11/13/2022: Fail Whale
本周的主题是 Fail Whale 鲸落。
恐怕大部分读者都不知道或者不记得 Fail Whale 的意思了。它是 Twitter 早期服务崩溃的时候会出现的一副图片,由于当时 Twitter 的性能和稳定性真的很差,无法和当时它火爆的增长曲线相匹配,所以这幅图片时常出现,以至于成为小圈子流行文化的一部分。
而这一切早已经随风而逝,不再有人为此津津乐道。
Elon Musk 召开了他作为 Twitter 所有者后第一次员工大会。本周的 newsletter 主要摘录和评论了这些会议记录中的部分内容。尽管抨击很多,但我对他执掌 Twitter 满怀期望——这完全是从观察和学习如何重新启动一个饱受争议的社交媒体平台角度来说的。无需赘言,我当然对此充满关切。
Venkatesh Rao 也写了一篇文章叫 The Muskening,开头就做了这么一段比喻:
In brief, an irresistible force has met an immovable object. An entrepreneur who has never met a venture he couldn’t master has met a platform that has never met a leader who could tame it. A legendary emperor has marched into a fabled graveyard of empires.
简而言之,一个不可抗拒的力量遇到了一个不可改变的对象。一个从未遇到过他无法驾驭的风险的企业家,遇到了一个从未遇到过可以驯服它的领导人的平台。一个传奇的皇帝进入了一个传说中的帝国墓地。
这是一个精彩故事的开头,让我们开始共同欣赏。
Essays
Elon Musk's First All Hands Meeting with Twitter Employees by The Verge
The Muskening by Venkatesh Rao
The Verge 发表了 Elon Musk 在收购 Twitter 之后首次员工大会的文字版。在收购之后,Musk 采取了一贯的强硬作风,大幅裁员,推动每月 8 美元的 Twitter Blue 订阅,引发了大量争议性舆论。
员工大会一般很少会被媒体公开,特别是对于一家已经被私有化了的公司。会议持续了 1 个小时,以 Musk 简短的开场白和员工 Q&A 的形式展开。我希望从这个文本中得到直接来自 Musk 本人关于 Twitter 未来走向的回答,而尽量少的被媒体报道所影响。
以下是我的一些摘录和批注。并不是完全按照会议中的发言顺序排列。
Twitter Blue
I think we need to spool up subscriptions. The reason I have extremely high urgency around subscriptions is that we are headed into, I think, quite a serious recession. And you can see that with basically every company doing layoffs. It’s not just Twitter. And in a recession, advertising is disproportionately affected. And then advertising where the advertiser cannot clearly trace how much they spent on an ad to what demand it generated, which is known as brand advertising, will be even more affected. This puts Twitter as we look ahead into a very dire situation from a revenue standpoint.
首先谈到了 Twitter Blue。Musk 谈到自己在 PayPal 和 Tesla 都经历过经济衰退周期,并坦言自己有衰退创伤综合症(recession PTSD)。上述发言的重点在于:在衰退中,广告会收到不成比例的影响。因此,订阅收入更为重要。
Verified accounts
It’s definitely just one step at a time. I intentionally limited things to only iOS, and just a small number of countries, to not have a massive amount of fraud that our systems… that we couldn’t handle. So the hardest thing to commit fraud on right now, relatively speaking, is if you require a phone and a credit or a debit card.
对于最近饱受争议的帐号认证,Musk 揭示了自己为什么会采用支付 8 美元的形式。这是因为他认为对抗虚假帐号最好的方式就是通过一个真实的电话号码,或者是一张借记卡。 但是显然,这个操作被认为是可以用花钱的方式来获得认证,的确在这个功能在上线以后也出现了一些新的虚假帐号。Twitter 也紧急叫停了这个功能。
某种程度上,Musk 并没有充分考虑到现在社交媒体上虚假帐号的行为动机与特征。他的这种思路更多是从他在 PayPal 年代积累的经验而来。发言比支付成本更低,但可能造成更严重的后果。
在这一点上,Musk 还需要更快的学习。
Payments
The full answer to the payment story is complicated. PayPal, as it’s known today, is a subset of a feature roadmap that we wrote in July of 2000. Roughly speaking, you want to basically populate a balance for every verified Twitter user. Give them some amount of money, like 10 bucks or something, that they can send anywhere in the system. We need money transfer licenses for that, which we’ve applied for.
在支付业务上,Musk 显然思考的更多,他提到 PayPal 实现的产品当时他们所写的路线图的一部分。显然,他有更大的一个蓝图要在 Twitter 实现。具体来说,他会利用 Twitter 上积累的社交图谱,开始给帐号建立余额概念,然后开始让用户在 Twitter 的系统内进行余额转移。同时,申请相应的金融牌照,并且开始把借记卡、贷记卡这样的功能绑定到 Twitter 帐号上,最终形成一个完整的金融服务。
他很强调私信(Direct Messaging)在支付产品中的重要性。或许这是从微信那里学习到的。而最终产品呈现的形态或许是 Jack Dorsey 本来想要在 Square + Twitter 这个组合中实现的。
E-commerce
Well, that goes hand in hand with payments. If you have a balance on Twitter or you have basically money in your Twitter account, then it’s actually very easy to buy things. So that’s a natural outgrowth of having an effective payment system, is making it easy to buy stuff.
也有员工问到电子商务的机会,Musk 的回答是电商与支付紧密相关,如果支付做得好,那么就会极大的提升用户在 Twitter 进行购物的转化率。
Video
And we also want to expand to be sort of a multimedia platform. We are that to some degree. We are the strongest when it comes to anything that’s writing and real time. But we also want to have that for pictures and video and not in a way that copies what others do. I don’t think we should not copy something just because someone else did it. We just want to focus on how we make Twitter as compelling as possible, where it’s maximum enjoyment, maximum learning. Where you spend an hour on Twitter, and you don’t regret it. Sometimes, you know, on some of the social media platforms, you might spend an hour, but you, like, kind of regret that, right? Not naming names but, you know, it’s like you want to spend an hour and be like, “Wow, that was a great hour.”
视频是 Musk 多次提到的另一个重点,但他似乎并没有精确的区分 YouTube 和 TikTok 到底谁更像是 Twitter 的竞争对手。一方面,他讲到很多 YouTube 创作者会在 Twitter 上发表内容,但视频却是放到 YouTube 上的,这是一个迁移的机会,通过更高的创作者补贴完成。另一方面,他认为 TikTok 的产品比 Twitter 简单,用户不需要进行繁琐的操作,就可以看到有趣的内容。
而他也提到了「最大化的娱乐,最大化的学习」的说法,一定程度上认为 Twitter 在内容上应该有一定的差异性,而不是让人们在看了一小时后感觉后悔。
视频的机会是兵家必争之地。前面已经有 Facebook 和 Instagram 灰头土脸的尝试,Twitter 目前也开发了上下滑的视频播放界面,但推荐算法还差得有点远。
Rao 这样批评 Musk 的做法:
Another tell is regression to familiar notions of businss value. His knee-jerk first response business model ideas for an “X app” seem to be PayPal era thinking (which he references explicitly) mashed up with vague notions of replicating the WeChat model, and a shallow copycat media design direction apparently inspired by TikTok and YouTube. What’s common to all the ideas is that precedents exist for them, at scale. They’re not new, untested ideas, and they’re also not ideas that Twitter is particularly well-positioned to pursue. If you wanted an everything payments-and-shopping app, buying Twitter would not be a self-evidently good first step. It is a lazy idea retconned onto a deal made for other reasons.
他对 "X 应用 "的第一反应是 PayPal 时代的思维(他明确提到了这一点),并掺杂了复制微信模式的模糊概念,以及明显受到 TikTok 和 YouTube 启发的浅层山寨媒体设计方向。所有这些想法的共同点是,它们都有先例可循,而且规模很大。它们不是未经测试的新想法,也不是 Twitter 特别适合追求的想法。如果你想做一个支付和购物的应用程序,收购 Twitter 并不是一个不言而喻的好的第一步。这是一个懒惰的想法,是为了其他原因而进行的交易。
All this is not only unimaginative by Musk’s own high standards, it is unimaginative by any standards. And a failure of imagination is the biggest overall sign of a collapsed OODA loop, because it is a symptom of loss of control over tempo. When you don’t control the tempo, you don’t have the freedom to be creative.
这一切不仅以马斯克自己的高标准来看是没有想象力的,而且以任何标准来看都是没有想象力的。想象力的失败是 OODA 循环崩溃的最大整体标志,因为它是失去对节奏控制的症状。当你不控制节奏时,你就没有发挥创造力的自由。
其中 OODA 是一个军事术语,是指 Observe-Orient-Decide-Act,Rao 用来描述 Musk 在收购过程中的一系列行为。
Priorities
I think just sort of proposing those things to me, I think, would be the right thing to do. But, you know, like I said, I think there’s a lot that is very, very obvious that we need to do. Like video content and compensating content creators in order to get content on Twitter and not have them be forced to put content elsewhere or not pay the bills, which is currently the situation. That’s a no-brainer. High priority. Improving search: high priority. We are obviously going to add payments capability to Twitter. That’s also a high priority. And if there is something you think we should do that I’m not aware of, just tell me about it, and I’m open to doing it.
Musk 也提到了他设定的一些重要优先级,其中包括视频内容、对内容创作数的补贴、搜索和支付。因为这是在问答环节当中进行的总结,所以并不是很全面,但至少可以让我们看到他最关心的一些问题是什么。
Metrics
I think we don’t want to get overly focused on metrics. It’s helpful to see these metrics as sanity checks. But if you’re over-optimized on a metric, then you end up being just like a foolish AI. You pursue a number without common sense. What actually matters is: are we making the product amazing? If you make the product amazing, people will use it.
Musk 认为,Twitter 不应该过度聚焦在指标上,避免针对指标进行过度优化。他相信产品做得好,人们就会使用。
这是他从 Tesla 的经历中学习到的成功经验。
Engineering-driven
I guess engineering-driven culture is really just are we advancing the core technology of the product. This is not engineering to the exclusion of everything else. Take Tesla, for example. We have an incredible design studio that I built from scratch with Franz von Holzhausen. We do have a project management organization, and then we’ve got a very large engineering organization. But product management, design, and engineering all report to me. It’s not just like everyone reports to engineering. Its design, project management, and engineering report to me, and they are of equal importance.
谈到组织文化,Musk 基本上借鉴了自己的 Tesla 的方法:工程师驱动。但他也解释说,这不意味着所有人都要向工程师汇报。设计、产品管理和工程三个团队会向他直接汇报,三者同等重要。
值得注意的是,Musk 在收购后立即解雇了一半的员工。Rao 对这个举动的看法是:
即使 Twitter 在两周前才人手过剩 2 倍,当你处理一个复杂的全球规模的技术平台时,弄清楚裁掉哪一半显然不是一夜之间就能做到的。这不是一家拥有庞大而明显的可替代劳动力的公司,可以很容易地以两位数的百分比被解雇。
Musk 之前与前任 Twitter CEO Parag Agrawal 讲道:
Frankly, I hate doing management stuff. I kinda don't think anyone should be the boss of anyone. But I love helping solve technical/product design problems.
Twitter 和很多过去十年的互联网产品一样,由于 1.0 过于成功,反而在后续的发展中缺乏真正的进化。我倾向于认为,这种产品、设计和工程驱动的文化能够让 Twitter 重新回到正确的演进轨道上。显然,Musk 对社交和内容的理解是不够的,Twitter 一定是一个比 Tesla 更复杂的系统。他需要更快的聚焦在少数优先级上,并把产品实现出来。
Rao 对 Twitter 的四大支柱分析:
Among what I think of as the four pillars making Twitter what it is, Discovery, Discourse, Dopamine, and Distribution, the last is the most legible, and most immediately amenable to authoritarian high-modernist redesign for yield. Pay-for-reach is a proposition everybody understands. Discovery is clearly on the chopping block if TikTok style algorithmic discovery, rather than discovery through social serendipity, is on the roadmap. Discourse is highly valuable to users, but not obviously monetizable, and the most compromised by his somewhat basic political opinions. And dopamine is primarily a cost and can be delivered in much cheaper ways than the current mechanism of actually enjoyable mutualism.
在我所认为的 Twitter 的四大支柱中,发现(Discovery)、讨论(Discourse)、多巴胺(Dopamine)和传播(Distribution),最后一个是最容易辨认的,也是最容易被专制的高度现代主义重新设计的。以钱换覆盖是一个人人都能理解的主张。
如果 TikTok 式的算法发现是在路线图上的话——而不是通过偶然性的社交发现,那么发现显然是砧板上的肉。 讨论对用户来说是非常有价值的,但不能很轻易的变现,而且他的一些基本政治观点对其影响最大。而多巴胺主要是一种成本,可以用比目前的互助机制更便宜的方式提供。
这些论点加起来,的确听上去更像是 TikTok 的样子。
Shortform
@balajis:
Four kinds of trust falling at the same time. Trust in vote counting Trust in verified tweets Trust in vouched reserves Trust in visual images Cryptographic verification is a partial solution. But also need high-trust startup societies.
前 Coinbase CTO,也是 The Network State 一书作者 Balaji Srinivasan 认为,信任至少在 4 个领域上都失效了:选举计票、Twitter 验证、担保准备金和图片。
Longform
本周在继续阅读 Inventor of the Future,Richard Buckminster Fuller 的传记。
在阅读这本书的过程中,我回到了 Apple Books 这个阅读软件。在 iOS 16 的升级过程中,它丢失了一个重要的导出笔记的功能。
或许这个功能后续还会被加回来,或许不会——出于版权保护的考虑或是其它。但它对我整理读书笔记曾经至关重要。我适应这个变化的方法是手动复制粘贴。
手动的好处在于可以在复制粘贴之余增加批注。这也让我在选择高亮内容的时候变得更加慎重一些。在阅读本书的过程中,我更重视梳理 Fuller 一生中重要的时刻,用小标题的方式列出一个大致的年表。在笔记大纲视图下,形成在以后容易查找的索引。
Fuller 在真正确立他的地位之前,多少显得有些投机取巧。作者在字里行间总是想要暗示这一点,告诉读者,某些关于主人公的传说在经过一些参考比对之后是存疑的,可能是来自于主人公自己的故弄玄虚。这些描述,和 Fuller 早年有些离经叛道的论调放在一起,更降低了对其历史地位的期望。在交通和住房上的创新往往并没有他宣称的那么实用。
转机出现在他结识 Henry Luce 之后,他开始在后者庞大的媒体帝国中建立新的人脉,在 Fortune 和 Life 杂志上发表文章,并由此开始得以接触到名流政要。
媒体,至少在那个年代,是能够赋予一个人足够大的影响力的。