I’m an optimist by nature. I instinctively look for the bright spot, silver lining, hopeful sign, even during the darkest of times.

This instinctive search has been increasingly hard to carry out in the last two weeks. But I managed to find one last Sunday night: the daughter of a Japanese host family I stayed with for one summer during college messaged me on Instagram and asked me in broken English if I’m doing ok, because she saw on TV that “America is being destroyed and burned down”. She never talked to me when I was living with her family 14 years ago. We somehow managed to be connected on Instagram, but never said a word to each other -- until last Sunday.

This tiny bright spot happened, of course, in the backdrop of a confluent of interconnected major events:

Yet my tiny bright spot could have only happened if Facebook and Instagram grew the way they grew.

Is Facebook the problem? Is Mark Zuckerberg the sole culprit?

Or is there a deeper issue: the algorithm that decides who should see what is beyond any control or accountability and has no legitimacy to that power.

No Due Process In An Algorithmic World

The debate about social networks tends to center on high-minded discussions about freedom of speech and censorship. The current one with regard to Trump’s vile “When the looting starts, the shooting starts” post is yet another example. What’s missing is a constructive discussion about the role that Due Process should play in technology products, especially algorithmically driven ones.

The most thoughtful take I’ve seen on the intersection of Due Process and technology is when two years ago Matthew Prince, the CEO of Cloudflare, explained his decision to terminate service for the right-wing website, the Daily Stormer. As a former lawyer, Prince drew an important distinction between freedom of speech and Due Process:

“The issue of who can and cannot be online has often been associated with Freedom of Speech. We think the more important principle is Due Process. I, personally, believe in strong Freedom of Speech protections, but I also acknowledge that it is a very American idea that is not shared globally. On the other hand, the concept of Due Process is close to universal. At its most basic, Due Process means that you should be able to know the rules a system will follow if you participate in that system.
Due Process requires that decisions be public and not arbitrary. It's why we've always said that our policy is to follow the guidance of the law in the jurisdictions in which we operate. Law enforcement, legislators, and courts have the political legitimacy and predictability to make decisions on what content should be restricted. Companies should not.” [Bold emphasis mine]

The reason Prince pulled the plug on the Daily Stormer was because it falsely claimed that Cloudflare, by providing services to the website as a vendor, was a supporter of its ideology. Prince did not “cancel” the Daily Stormer, because he didn’t like what was on the website. He felt the need to explain his decision, because it might set a dangerous precedence. He did not want to be the arbiter of who gets to be online, much in the same way that Zuckerberg does not want to be the “arbiter of truth”.

I think Prince’s perspective that Due Process is more universal than Freedom of Speech is the correct lens. And to make Due Process work, it requires that decisions be public and not arbitrary.

The algorithms that decide which tweet appears in my feed or which YouTube video I should watch next is anything but “public and not arbitrary.” In fact, they are the carefully-guarded secret sauce that’s supposed to fuel user growth and engagement, which in turn drives more advertisers and higher ad rates.

An algorithm is a decision making process. If an algorithm has no Due Process, its decisions have no legitimacy.

You can apply the same framework to the CEOs of any of these social media companies.

People are mad at Zuckerberg because they don’t agree with his decision to not remove Trump’s post, but there’s no Due Process that confers Zuckerberg any legitimacy to decide.

People are applauding Jack Dorsey’s decision to flag the same post on Twitter, because it aligned with their opinion about Trump, but there’s no Due Process that confers Dorsey that legitimacy either.

Had the decision been reversed: Zuckerberg removed Trump's post while Dorsey left it unflagged, the decisions would’ve been equally lacking in Due Process power. Had the decision been about another politician saying some other incendiary things, it would be no different.

While we aim our anger, frustration, and blame at these CEOs, the algorithms that fuel their social networks -- imbued with the features engineered by human beings with their own intentions, emotions, and biases -- have become far more powerful and well beyond their control. But these algorithms are faceless, opaque, and harder to reason about, thus harder for us to blame.

A Social Network with Due Process

There has been a chorus of investors publicly shaming Zuckerberg and looking to fund a Facebook alternative. I don’t think another VC-backed social network startup will solve the lack of algorithmic Due Process problem.

If a social network startup (or any startup) takes VC investment, it’s expected to eventually generate hypergrowth and outsized financial returns. That is the nature of the economics of venture capital, and the sole professional purpose of these investors. For a new social network to grow quickly to meet this expectation, a highly efficient algorithm engineered solely for the purpose of user engagement morphed into ad dollars is still the best playbook.

Look no further than the newest and hottest entrant into the global social media space: Bytedance, in the form of TikTok. To be far, Bytedance has been hot in China since 2013 when its first news aggregator app, Jinri Toutiao (meaning: Today’s Headlines), took off but western tech media largely ignored it. Its second big hit, Douyin, was launched in 2016 and is the direct precursor to TikTok after the company bought Musical.ly. Whether it's news headlines or short videos of Gen Z’s lip-syncing, Bytedance’s apps are all powered by opaque AI algorithms. This playbook generated $3 billion USD in profit last year on $17 billion of revenue, mostly from ads.

To be a real alternative to Facebook at the core, the social network’s algorithm has to be public and not arbitrary, transparent and not faceless. It has to have Due Process.

Ironically or perhaps appropriately, the only effort of any scale that resembles this alternative is Libra. (Social networks like Minds are still very small.) The Libra project has met with waves of challenges and detraction since its launch. But judging from the measuring stick of Due Process, its codebase was open-sourced from day one, it’s built on the blockchain which is by definition an open protocol, and it’s governed by an association of organizations, reducing the possibility of arbitrary decision making by Facebook the company or Zuckerberg the person.

To be clear, Libra is still more of a fledgling concept than a production-ready application. It’s also a payment system, not a social network. It’s too early to say whether Libra has algorithmic Due Process and the legitimacy that that confers. But if Zuckerberg’s privacy-focused vision for a social network were to succeed -- a place where a lost acquaintance of 14 years can find me without being lab rats of an algorithmic ad machine -- Libra along with Whatsapp will undoubtedly be critical pieces to that future.

The engineering and technical challenges to implementing that vision will be immense and take years. Thankfully Zuckerberg has more money and patience than all the VCs combined. Unfortunately, our world may not have enough time for him or another entrepreneur to figure all this out before we destroy ourselves. But I’m still an optimist.

I know this analysis may upset a lot of people. It’s even more upsetting for me to write it, as someone who’s spent most of his 20s working for Barack Obama and cannot stand Donald Trump sitting in the White House. But for any analysis to be useful, it must be dispassionate. My feeling is irrelevant to this post, in the same way that Matthew Prince’s feeling is irrelevant to who should be online, and that Mark Zuckerberg’s feeling is irrelevant to what you can see on Facebook.

If you like what you've read, please SUBSCRIBE to the Interconnected email list. New posts will be delivered to your inbox (twice per week). Follow and interact with me on: Twitter, LinkedIn.

算法才是问题，扎克伯格不是

我天生是个乐观主义者。即使在最黑暗的时候，我也会本能地去寻找亮点和希望。

过去两周，这种本能的搜索越来越难了。但上周日晚上，我还是找到了一个：我在大学期间的一个暑假曾经在日本学习，当时在一个日本家庭寄宿。这个家里的大女儿在Instagram上给我发了一条信息，用蹩脚的英语问我是否还好，因为她在电视上看到“美国正在被毁灭和烧毁”。14年前，我在她家住的时候，她从没和我说过话。我们虽然在Instagram上是“好友”，但也从来没发过任何信息，直到上周日。

当然，这一小亮点是在一系列相互关联的大事的背景下发生的：

  • COVID-19
  • 全美国反对系统性种族歧视的抗议活动
  • 美国民权领袖、Facebook员工以及许多民众都批评扎克伯格没有像Jack Dorsey那样“核实”特朗普的帖子
  • 特朗普的新行政命令挑战保护社交媒体的法律条款（Section 230 of the Communications Decency Act)

然而，我这个小亮点能出现也是因为Facebook和Instagram的扩大和发展。

Facebook是问题所在吗？扎克伯格是唯一的罪魁祸首吗？

或者还有一个更深层次的问题存在：决定谁应该看到什么内容的算法，已经超出了任何控制或责任，而它这种巨大权力的背后没有任何公认的合理性。

毫无正当程序的“算法世界”

关于社交网络的讨论往往集中在对言论自由和话语权的封杀等高瞻远瞩的角度上。目前关于特朗普“抢劫开始时，枪击开始”的恶言秽语的帖子就是个例子。缺乏的是关于“正当程序”（Due Process）在技术产品中应扮演的角色的建设性讨论，尤其是算法驱动的产品。

在“正当程序”和科技技术的交叉点上，我看到的最深思熟虑的一篇思考是两年前，Cloudflare 的首席执行官 Matthew Prince 解释了他终止对右翼网站《每日风暴报》（Daily Stormer）的服务的决定。作为前律师，Prince对言论自由和正当程序作了重要区分：

“谁能上网、谁不能上网这个问题，往往与言论自由有关。我们认为更重要的原则是正当程序。 就个人而言，我相信强有力的言论自由保护，但我也承认，这是一个非常美国的想法，并没有全球共识。另一方面，正当程序这个概念更接近普世。最基本的正当程序就是，如果你参与了一个系统，你应该能够知道该系统将遵循的规则。
在正当程序下，所有的决定都是公开的，不是武断而毫无理由的。这就是为什么我们一直说，我司的政策是在所有我们有业务的地方，遵循当地法律的管辖。执法部门、立法人士和法院有政治合法性和可预见性，可以决定哪些内容应该受到限制。公司不应该有这种决策的权利。”

Prince撤销《每日风暴报》的原因是，该报谎称Cloudflare作为给网站提供服务的供应商，同时也支持网站内容的意识形态。Prince并不是因为他不喜欢网站上的内容，而“取消”《每日风暴报》。他觉得有必要解释他的决定，因为这可能会造成危险的后果。他不想成为“谁可以上网的仲裁者”，就像扎克伯格不想成为“真相的仲裁者”一样。

我觉得Prince认为正当程序比言论自由更普世这个观点是正确的，也是我们应该用的视角来看这些问题。一套正当程序能发挥作用的前提是，决策要公开，而不武断。

决定哪些tweet出现在我的feed中，或者我下一个YouTube视频应该看那个的算法根本不公开，而是绝对武断的。这种算法是这些公司精心保护的“秘方”，来促进用户增长和参与度，从而召集更多的广告商和卖更高的广告费率。

一个算法就是个做决策的过程。如果一个算法没有正当程序，它的决策就没什么合理性。

你可以把同样的框架套在这些社交媒体公司的CEO身上。

人们对扎克伯格很生气，因为他们不同意他不删除特朗普的帖子或给予警告，但没有任何正当程序赋予扎克伯格做这个决定的权力。

人们对Jack Dorsey决定在特朗普发的同一个帖子在Twitter上给予警告而表示赞赏，因为这与他们对特朗普的看法一致，但也没有任何正当程序赋予Dorsey这种权力。

如果这个事情倒过来看：扎克伯格对特朗普的帖子给予警告，而Dorsey却没有，这两个决定也同样缺乏只有正当程序能给予的权力。如果此事发生在另一个政客人物说的另一套有煽动性的言论的话，也应该是一样的。

当我们把愤怒、懊恼和责怪的矛头指向这些CEO时，那些为他们的社交网络提供燃料的算法——充斥着写它的人们自己的意图、情感和偏见而设计的功能——已经变得过于强大，远远超出了他们的控制。而这些算法是不公开的，不透明的，很难解释，因此我们也更难责怪它。

有正当程序的社交网络

有一大群投资人已经开始公开羞辱扎克伯格，并渴望投资一个能替代Facebook的产品。我不认为另一家由风投支持的社交网络公司能解决这个“缺乏算法正当程序”的问题。

如果一家社交网络创业公司（或任何一家创业公司）接受风险投资，它最终必须要有快速的增长从而给投资人们丰厚的回报。这是风险投资本身的经济原理的本质，也是这些投资人唯一的职业目标。为了让一个新的社交网络快速发展来满足这一期望，设计一个专为增长用户参与度、再把参与度转变成广告费而设计的高效算法，仍然是最好的攻略。

只要看看全球社交媒体最热门的新手，字节跳动的TikTok，就可以看到这一点。当然，自从2013年发布的今日头条开始，字节跳动在中国就一直很火爆，但西方科技媒体基本上忽略了它。它的第二个主打，抖音，于2016年推出，也是收购Musical.ly之后TikTok的直接前身。无论是新闻头条还是九五后的假唱短视频，字节跳动的所有产品都是由不透明的人工智能算法驱动的。这套攻略去年以170亿美元的收入创造了30亿美元的利润，其中大部分来自广告。

要真想从核心替代Facebook，这个社交网络的算法必须是公开的，而不是武断的，必须是透明的，而不是毫无面目的。换句话说，它必须遵守正当程序。

既讽刺也有些恰当的是，目前唯一有些规模的Facebook替代品就是Libra。（像Minds这种其他社交网络的规模还很小。）Libra项目自推出以来，遭遇了一波又一波的挑战和诋毁。但从正当程序的衡量标准来看，它的代码从第一天开始就是开源的，它建立在区块链上，而区块链的本质是一个开放的协议，而且它是由一个会员组织协会管理，减少了Facebook公司或扎克伯格个人单方做决定的可能性。

当然，Libra总体还是一个初出茅庐的概念，并不是一个可以用的程序。它本身是一个支付系统，不是一个社交网络，现在判断Libra是否具备有“算法正当程序”的合理性还为时过早。但如果扎克伯格的以隐私为中心的社交网络的伟大设想要能获得成功——既可以让一个14年没有来往的人随时可以找到我，又不会她继续做一个算法驱动的广告机器里的实验小白鼠——Libra和Whatsapp无疑会是这种未来的关键。

实现这种未来而需要的工程和技术挑战将是巨大的，需要数年时间。谢天谢地，扎克伯格比所有风投加起来都更有钱，也更有耐心。也许不凑巧的是，我们的世界可能没有足够的时间等他或另一个创业者在我们毁灭自己之前把这个事情搞定。但我仍然是个乐观主义者。

我知道看完这篇分析后很多人可能会不开心。作为一个多年为奥巴马工作而无法忍受特朗普坐在白宫里的我来说，在写这篇分析的过程中我也很不开心。但任何有用的分析都必须带有一个冷静而无情感色彩的视角。我的个人情感不应该影响这篇文章的内容，就像Matthew Prince的个人情感不应该影响谁可以在网上，就像扎克伯格的个人情感也不应该影响你在Facebook能看到什么一样。

如果您喜欢所读的内容，请用email订阅加入“互联”。每周两次，新的文章将会直接送达您的邮箱。请在Twitter、LinkedIn上给个follow，与我交流互动！