By now, the entire world knows that Elon Musk wants to buy Twitter. During his first interview after making his intention known, Elon made it clear that he wants to open source Twitter’s algorithm. This announcement was met with long, thunderous applause.
There’s no doubt that if Twitter’s algorithm was open-sourced, it would bring some much-needed transparency to the opaque social media platform, which would be a leap of progress for humanity. The idea of open-sourcing algorithms, though, is not a new idea. Many people have been advocating it for years. In this very newsletter, I’ve advocated for it in the context of both TikTok and Facebook.
However, this idea has not been implemented widely, because it is often framed as a magnanimous decision that may be good for humanity, but is bad for business. I think this is a false tradeoff built on faulty logic.
Open-sourcing Twitter is actually good for humanity and good for business – a rare case where you can have your cake and eat it too!
Algorithm is Not the Moneymaker
The algorithms that power a social media platform are often regarded as its “secret sauce”. This secrecy is misplaced, because the algorithms themselves are not the moneymaker.
How so?
Here’s a non-technical, plain-language explanation of why these algorithms’ value is misunderstood and often inflated. The algorithms that power an ad-driven social media platform are akin to maps. Their goal is to:
- Route the right ads to the right users, so these users will do something that the advertisers who paid for the ads care about
- If the algorithmic route is successful, the advertisers are happy, Twitter generates revenue, and Twitter can try to charge a higher ad rate from these advertisers to make more money
- If the algorithmic route fails, it quickly evolves (sometimes with human intervention, sometimes automatically) by changing the routes, so the next time an ad comes through, the algorithm can do a better job of routing, otherwise Twitter will lose this advertiser as a customer
The algorithms are the “rules of the road” that lay out all the routes. However, these routes are inherently not that valuable and not the “secret sauce” to building an ad-driven social media business like Twitter. Most AI frameworks are already open-sourced (e.g. Tensorflow, PyTorch, Keras, etc.). Most of the mathematical equations that underpin these frameworks are published in academic journals, open to anyone who can understand them to implement, verify, and backtest. Twitter’s algorithms are likely built from these free, open components.
What is valuable is the user-generated content (i.e. tweets) and the data that gets attached to that content (i.e. who viewed, liked, retweeted, replied, etc.).
“Liked” a Vitalik tweet about Ethereum? Route an ad about a web3 conference to this user. User registered for the conference. Success! Retweeted a tweet about Tesla by Cathie Wood? Route an ad about the new Prius paid by Toyota to that user. User ignored the ad (probably wants a pure EV, not a hybrid). Fail! Update route to send a different ad about a pure EV paid by Nio. User clicked on it. Success! You get the picture.
This data produces the signals that these algorithms need to do the routing and improve their routes over time, so they can get better at routing future ads.
An algorithm without data is like a highway system without cars – cool-looking but useless.
Thus, open sourcing the algorithms won’t hurt Twitter’s business, because it is really not that business critical. What is business critical is growing its user base, frequency of user-generated tweets, and interactions with those tweets, so the algorithms have more and better signals to improve their routes for the next ad!
Due Process for Tweeters and Humanity
There was a telling moment in Elon’s interview where the interviewer, Chris Anderson, compared two tweets he wrote on the same day to promote the interview – one was generic, one tagged Elon’s handle. The one with Elon’s handle had 100-times more likes and engagement than the generic tweet. Both Elon and Chris were dumbfounded by the difference.
Most of the discussions around Elon’s bid to buy Twitter center on our freedom of speech – as in people’s right to speak their minds in this proverbial “town square”. What’s less discussed, but equally important, is whether we can ensure that different pieces of speech can be equally heard, and why one piece of speech (Chris’s second tweet) is louder than another (Chris’s first tweet). It’s not a “right to a freedom” question; it’s a “due process” question.
The most thoughtful tech CEO on the topic of due process and technology is Matthew Prince, the CEO of Cloudflare. When Cloudflare terminated service for the right-wing website in 2017, an early version of “de-platforming”, here is how Prince explained the decision, drawing 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]
Open-sourcing Twitter’s algorithms meets the due process requirements, because the “rules of the road” are now public. And if these rules are indeed arbitrary, where certain powerful people within Twitter are putting their thumbs on the scale in favor of some people’s speech over others, we now have proof! This proof can then be used to provide evidence to improve AI regulations (which Elon is in favor of) and improve the UX for everyone exercising their freedom of speech on Twitter (which is good for all tweeters).
All this can happen without hurting Twitter business. It can even strengthen the business, as the platform becomes more transparent and trustworthy, thus attracting more users, more tweets, and more advertisers. Open source itself is not a business model. But open source projects have been the foundation of many innovative business models – open core, cloud SaaS, enterprise services – that have powered many multi-billion dollar businesses, like MongoDB, Elastic, GitLab, Hashicorp, to name a few. There is no reason why Twitter cannot become the next big “open source” company.
As a company, Twitter has been contributing to open source for many years, has a dedicated open-source team, and maintains an active presence on GitHub. It already has the domain expertise to open source its algorithms and to do it well. The only thing that’s lacking is leadership and a clear understanding of its algorithms’ lack of business value.
An Age of Trust, Not Just Convenience
In 2013, Ev Williams, one of the founders and ex-CEO’s of Twitter, shared a profound insight that the Internet was just about enabling convenience. Nothing more. That insight fueled the rise of his first startup, Blogger, then Twitter. That insight of maximizing convenience for people to do what they’ve always done for millennia – express themselves, find information, create memes, seek entertainment, troll other people – is at the heart of every mainstream social media platform.
While the need for convenience has not changed, what I believe has changed is a stronger demand for trust. People want to be on a platform that has a stronger due process that they can trust more, so they can continue to express themselves, find information, create memes, seek entertainment, and troll other people.
Open-sourcing its algorithms can help Twitter build that trust overnight. Only a tiny percentage of Twitter users will regularly go into the codebase to scrutinize the latest changes, but the fact that they can, whenever they want, to is all that matters.
As fiduciaries of Twitter’s shareholders, the Twitter board should like the idea of open-sourcing its algorithms, because it will be good for Twitter’s brand, public trust, and business growth. Having Elon’s ownership catalyze this industry-shifting change should be a welcoming prospect. Instead, it adopted a “poison pill” to prevent Elon from acquiring the company.
Elon has hinted in his interview that if his bid to buy Twitter fails, he has a Plan B. That plan may, in fact, be building an open-source alternative to Twitter. And he will take all the Twitter users, who trust him more than Twitter, away.
为什么开源Twitter的算法，对人类好，也对生意好
全世界都知道Elon Musk想买Twitter了。在他公开收购意图后的第一次采访中，Elon明确表示，他想将Twitter的算法开源。这一公告得到了雷鸣般的掌声。
毫无疑问，如果Twitter的算法被开源，它将为这个不透明的社交媒体平台带来一些急需的透明度，这会是人类进步的一个飞跃。不过，“开源算法” 这个想法本身并不新，许多人多年来一直在倡导它。我在《互联》上也曾以TikTok和Facebook为例，倡导过同样的想法。
然而，开源算法目前并没有被广泛落实，因为它经常被形容为一个“宽宏大量”的决定：也许为造福人类有益，但对商业利益不利。我认为，这是一个建立在错误逻辑上的错误权衡。
开源Twitter实际上对人类和商业都有好处——这是个挺罕见的情况，鱼和熊掌可以兼得。
算法不是赚钱的法宝
社交媒体背后的算法通常被认为是其“独家秘方”，但“秘制”的并不是它，因为算法本身并不是赚钱的核心。
我打一个非技术的、通俗的比方，来说明为什么这些算法的价值常常被误解并夸大。一个由卖广告驱动的社交媒体平台的算法，其实类似于地图。目标是：
- 将广告引向到正确的用户面前，导致用户会做一些为卖广告的广告商在乎的行为
- 如果算法路线成功，广告商高兴，Twitter就会有收入，还可以尝试向广告商收取更高的费用，赚更多的钱
- 如果算法路线失败，就会迅速进化（有时是人工的，有时是自动的），改变路线；这样下一个广告来的时候，算法就可以做得更好，否则Twitter会失去这个广告商作为客户。
算法是“道路规则”，规定了广告怎么从到用户面前的所有路线。然而，这些路线本质上并没有什么商业价值，也不是建立像Twitter这样由广告驱动的社交媒体业务的“秘方”。大多数AI框架和算法已经开源了（例如Tensorflow、PyTorch、Keras等），支撑这些框架的大多数数学方程也都已发表在学术paper上发表了，任何一位能够看懂它们的人都可以随意免费的去验证、回测或实施。Twitter的算法很可能用了很多这些免费、开放的组件来构建的。
真正有价值的，其实是用户生成的内容（即推文）和附加在这些内容上的各种数据（即谁查看、点赞、转发、回复了等等）。
给Vitalik关于以太坊的推文“点赞”了？那就推个web3大会的广告给该用户。用户注册了该会议——成功！“转发”了Cathie Wood关于特斯拉的推文？那就推个丰田公司付费的新普锐斯的广告给该用户。用户忽略了这个广告（他可能想要纯电动车，而不是混合动力车）——失败！更新路线，改推由Nio付费的纯电动汽车的广告。用户点击了它——成功！您应该明白我的意思了。
这些数据制造了算法做路由所需要的信号，并随着时间的推移不断改进算法的路线，以在未来更好地路由广告。
没有数据的算法就像没有汽车的高速公路系统——看起来很酷，但毫无用处。
因此，开源其算法不会损害Twitter的业务，因为它对生意没那么重要。对生意更关键的是增长其用户群、用户发推的频率以及与用户与推文间的互动，这样算法就能有更多、更好的数据作为信号来改进下一个广告的路线！
为推客（Tweeters）与人类的 “正当程序”
在对Elon的采访中，有一刻很耐人寻味：采访者Chris Anderson比较了他在同一天，为宣传这次采访而写的两条推文——一条很普通，另一条则标记了Elon的Twitter用户名。带有Elon用户名的那条推文的点赞数和参与度竟达到了普通推文的100倍之多，Elon和Chris都对这种差异目瞪口呆。
围绕Elon收购推特的大部分讨论都集中在言论自由上——就是人们在Twitter这个数字 “广场”，怎么保护言论自由的权利。讨论相对较少但同样重要的是，我们是否能确保不同的声音被平等地听到，以及为什么有些言论（Chris的第二条推特）比另一些（Chris的第一条推特）更响。这其实不是“自由权利”的问题，而是“正当程序”的问题。
在正当程序和科技交叉的问题上，最有想法的科技界CEO是Cloudflare的Matthew Prince。当Cloudflare在2017年终止对一家右翼网站的服务时——也是 “去平台化” (deplatforming) 的早期前身 —— Prince对这一决定提供了以下解释，在言论自由和正当程序之间做出了重要区分：
“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。”【粗体字强调是我另加的】
开源Twitter的算法符合正当程序的要求，因为“道路规则”就会被公开。如果这些规则确实是武断的，也许在Twitter内部某些有权人有意影响某些规则，插手让一些人的言论获得比其他人更多的关注，我们现在就有证据了！这些证据可以被用来改善AI的管制法规（Elon赞成某种程度AI管制），以及每个人在Twitter上行使言论自由的用户体验（对所有推客都有好处）。
所有这些都可以在不损害Twitter业务的情况下进行。它甚至可以促进增长业务，因为平台变得更加透明，可信度更高，就会吸引更多的用户，更多的推文，更多的广告商。开源本身不是一种商业模式，但开源项目一直是许多创新商业模式的基础——Open Core、云SaaS、企业服务——是许多市值数十亿美元的企业背后的动力，比如MongoDB、Elastic、GitLab、Hashicorp等。Twitter没有理由不成为下一个大的“开源”公司。
作为一家公司，Twitter多年来一直在为开源做贡献，拥有一个专门的开源团队，并在GitHub上保持活跃。它已经具备了将其算法开源并做好的专业知识，唯一缺乏的是上层领导力和对其算法的商业价值的清晰认知。
一个信任的时代，不仅是便利的时代
2013年，Twitter的创始人和前CEO之一Ev Williams分享了一个深刻的见解：互联网只是为了实现便利，仅此而已。这一洞察推动了他的第一家公司Blogger的崛起，然后是Twitter。这种见解，即最大限度地方便人们做他们几千年来一直在做的事——表达自己、寻找信息、玩梗、找乐子、嘲弄他人——是每个主流社交媒体平台的核心。
虽然对便利性的需求一直没有改变，但我认为正在改变的是，人们对信任的要求比以往更强烈了。大家都希望社交媒体平台有完善的“正当程序”，赢得用户的信任，这样就可以继续安心地表达自己、寻找信息、玩梗、找乐子，以及嘲弄他人了。
开源其算法可以帮助Twitter在一夜之间树立这种信任。只有极少数的Twitter用户会定期进入代码库，仔细研究最新的变化，但 “只要想，就可以随时查看” 这一事实才是最重要的。
作为Twitter股东的受托人，Twitter董事会按理说应该欢迎开源其算法这个想法，因为这对Twitter的品牌、公众信任和业务增长都有好处。让Elon加入并推动这一能改变整个行业的举动，应该大受欢迎。但事实恰恰相反，Twitter董事会采用了“poison pill”来阻止Elon收购公司。
Elon在采访中暗示，如果收购Twitter失败，他还有个Plan B。事实上，这个计划很可能是建立一个替代Twitter的开源版。如果走出这一步，他将带走所有更信任Elon而不信任Twitter的Twitter用户。