事。
轨道加注是去火星的前置条件
除了完全复用,还有一件基础性的事:轨道加注(orbital refilling)。
"你得在地球轨道上加满油再出发。这样你所有的质量分数、所有的速度增量都可以在地球轨道上重新配置。这是去火星的关键。"(马斯克)
有人问利用月球资源,比如水、氧气、土壤,能不能支撑火星任务。
马斯克的回答是:除了轨道加注,暂时不考虑。
月球的位置也很尴尬。
"月球不在这条路的中点上。用它就像说,你要横渡大西洋,也许先去冰岛。去一趟可以,但这不是一个必须停靠的站。"(马斯克)
为什么星舰用不锈钢,这笔账值得算一遍
这是整场访谈信息密度最高的部分。
星舰一开始打算用先进复合材料。成本每磅 60 美元左右,每公斤 60 多美元,可能到每磅 130 美元。储箱要铺 60 到 120 层。进度慢得让人绝望,钱烧得飞快。
换铝锂合金。猎鹰9的储箱就是铝锂合金。问题是锂的活性让它很难焊接。
那就用钢。
马斯克列了一串理由。
钢非常好焊。
钢不用喷漆。喷漆在低温下会弯曲开裂,漆"恨不得立刻掉下来"。用特种漆?特种漆还有别的问题:飞行器超音速飞行时会产生静电积累(triboelectric charging),用错漆可能会电到自己。而且漆有重量。航天飞机的外储箱原本喷成白色,后来为了减重干脆保持橙色。
马斯克提到了早期宇宙神(Atlas)火箭,用的是 301 不锈钢。他专门致敬了一个人:Charlie Bossert。马斯克说这哥们被严重低估,"他太牛了",读他的东西很有收获。
301 不锈钢不新,但马斯克算了一笔账。材料手册上通常不会强调,301 在冷作硬化后强度大幅提升,在低温下强度还会再大幅提升。两个效果叠加,有效强度重量比和先进复合材料差不多。
复合材料这边有个常见的新手错误:只看材料数据表,忘了算减缩系数(knockdown)。复合材料本质上是"纤维加胶水",你得留冗余。设计需要四层碳纤维,实际可能得铺五层六层,因为要防某一层损伤、要算脱粘的最坏情况。算下来,复合材料的实际减缩比金属结构更大。
所以人们经常高估碳纤维。
还没算喷漆。钢不用喷。
也没算耐热。可复用飞行器会发热。复合材料大概 150 到 200 摄氏度还舒服,250 摄氏度就开始危险,得上特种树脂。钢在 500 摄氏度全天候没问题,短时间 1000 摄氏度也不慌。
这一条直接改变热防护设计。钢的背风面可以完全不要隔热瓦,因为结构本身扛得住辐射加热。迎风面的隔热瓦可以减薄,因为瓦的厚度取决于瓦背面、也就是安装到主结构那一面的温度。主结构耐热,瓦就能薄。
马斯克的比喻:隔热瓦像烤箱手套。你的手能扛多热,决定了手套要多厚。手不怕烫,手套就能薄。
"算下来,一艘可复用的钢制飞船,总质量比你能想象的最先进碳纤维飞行器还要轻。"(马斯克)
然后他补了一句关键的:这个发现是意外。
"这听起来像什么了不起的洞察,其实完全是个意外。我们当时在复合材料上进展太慢,我说不行,再这么慢公司要破产了,直接用钢。设计必须围绕解决问题,否则你会花太多时间琢磨。"(马斯克)
由此引出他那句口头禅:管理靠押韵(management by rhyming)。
"如果进度表很长,说明设计错了。"(马斯克)
他还提了一句,接下来一两个月可能开始换合金,调整成分。2020 年做出来的钢,没道理比 50 年代差。
星链不是来干掉电信运营商的
SpaceX 的终极目标是让生命多行星化。但发射火箭、发射卫星、给空间站提供服务,这些业务的天花板大概一年 30 亿美元。
宽带业务的天花板高一个数量级,大概一年 300 亿美元。
即便如此,星链在这个市场里的份额可能还不到 5%。马斯克反复强调:星链对电信运营商不构成巨大威胁,反而会帮到他们。
逻辑很直接。星链服务的是电信运营商最难覆盖的那 3% 到 4% 客户。铺光纤不划算的地方,建基站也不划算的地方,或者根本没有网络的地方。
5G 适合高密度场景,华盛顿特区、纽约、旧金山这种地方。5G 不适合乡村,覆盖距离不够。稀疏环境里 5G 很吃力。
他甚至开玩笑说,星链会把传统运营商最难啃的那部分客户接走,替他们卸掉一块负担。
用户体验的目标是什么?
延迟低于 20 毫秒。这个门槛够打竞技级快响应视频游戏。带宽足够看高清电影、打游戏、做想做的事,感觉不到卡。
瓶颈在蜂窝尺寸。天基系统的"蜂窝"覆盖范围巨大,带宽密度上不去。洛杉矶会有一些用户,但不可能有很多,因为单个蜂窝的带宽不够。
地面终端长什么样?马斯克说,像个"棍子上的 UFO"。第一版终端带执行器,能自动调整指向精度。安装不需要专业人员。
"盒子上就两条指令,顺序随便:指向天空,插电。反过来也行。真的即插即用。"(马斯克)
卫星反光这件事,马斯克赌的是零
2020 年星链最大的争议之一,是卫星反光影响天文观测。
马斯克的判断很干脆:
"我确信我们不会对天文发现造成任何影响。零。这是我的预测。如果超过零,我们会采取纠正措施。"(马斯克)
他解释了为什么早期看起来那么糟。卫星刚发射时还在翻滚,没稳定下来,会闪烁。而且那时候在升轨,轨道高度比最终低,反射方式不一样。
"现在卫星都在轨道上了。如果有人能告诉我它们全都在哪儿,我会很佩服。我还没遇到一个人能说清楚所有卫星的位置。连一个都没有。所以这事没那么大。"(马斯克)
SpaceX 在和科学界资深人士、资深天文学家合作,减少反射。具体动作:把相控阵天线涂成黑色而不是白色。研发遮阳罩,因为某些太阳角度下会有反射。
马斯克还补了一句:从审美角度也不该有影响。
关于彭博社报道的"星链从 SpaceX 分拆",马斯克连说几个零。完全没在考虑,得先让这东西运转起来。
他顺手给低轨通信星座行业泼了一盆冷水:
"低轨通信星座,猜猜有多少家没破产?零。铱星现在还行,但铱星一号破产了。ICO 破产,全球星破产,Teledesic 破产。还有一堆没走多远也破产了。所以先让星链进入非破产类别,这是个大台阶。"(马斯克)
SpaceX 会不会因为自家星链,对发射竞争对手的卫星有顾虑?
马斯克:完全没问题,还给好价钱。世界对带宽的胃口看起来是无止境的。
软件会腐烂,旧代码该退休就退休
主持人提到游戏设计师 Jonathan Blow 的一个观点:技术会自然衰退,因为技能会消退,软件会退化。去火星的路上,软件怎么办?
马斯克同意。
软件在任何技术里的占比都在上升。特斯拉汽车"本质上是一台带轮子的笔记本电脑",可配置程度极高。
"全自动驾驶,唯一的限制因素是软件。硬件早就齐了。最后一块拼图是升级计算机、增加算力,那是几年前的事。"(马斯克)
然后他说到自己作为一个 iPhone 用户的体验:最近几次软件更新不怎么样,把他的邮件系统搞坏了。他说这挺要命的,邮件是很基础的功能。
"技术不会自动变好。人们习惯了手机一年比一年好,但这不是自然规律。"(马斯克)
大量软件的作者已经退休,甚至不在了。怎么修?
马斯克的观点:需要更多聪明人写软件,但不能只修旧问题。旧代码库该退休就退休,不能永远维护。维护难度到某个点会指数上升,该重写就重写。
给 20 岁的自己一张清单
斯坦福研究生 Jaden Zundel 问:回顾航天生涯,最意外、最没想到的挑战是什么?如果能回到过去对 20 岁的自己说话,会做什么不同的事?
马斯克说,会给一张清单,上面列着所有即将犯的蠢事,请千万不要做。清单会非常长。
他提炼出几条。
第一,简化设计。很多次他们把设计搞得太复杂。检验标准就一条:这东西要多久才能飞起来。如果要很久,就别做,换个方案。
第二,别优化不该存在的东西。这是聪明工程师犯蠢的典型方式。大学训练你回答所有问题,教授出题,你必须答,不答就没分。但现实中很多时候你该说:这个问题本身就是错的。
"你的设计在某种程度上一定是错的,而且错得比你想象的多。你的目标是让它随着时间少错一点。"(马斯克)
第三,别接不该接的资源。资源会自动填满所有可用空间,有时候得学会说不。
他举了猎鹰9的例子。铝锂合金储箱,非加压结构用碳纤维复合材料。热膨胀系数(CTE)差异巨大,一边是从室温到低温会大幅收缩的金属,一边是几乎零膨胀的碳纤维。接头成了噩梦。解决办法是超贵超重的螺栓,像悬臂梁一样承受弯矩,把载荷传进级间段。这些螺栓随时想被剪断或者折断。
"直接用连续的金属结构,这是显而易见的,好得多。"(马斯克)
猎鹰1的原始团队,包括整流罩、储箱、发动机,全部加起来一百多人。现在 SpaceX 大约 6000 人。产品要尽可能简化。
星舰真正的难点是产线
有人问星舰在博卡奇卡(Boca Chica)的进展。马斯克说,当天迟到就是因为在那边的产线上干活,"挺酷的,我喜欢那地方"。
他把话说得很直:
"设计火箭本身没那么难,尤其是不可复用的火箭。你可以直接看书,书上会告诉你具体怎么做。难的是把它造出来,哪怕只造一枚都难。而造一条产线,难度至少是一千倍。"(马斯克)
紧接着是一句判断:
"制造和生产整体上被低估了,尤其是在美国。我们应该花更多注意力、更多心思在制造上。这是一份实打实的、诚实的工作。"(马斯克)
大学不是为了学习
一位叫 Julie 的听众问:马斯克说过大学学位不该那么重要,特斯拉的招聘启事也体现这一点。但航天业、卫星开发领域的很多职位要求至少本科学历,最好硕士。学费越来越贵,奖学金没怎么涨,怎么让大学更负担得起,同时保证毕业生和员工收入,还要让弱势群体有渠道进好大学?
马斯克的回答分两层。
第一层:你不需要大学来学习。想学什么,基本都能免费学到。问题从来不是获取知识。
第二层:大学有价值,但价值不在教学。一是证明你能坚持干完一堆烦人的作业,能熬过去,能交付。二是让你和同龄人待一段时间,而不是直接进职场。
"所以大学基本上是为了好玩,以及证明你能做完你的杂活。不是为了学习。"(马斯克)
他顺势说,要确保特斯拉的招聘不写"要求大学学历",那很荒谬。
"但我们会要求卓越能力的证据。你要做非凡的事,就得有非凡能力的证据。我不认为上大学是卓越能力的证据。理想情况下,你辍学了,然后去做成了点什么。"(马斯克)
盖茨很聪明,辍学了。乔布斯很聪明,辍学了。拉里·埃里森很聪明,辍学了。莎士比亚可能也没上过大学。
他希望去火星的时候自己还没死
主持人问:你的使命是去火星,这是长期项目,我们这代人可能看不到人类定居火星。你怎么培养下一代领导力,让他们接着干?
马斯克的回答半开玩笑半认真:
"我希望人类去火星的时候我还没死。那会是个很好的结果。但如果我们不提高进步速度,我肯定在去火星之前就死了。这是我的 aspiration:去火星的时候还活着。"(马斯克)
他重复了开头的判断:光是把第一批人送上轨道就花了 18 年。按过去的趋势,他肯定看不到火星。所以必须大幅提高创新速度。
SpaceX 有很多优秀的人,有时候他会想,怎么才能最好地发挥这些人的才能,因为现在很多才能没被用好。
这场访谈从头到尾其实在讲同一件事:进步不会自动发生。火箭不会因为时间推移自己变得便宜,软件不会因为用户多就自动变好,产线不会因为设计图漂亮就自己跑起来,一个人也不会因为上了大学就变聪明。每一件事都需要有人具体地、笨拙地、反复地把它往前推一点。
马斯克自己给出的检验标准,简单到近乎粗暴:如果这件事要很久才能飞起来,那设计就是错的。换个设计。
这句话放在火箭上成立,放在任何一件被拖了很久的事情上,大概也成立。
可以立刻用起来的几条
- 给手上每个项目算一笔"飞行时间":从动手到能真实验证要多久。超过几个月的,先怀疑设计,别急着加人加钱。
- 把团队正在优化的所有事项列出来,逐条问一句"这个东西本来就该存在吗"。答不上来的,砍掉。
- 给旧代码库设一个明确的退休日期。维护成本到某个点会失控,重写比续命便宜。
- 招聘启事里删掉"要求本科及以上",换成"要求卓越能力的证据",并说清楚证据长什么样:做成过什么、怎么做到的。
- 要做连接结构,尽量用同一种连续材料。别把热膨胀系数差一个数量级的两种材料硬接在一起,接头会成为长期的麻烦来源。
- 评估任何硬件项目时,把产线难度单独算一项,权重给足。设计和量产是两种完全不同的能力。
- 新业务先想清楚服务谁。挑现有玩家最难覆盖的那一小块,比正面抢核心用户更容易活下来,也更容易让别人愿意跟你合作。
0:08[Music] you [Music] all right hi everybody welcome to satellite 2020. sorry we got off to a late start dc traffic is a killer but my name is jeffrey hill i'm the conference chair and i'm here with elon musk chief engineer and founder of spacex elon thank you so much for being with us thanks thanks for thanks for having me i guess it was a 11 years ago that we met yeah and and when we met 11 years ago we were talking about falcon 1 falcon 9 at the time and when i asked you what's what's the point of all this you said to send human beings back into space on a u.s built rocket from a u.s built facility so that we could permanently resettle other other
1:48planets and here we are we're right on the the brink here of sending humans back into space the the crew dragon is in cape canaveral how do you feel that you're on the doorstep here yeah it's great that we're about to launch people to to orbit it's been a long time long road 18 years yeah kid could be a kid could be in college by now is it like sending your kid off to college well we haven't done it yet but it's a long time you're you're packing you're packing the bags you're ready to go can you talk about the the talk about the the road from the space shuttle to the crew dragon talk about the the what are some of the challenges that you faced in creating a
2:41a human rated space spacecraft for for human space flight what were some of the challenges that you encountered along the way well the thing that the thing that concerns me most right now is that unless we improve our rate of innovation dramatically then there is no chance of a base on the moon or a city on mars not i'll be like yeah
3:05this is my biggest concern i think crew dragon was we we've already taken it to the space station and back a lot of people want to wear that we just had like a basically a dummy you know so and then since what then we've done massive amount of testing you know pushing all the corner cases and just a truly ridiculous amount of of testing it's like definitely hat is off to the dragon engineers and sporting team at nasa for going through a truly staggering number of tests now now that that said
3:53dragon really is just a low earth orbit transport vehicle it's it's really just it's capable of taking a few people at what is still a very high cost to earth orbit i mean technically we could send people around the moon on dragon but i'm not sure we'd want to it's too small you know so
4:23it it's good to get this done but it's i i think we need to be very careful of getting stuck in a local maximum and you know the space shuttle was something that was really stuck in a local maximum for a long time and yeah we don't want to be that situation i mean frankly why is why does soyuz still fly i mean carlov is probably turning in his grave right now right interesting it's the string question it was designed in the 50s yeah right right if you told them if if you told carlos and the other guys that they'd still be we'd still be flying soyuz in 2020 they'd be like that's crazy yeah we are so we don't want to be that situation you know it's a solid vehicle it's just
5:14like it's time to move on right right and so we you know i know we started late i sourced a lot of these questions from the from the public in the audience we're also going to do a q a here so i'm just going to jump right into the questions that we received so i mean the most popular question we got was like what are the greatest challenges or the biggest challenges we face in expanding our presence in space and exploring and eventually resettling new worlds there's really just one thing that matters that is a fully and rapidly reusable rocket that that's the one thing that matters and it needs to be reasonably big
5:58or your payload to non-payload ratio will be kind of whacked you know what won't be good so just like you wouldn't want a super tanker growing like you when you you know container ships you have a container ship with thousands of containers you don't you know have like a bunch of tiny ships with little outboards on them cruising across the pacific that would be silly so you have big shifts when you want to go long distances with serious cargo so we need a fairly big but definitely rapidly and completely reusable rocket this is the fundamental thing without that we're going nowhere and what level of reusability is spacex actively pursuing for
6:46for falcon 9 for dragon for starship i think falconei and dragon have their asymptoting that their tech their technology architecture is asymptoting meaning like it it really would not make sense to have a block six falcon line you know from where we are right now it just doesn't make sense that's why we have a big focus in terms of new technology development on starship for falcon and falcon and dragon are kind of like operational vehicles at this point so they're they're good products they're operational but but there's not really
7:36we need a whole new architecture and that's what starship is about and starship needs to be fully and completely reusable and rapidly so i mean it's it's being designed for about and you know to be really relaunched in hour after landing with zero nominal work like if you could have scheduled maintenance you or you could have like something like a spork issue just like commercial aircraft but you're expected the only thing you expect to change on a regular basis is propellant and it's got to be fast so yeah now that for the ship you you you've got to wait if unless you're launching dewey's from the equator you've got to have figure out some way
8:29to get the ship orbital ground track to pass over the landing site otherwise you're too far away so the ship maybe it might take you know three orbits four orbits maybe to get back over the the launch site but i think we want to aim for a capability of three flights a day for the show wow most of which is taken up with getting the orbital you know ground track to come over the launch site yeah and then an hour for everything else and you know everybody's in interested in in the the mission to mars planetary resettlement we're talking about reusability for the launch vehicle what are your thoughts on in-space resource utilization for example water oxygen soil
9:28from the moon perhaps maybe to go to mars do you have any plans to utilize resources in space for the mission to mars nope i mean apart from orbital refilling i think that's very important so you've got to so there's one x besides a fully fully rapidly reusable rocket you need to also have orbital refilling or re-tanking that's got to be that's fundamental because then you can essentially recruit all of your mass fraction delta v in earth orbit you can leave with full tanks and it could be from immediate low earth orbit or you know something that's maybe elliptical or something like that if you want to go higher energy but that's that's crucial for getting to
10:25to mars the moon is neither here in all there i mean using the moon would be like okay if you want to cross the atlantic maybe you want to go to iceland probably not you know but you know to visit sure but you know it's not like a mandatory stop you know so and also for the mission to mars what advancements we talk a lot about hardware and physics problems and then what about advancements in software you know the reason i bring this up recently are you familiar with the game designer jonathan blow he referenced you in a keynote he was given and he said that you you would talk about technology naturally decays because skills naturally fade and one of
11:16the things he identified was a decay in software degradation in software is this something we have to address in in doing something like going to mars and since all of this stuff runs on software well software is an increasing part of any piece of technology i mean tesla the car is extremely configurable it's basically like a laptop on wheels so software matters enormously there and really for example for full autonomy the only gaining factor is software the hardware is all there that's required it has been for last couple years well the final piece of hardware was upgrading the computer to have more compute power
12:07so software is extremely important the point you're leading to which is that you know what i was referring to is technology does not automatically improve right people are used to the phone being better every year well and i'm an iphone user but i think like some of the recent software updates have been like not great certainly feeding it to that point it's like broke my email system like what the this is like quite fundamental so yeah there sure is a lot of software out there and some of it's like the people that wrote it are retired or maybe dad you know so like now how do you fix it it's going to be an issue i think we definitely need a lot more smart people
12:59working software and not just troubleshooting old problems don't just travel shooting old problems it's actually very important to retire old code bases and not just maintain them forever because the difficulty of maintaining them becomes extremely high at a certain point you just got to redo the code base so we'll come back to the mars mission because i know we've got some audience questions on that you know we're at a satellite conference so i'm going to ask you some questions about satellites starlink what's the long-term vision for starlink how do you see the role of starlink as it relates to mobile broadband and 5g sure so i mean the whole purpose of spacex is really to
13:48help make life multi-planetary and then but the revenue potential of launching rock launching satellites servicing the space station what not that's you know taps out around three billion dollars a year but i think providing broad brand is is more like an order of magnitude more than that probably 30 billion a year as a rough approximation and we're still like probably below five percent at that point so it's not like i want to be clear like sound like star lake is some huge threat to telcos i want to be super clear it is not in fact it will be helpful to telcos because starlink will will serve the hardest to serve customers that telcos otherwise have trouble
14:37doing with with landlines or even with with cell radio stations you know with cell cell towers 5g is is great for high density situations like being here in dc or you know new york san francisco that kind of thing 5g is great for high density situations but it's actually not great for the countryside you know for for rural areas it's it's not it's not great you need you need range and so any any kind of sparse environment 5g is really not not well suited but it's great great for in it for for city density situations so starlink will effectively serve the i don't know three or four percent hardest to reach customers for telcos or people who simply
15:26have no connectivity right now or the connectivity is really bad so i think it will be actually helpful and take a significant load off the traditional telcos okay and i was i was going to ask you what customers you know were ideally suited for starlink but i guess since you mentioned that it would be it would be these three to four percent at the very like that at the very edge what is the customer experience like then for those people and what's the cost of of acquiring those services well it will be a it should be a good experience because it'll be very low latency and we're targeting latency below 20 milliseconds so somebody could could play a fast
16:15response video game at a competitive level right like that's the threshold for the latency so then and bandwidth the bandwidth is a very complex question but let's just say somebody will be able to watch high def movies play video games and do all the things they want to do without noticing speed right and then the the challenge for anything that is space based is that the the size of the cell is gigantic you know so it's like i said it's great for for very low to maybe maybe mediums sort of sparsity situations but it's not it's not good for high-density situations we'll have some small number of customers in la but we can't do a lot of customers in la because the bandwidth
17:13per cell is simply not high enough what is what does the equipment on the ground look like for this what are yeah so the the ground equipment just looks like well i think it's like saying it looks like a little looks like a ufo on a stick so at least the version one of the user terminal will actually have actuators on it so that it can it can improve the pointing accuracy so you don't have to it's very important that you don't need a specialist or something to install the goal is that this the the instructions on the box will there's just two instructions and they can be done in either order a pointed sky in
18:02and we do it either order sequence doesn't matter and it will work plug and play literally but also point at sky you can't see the satellites you can't see the satellites you can't see you just wanted to talk about just some of the design concerns that were raised by astronomers you can talk about a little bit about how you working maybe working with astronomers to alleviate these concerns or are you working on the design or altering it or is are the concerns overblown i mean how do you feel about what has been raised i i am confident that we will not cause any impact whatsoever in astronomical discoveries zero that's my prediction so you're not
18:54watching we'll take corrective action if it's above zero [Music] so you're not giving like orion a hat or anything like that everything's no i mean this there's a sometimes people get a little excited because when the when the satellites are first launched that they're they're tumbling a little bit so they're like they're kind of like they're going to blink and because they haven't stabilized and then and they're they're raising their orbit so they're lower than you'd expect and they're kind of necessarily going to reflect in ways that is not the case when they're on orbit but now now that the satellites are on orbit i'll be impressed if if somebody can actually tell me where
19:37where all of them are i've not met someone who can tell me where all of them are not even one person interesting so i mean it can't be that big of a deal we are actually working with senior members of the the the the science community and senior astronomers to minimize the potential for reflection
20:08of the satellites so you know we're we're running a bunch of experiments to for example just have a paint the phase array antenna black instead of white and we're working on a sun shade because there are like certain angles where if the sun gets you know just sort of just right and there's not like a little sun shade we're not talking about a lot here then you can get a reflection and so we're launching sunshade changing the the color of the satellites and otherwise minimizing the the potential for any impact even like at this aesthetically this should not be an impact i think recently gwen chatwell in bloomberg was quoted as saying that you know you were exploring splitting
21:14starlink from spacex could you talk a little bit about that why that would happen and how you see both of these independent companies functioning and just talk a little bit about that we're thinking about that zero was that we're thinking about that what zero zero zero not thinking about it at all we need to make the thing work it's it's far from obvious that i mean it's real important to just set the stage here for leo communications constellations guess how many leo constellations
21:51didn't go bankrupt zero right zero yet iridium is doing okay now but the iridium one went bankrupt obcom went bankrupt global star bankrupt television bankrupt am i leaving anyone out this bunch of others that didn't get very far also in bankruptcy anyway they weren't bankrupt so you're focusing on making it work first not bankrupt right and not going there's a big that would be a big step let's have like more than zero in the not bankrupt category how would you i mean with how does it work then with the the the business of spacex since you're like i mean you are launching other constellations is that is that a an issue does that cause like a conflict or
22:45we're launching other constellations well you're launching other satellites oh yeah sure whatever yeah no problem of course right so there's no like you know we're even giving them a good deal by the way like no problem anyone watch constellation on spacex sounds good to me so i mean i think that you know i think there's this the world seems to have an insatiable appetite for bandwidth so we're certainly happy to launch other satellites and you know we don't think starlink is not going to destroy all of the satellites or something like that definitely not all right yeah we just want to be in the not bankrupt category that's our goal since you're since you're now
23:31the company of spacex now you know you're building a launching satellite are you are you looking at expanding the business of spacex and other areas of commercial satellite connectivity maybe like we talked a little bit about the you know you're already building like technology on the ground are there other areas that you're looking to get into in terms of commercial space connectivity or satellite services no we're there's just there's two major new technology programs at spacex that's starlink and starship well like it's kind of star in the name of it too much if i maybe just call it like lincolnship if you divide the one by the other the stars net out and then it's just
24:13making sure but that's it as i know anyway and there could be some secret project that's so secretive and i don't know about it you didn't have a business under the chairs yeah i don't think there's anything major okay so i want to i want to give time for people in the audience to ask some questions but you know we talked about starship so development at boca chica is moving along pretty quickly and yeah actually that was the real reason i was late is because that book chica my apologies i was just working on starship with the team there so it's pretty cool out there actually i like it and tell us a little bit about the work that's underway what we can expect in
24:56the future for for starship well we're building a production line production line is the hard part you know making one of something is well at this point you know like frankly designing rockets is not that hard especially if it's an expandable rocket it's not not really a hard problem you can literally read books that'll tell you exactly how to do it the the hard part is now actually building that thing even once is hard and then building a production line is a thousand percent harder like at least a thousand percent harder yeah maybe more so just in general production and manufacturing is underappreciated i think especially in the u.s frankly
25:48so we should really pay a lot more attention and care a lot more about manufacturing this is yeah this is an honest day's work let me tell you also what what inspired some of the design aesthetics of the spacecraft means stainless steel it's a it's a striking design first for a spacecraft but what inspired you know your vision for the way it looks the way it functions like why stainless steel why
26:26well we were going to make it out of advanced composites and the advanced composites that cost like sixty dollars a pound or sixty dollars a kilogram like a little more than that maybe a hundred thirty dollars a pound and there was 60 to 120 plies for the the tank it was taking forever you weren't making good progress cost crazy money and i was like okay switching to aluminum lithium there's also pain in the neck we do that's what we use for the falcon 9 tanks because it's hard to weld because of the reactivity of the lithium so you know what's easy to weld steel seals really easy to weld and stainless steel doesn't even require paint that sounds great because the
27:17paint chops the paint in the neck and you want to try painting something that's got to go to drop to cryogenic temperatures and then bend a lot it's like forget it i mean that paint wants to come off like there's no tomorrow it does not like to stick so then you could use special paint and then the special paint also can't get like when you're going vertically at like supersonic you get the basically static electricity buildup called tribal electrification although it reminds me of the trouble of troubles but you can basically zap yourself if you have paint that with the wrong paint you know so anyway no paint is great yeah because we need a big freaking big-ass paint job
28:07for starship you know one less problem the thing one less problem and paint doesn't weigh zero you know they used to paint the the shuttle external tank white but then like like well we're adding a lot of weight to this thing and it's a big pain in the neck so we'll leave just have it stay orange so just not painting it is great
28:31so then you know and we're not the first to use steel like they used 301 in the early atlas program charlie bossert i think it was yeah i think it was his idea obviously other people involved but charlie bastard by the way that guy's underappreciated he kicks ass it's really great to read about his his stuff he's awesome so he's 301 so obviously it's not a new alloy i think we're going to start switching to a different alloy pretty soon and then just tweak the alloy constituents because we should be able to do better in 2020 than they were that they did in like the 50s you know so i mean come on so i think we'll probably start switching away from 301
29:22maybe the next month or two okay now the funny thing is that like i actually knew that steel especially 301 full hard steel couldn't be that heavy because the original atlas had a very good mass fraction right so it can't be that wrong bad and if you look at the normal normal sort of standard material sheet for 301 it will usually not tell you what that it work hardens dramatically and improves the strength dramatically with work hardening and also at cryogenic temperatures it it improves strength dramatically so then the if you combine the work hardening with the the cryo strength improvement you get an effective strength to weight that is about the same as advanced composite
30:21now people generally make a mistake with composites because they'll look at the material sheet and not realize that okay with composites you've got to have a big knockdown because you basically have composites or string and glue and so you you can't just like have like let's say your problem calls for having four pliers of carbon you can't just have four pliers you need like five or six because in case you damage one or something like that and you say what's your worst case allowable for a d-bond or something like that so the actual knockdown you end up taking for composites is more than you would for a metal structure so people often so it's like a classic newbie mistake is
31:04to overrate carbon fiber because you just you look at the material data sheet and it looks like an obvious move but it's not so anyway so at cryogenic temperatures the steel is has a de facto strength weight about the same as advanced composite but and that does not even counting the fact that you have to paint the composite you don't have to paint the steel right then there's a another factor which is if you want to have a reusable vehicle it's going to get hot composites don't like getting hot so
31:44you know typically your your composite maybe is comfortable up to around 150 200 celsius something like that you know and things start getting pretty sketchy around 250 ec i mean like you start having to use advanced resins and all that kind of thing so whereas steel's pretty happy at a thousand c you know no and no problem at 500 considered 500 c all day and brief periods of 1000 c no problem so then for a reusable vehicle you now need zero heat shielding on the leeward side normally you need some heat shielding just due to rate of heating on the back
32:34so you don't have a lot of convective heating at hypersonic but you have radiative heating and then you can thin out the windward side of the heat shield because the thickness of the of the heat shield tile is driven by the temperature on the backside of the tile where it mounts the primary structure so if you your primary structure can take a high heat that means you can thin out the tile so i think it i think think of it like these like like oven mitts or something you know if you have like how hot can your hand go and that sets how thick your oven mite is right so then you can have like i said no no heat shield on the leeward side on on a
33:19and a thin heat shield on the windward side so now your actual total mass of a steel of a reusable steel spacecraft is less than that of the most advanced carbon fiber vehicle you could possibly imagine yeah wow this happened by accident by the way this may sound like some great insight but it actually happened because we were moving too slowly on composite and i was like we cannot move this slowly or we'll go bankrupt so just do this with steel so yeah i mean the design has to be focused on problem solving otherwise you're going to spend too much time trying to figure you don't start with a yeah yeah i'm like sort of taken to management management
34:02by rhyming if the schedule is schedule is long your design is wrong right this is very true good point yes with that i want to go to some audience questions we asked the audience through the the app to submit us questions so i believe we have a few that we've pulled here so we've got some over here can we and can you come closer over here so we can we can see you i'll say
34:32all right our first question hello i'm jaden zundel i'm a graduate student at stanford my question for you is as you look back on your career in the space industry what has been the most surprising or unexpected challenge that you faced and along those lines if you were to go back in time and talk to your 20 year old self would you do anything differently
34:54go back in time to your 20 year old self i mean i think i'd if i could i think it would make a far fewer mistakes obviously if i could go like here's a list of all the dumb things you're about to do please do not do them wouldn't we all be a very long list and like here let me write it down or something you know i mean it's hindsight's 20 20. so it's hard to say i mean number i've made so many foolish mistakes i had a lot count honestly
35:33i mean some of these things i just wish i like like that's simple sort of mantra management by rhyming i mean it worked for homer okay management by rhyming is the thing i'm saying like if but if the schedule's long the design is wrong we've overcomplicated the design many times
36:01and i think we should have just gone with a simpler design with the acid test being how long will it take to for this to fly and if it's going to take a long time don't do it do something else if you look at say falcon 9 you know it's got a aluminum lithium tank but then the unpressurized structures are carbon fiber composite and really one of the worst possible things you could do to a joint is take something with a high coefficient of thermal expansion high cte put it go take it from room temperature to cryo and then connect it to something that has zero cte you know basically zero like a carbon fiber so now you've got a real pain in the ass joint basically
36:59right so in order for that to work you've got the tanks got to shrink radially and you've got these super expensive heavy bolts that are like a beam and bending across you know that are then taking load into the interstage and they desperately want to share off or snap off this is crazy you know really just have a continuous metal structure that that's obvious that should be done that'd be way better you know things expand to fully available resources so then like sometimes you should say no to things that you that you don't
37:50you know like the original falcon one team which the the fairing tanks engines everything pretty much was maybe a little over 100 people okay and now spacex is like 6 000 people i think something like that so really just simplify your product as much as possible you know and then like if i think of some of the ways which how does a smart engineer make dumb mistakes including you know
38:32is optimize something that shouldn't exist don't optimize something that shouldn't exist if people are trained to do this in college you can't say no to the professor you know the professor is going to give you the exam and you've got to answer all the questions or they will get angry so and give you a bad grade so then you you always optimize the you always answer the question a lot of the times you should say this is the wrong question right in fact the question is definitely wrong to some degree just how wrong and i think just generally taking the approach that your design is some degree wrong probably a lot more than you think your goal is to make it less wrong over time
39:22do we have that let's go to another question hi my name is julie seven sage and mr musk you've said in the past that you think that college degrees shouldn't be that important and that has been shown in job listings in places such as tesla however in places like cis industry including even at spacex in the satellite development area many of the job listings say that you need at least a bachelor's degree and prefer at least a master's degree so my question to you is with more jobs asking for higher levels of
40:07degrees without scholarships are not changing amounts and that is getting harder and harder every year to pay tuition even with using scholarships how can colleges and industries make it easier to afford college but at the same time being able to pay grad students and employees well and also to make sure that there is a large-scale access to good colleges especially to underprivileged communities so that everyone can be a part of the future we're building thank you well first of all you don't need college learning learn stuff okay everything is available basically for free you can learn anything you want for free it is not a question of learning
40:54there there is a value that colleges have which is like you know seeing whether somebody's is can somebody work hard at something including a bunch of sort of annoying homework assignments and still do their homework assignments and and kind of soldier through and and get it done you know that's that's like the main value of college and then also you know if you you probably want to hang around with a bunch of people your own age for a while instead of going right into the workforce so i think colleges are basically for fun and to prove you can do your chores but they're not for learning
41:38there it is i know we started late and i know we we don't have much time left but to build on julie's question here how does somebody like you with a very long-term mission of going to mars how are you cultivating the next generation of leadership to take you there because i mean this is this is a long-term project you might we might not be around to see us finally resettle on mars i mean or may maybe no i mean i i hope i'm not dead by the time people go to mars that would be a great a great outcome i think it might be you know if we don't improve our pace of progress i'm definitely you know going to be dead before we go to mars so i'm just like would like to not be dead
42:21when by the time we go to mars that's my aspiration here so if it's taking us 18 years just to get ready to do the first people to orbit we better improve our rate of innovation or you know based on past trends i am definitely going to be dead before mars so so we're going to improve our pace of innovation a lot so yeah
42:55i guess what i would say is i can tell you can see you how do you communicate that vision you have to that to the somebody who could maybe take over for what you're doing and to see things the way you're seeing them in terms of the the mission well we have a lot of good good people at spacex that
43:13a lot of really talented people in fact i wonder like sometimes how we can make use of their talents in the best way because you know i think we're often not using their talents in the best way yeah but you know to the point of the question i was just asked i want to make sure tesla recruiting does not have anything that says requires university because that's absurd but there is a requirement of evidence of exceptional ability like you just can't if you're trying to do something exceptional they must have evidence of exceptional ability i don't consider going to college evidence of exceptional ability in fact ideally you dropped out and did
44:01something i mean obviously you know you look at like you know gates is a pretty smart guy he dropped out john was pretty smart he dropped out you know larry ellison smart guy he dropped out i'm like obviously not needed so did shakespeare even go to college probably not i mean well i i'm thank you so much i wish we could take more audience questions i know we have we have a hard stop but thank you elon for stopping by thank you let's give him a round of applause for stopping by and speaking to us
44:38enjoy the rest of your evening and thank you very much
资源与出处
- 视频在 YouTube 收看
- 视频在 B 站收看UP 主「swjslhb」搬运
- 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