The impending ban or acquisition of TikTok and ByteDance’s lawsuit against the Trump administration have sucked up all the oxygen in the US-China tech zeitgeist. A gigantic IPO, possibly the largest in the world, by Ant Financials is also looming on the horizon. Consequently, one story with potentially much farther-reaching impact got buried.

That story is the Chinese Ministry of Industry and Information Technology (MIIT)’s anointment of Gitee, a homegrown Git-based developer collaboration platform, as the domestic “national champion” to drive open source growth in China. It got a blip on TechCrunch, and I included the original announcement in last week’s Interconnected Weekly. That's about it.

To understand why this development could have global significance, we need to first understand Gitee, MIIT’s role in China’s technology landscape, and open source’s own “gift culture” and global reach.

What is Gitee?

As readers of Interconnected know, I loathe the intellectually lazy comparisons that are common when talking about Chinese and American tech companies and try to avoid the “X is the [American company Y] of China” shorthand. Sure there has been plenty of cloning by Chinese companies of American success stories (and increasingly in the opposite direction), but both economies, cultures, and consumer behaviors are complex in their own ways and ought to be treated with nuance.

Sadly in Gitee’s case, this lazy framework happens to be rather accurate. It is the GitHub of China and nothing more.

History and Product

Gitee started in 2013 and was incubated out of OSChina, the country’s largest open source community website that started in 2008. OSChina owns Gitee, and even today the main contact email on Gitee’s homepage is a `oschina.cn` address. When Gitee started, GitHub (the current incumbent in the code repository hosting space) was a five-year-old Series A startup that was already gaining traction in various open source communities, especially the Ruby on Rails community.

If GitHub didn’t revamp it’s own UI just a couple of months ago, the two platforms would’ve looked virtually identical, except for a different color scheme and one displaying many more Chinese characters than the other. (The degree of similarity reminds me a bit of RenRen and Facebook in the early days of social media in China.)

It’s not completely fair to say Gitee has done nothing new. It does have more social tools, like direct messaging, which GitHub does not support. It also has an embedded web-based IDE (Integrated Development Environment, aka text editor for developers) though it’s not clear how many developers use it in place of existing popular solutions, like VSCode or IntelliJ. GitHub continues to add new features at a rapid clip, even after the Microsoft acquisition, like GitHub Actions and Discussions (in beta), so it’ll take quite a few years before Gitee can reach feature parity, if ever.

(Before writing this post, I tried to sign up for a Gitee account to experience it fully as a developer, but got stumped by a bizarrely difficult captcha process that took me at least five tries without success. I didn’t know I was this bad at reading fuzzy letters. This level of friction and poor user experience would’ve made any developer balk and leave.)

Market Traction

For the most part, Gitee is a typical venture-backed startup though it has not attracted much investment so far. In 2019, Gitee received a small strategic investment from Baidu and, in exchange, Baidu migrated some of its flagship open source projects, like its self-driving framework Apollo, to Gitee to seed more adoption. It claims to be the 2nd largest code repository platform in the world. This claim is not that far-fetched considering that China-based developers’ activities on GitHub is the 2nd largest group by a wide margin. If you include Hong Kong-based developers with China, it may be the largest cohort, exceeding that of the United States. It’s likely that a solid proportion of Chinese developers have Gitee accounts, along with GitHub, which would justify Gitee’s 2nd place claim.

Source GitHub Octoverse 2019: https://octoverse.github.com/

From the business side, Gitee’s go-to-market approach is a classic enterprise strategy, which is, again, not that different from GitHub, GitLab, or other similar products. It counts Huawei, China Telecom, and a handful of banks as enterprise customers -- all companies with national strategic importance. With MIIT’s official endorsement, Gitee’s enterprise traction within China will grow. That also means it will have no traction outside of China, but doing business globally was never part of Gitee’s ambition, though that contradicts its claimed purpose to foster open source (more on that later).

When I was writing my 3-part series “Open Source in China'' in May, I only briefly mentioned Gitee as one of the minor players in the ecosystem. MIIT’s anointment will boost Gitee’s profile and status, because it is now the designated “national champion”. There will be more government support for the platform. The company will get more investment from funds that like to align themselves closely with government initiatives. I can even see a day when Gitee will IPO on the Shanghai STAR exchange, similar to other much bigger “national champions”, like SMIC in the semiconductor industry.

There are still lots of unknowns to how Gitee will grow as a product. What is clear is that Gitee is now a “policy-driven” company that will rely heavily on the government direction-setting and protectionism, quite similar to its investor Baidu. While all current Chinese tech giants have benefited from some level of protectionism in the past, Alibaba, Tencent, Huawei, and now ByteDance have all grown their core technologies and products to a level where they can compete with other global players on those terms alone. Not Baidu. And not Gitee.

MIIT’s Role

MIIT is the most relevant central government ministry to pay attention to when it comes to China’s technology sector. It has embraced open source in an official policy planning document in 2016, titled: “2016-2020: Development Planning of Software and Information Technology Services Industry” (unofficial translation). This document laid out a few initiatives (more like proclamations) where open source was the centerpiece:

“...promote the effect that the open source community has to support innovation, strengthen the real-world application of open source technology in innovation, build an open source ecosystem that is advantageous to innovation in an open, collaborative, and global fashion...” (unofficial translation)

The endorsement of Gitee is the result of an official government project procurement in May that called for proposals of solutions for an “open source hosting platform”. This is perhaps the first step that MIIT took to concretely involve itself in open source.

MIIT official project procurement announcement

In the end, the group that won the project was a consortium of 10 organizations that included: Huawei, a few technical universities and think tanks, a few other companies, and, of course, OSChina. Thus Gitee.

Why did Gitee win? It’s really the only game in town. I highly doubt GitHub or GitLab submitted a proposal.

If MIIT was simply interested in embracing open source and lending its support to Chinese open source developers where they are working and collaborating now, GitHub would’ve been the obvious choice. On Zhihu, a Quora-like service in China where many developers congregate to discuss technology topics (thus a good proxy for assessing interest level), more than 84,000 people follow GitHub. Gitee? 243 (at the time of this writing).

How will all this play out?

There are not enough details to speculate. As the current four-year planning period ended in 2020, MIIT is likely working on the next four years of planning, where Gitee will play a significant role.

However, it is likely that any open source project that MIIT or its various sub-bureaus will use in the future will have to be hosted on Gitee. If the project is already on GitHub, the company or team that runs the project will probably have to maintain another copy on Gitee. This requirement may extend to many of the strategic tech sectors that are under MIIT’s jurisdiction.

All this is done to exert some level of national “control” on the open source technologies that are important to China’s path towards self-reliance, just in case GitHub becomes inaccessible due to more sanctions or policy changes from Washington.

This fear is not unwarranted. Right now, GitHub is accessible albeit with performance issues here and there, but no one can guarantee that will continue in this current environment between the US and China. Being a subsidiary of Microsoft, GitHub may also be caught up in the potential Microsoft-TikTok deal in ways that no one can predict.

There’s a lot to untangle, and the future is murky. Thus, some contingency planning is quite reasonable.

But is this attempt to “nationalize” open source in China going to work?

Open Source Can’t Be “Nationalized”

As you might’ve guessed by now, the title of the post is a rhetorical question. You can’t “nationalize” open source.

But the bigger point is: it’s dangerous to think you can.

It’s smart for a government to embrace the innovative energy of open source to boost its own technological advancement. China is doing it. I’ve advocated in this Wired opinion piece that the U.S. should be doing the same.

But no government can or should try to “nationalize” open source, because it contradicts the very nature and culture of open source that made it the technological force that it is today.

Eric Raymond, one of the founding members of the free software and open source movement, wrote a profound essay called “The Hacker Milieu as Gift Culture” in the 90s. In it, he explained that the open source hacker culture is predicated on the “gift culture”, where one gains status and respect by giving away work product and creativity. The “gift culture” stands in contrast to two other models: “command hierarchy” (aka top-down control) and “exchange economy” (aka the free market).

By controlling or “nationalizing” open source, or even trying to do so, you are converting something that’s based on a “gift culture” to either a “command hierarchy” or “exchange economy”. If that happens, then open source is no longer open source.

Now, “gift culture” only works if there’s a presumption of abundance, not scarcity. And Raymond’s essay cited fascinating anthropological examples where a “gift culture” is adopted in some aboriginal cultures, where they live in mild climates with abundant food.

I don’t blame the Chinese government for not feeling like they are living in “mild climates with abundant food”. Despite how rich China has become in a GDP sense, it is sorely lacking in core, industrial technology and still cannot stand on its own two feet without reliance on the United States and other countries. It is operating in a climate of scarcity, not abundance.

Thus, it is rational to approach these challenges transactionally. However, forcing a transactional or top-down model onto open source will kill the very thing the Chinese government hopes open source will fuel, as stated in MIIT’s policy document: innovation.

Bifurcating open source, by anointing Gitee and pushing other “nationalization” initiatives, will only slow down innovation, not speed it up.

The top-tier engineers in China (there are many) will continue to innovate on GitHub or other global platforms, and hop over the GFW if they have to, because that’s where they find the best information and collaborate with the best talent from around the world. (In fact, some top tier commercial open source companies implicitly make GFW-hopping a recruiting requirement by purposely hosting job descriptions as a Google Doc. If you can’t find a way to read the doc, you are probably not a good fit.)

Without the mindshare of your top-tier engineers, no amount of technology transfer will matter, because quite frankly the code isn’t worth much if you don’t have enough qualified people to understand, improve, and innovate on top of it.

In turn, China’s innovation will suffer and global innovation will suffer, because China’s best and brightest have been making important contributions globally in fields like AI, cloud computing, climate science, and many others. For China’s sake and the world’s sake, I hope MIIT understands that it cannot “nationalize” open source.

China’s posture towards open source is perhaps the most important thing that no one is paying attention to right now.

Until people do, I’ll just be working on my fuzzy-letter-reading skills, so I can hopefully pass Gitee’s onerous captcha test one day.

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.

能把开源“国有化”吗?

即将被禁或收购的TikTok和字节跳动对特朗普政府的诉讼,差不多吸干了中美科技舆论所有的氧气。蚂蚁金服的上市,很可能是全球规模最大的IPO,也会在不久的未来发生。因此一条影响更深远的故事被忽略了。

这就是工信部指定Gitee为国内推动开源增长的默认平台一事。这条消息在TechCrunch上有个简短的报道,我也把官宣放在了上周的《互联周刊》里。基本上就这么些报道。

要充分理解这条新闻为何具有全球意义,我们首先需要了解Gitee、工信部在中国科技领域般的角色,以及开源自身的“贡献文化”和全球影响力。

Gitee是什么?


《互联》的长期读者都知道,我很讨厌在谈论中国和美国科技公司时常见而懒惰的那种比较方式,也尽量避免用 “X是中国的(Y美国公司)”的那种简写。诚然,中国很多公司确实大量克隆了美国成功的产品(最近有越来越多地“反方向克隆”),但无论是经济、文化还是消费者习惯,两国都有其自身的复杂性,应该有更细致的分析去对待。

可惜的是,在看Gitee的时候,懒惰的框架碰巧也相当准确。它的确就是“中国的GitHub”,仅此而已。

历史和产品


Gitee成立于2013年,是OSChina孵化出来的。至今Gitee还是OSChina的产品之一,就连Gitee主页上的电子邮件箱还是`oschina.cn` 。Gitee刚起步时,GitHub(当今代码托管平台服务的行业老大)是一个仅成立五年的A轮公司,但已经得到许多开源社区的认可和采用,特别是Ruby On Rails社区。

如果GitHub前几个月前没改进自己的UI的话,那这两个平台看起来几乎是完全一样,除了颜色有点不同,一个的汉字比另一个多很多而已。这种“类似”让我想起了人人网和Facebook在中国社交媒体早起时的相似。

说Gitee什么新东西都没有也不算公平。它有更多的社交工具,比如私信,GitHub是不支持的。它也有个自己的Web IDE(集成开发环境,程序员的文字编辑器),虽然还不清楚有多少用户使用这个IDE,代替现有的更流行其他选择,如VSCode或IntelliJ。虽然被微软这个巨头买了,但GitHub仍然在快速地增加新功能,比如GitHub Actions和Discussions(beta),因此Gitee要达到功能对等还需要几年时间,甚至永远赶不上。

(在写这篇文章之前,我试图注册了一个Gitee帐户,想作为开发者的身份充分体验一下产品,但却被一个某名奇妙困难的验证过程阻止了,至少试了五次都没过。没想到我读模糊字母的能力有这么差!这种程度的摩擦和糟糕的用户体验应该会让绝大部份的开发者望而却步。)

市场和商业业务

在很大程度上,Gitee是一家典型的风投创业公司,虽然到目前为止,还没有吸引到太多的投资。2019年,Gitee从百度获得了一笔小规模的战略投资,之后百度把一些主要的开源项目(如自动驾驶框架Apollo)迁移到Gitee,以促进用户增长。它号称是世界上第二大代码库平台。考虑到中国开发者在GitHub上的活动程度以是当之无愧的世界第二,这说法并不牵强。如果再把香港的开发者加在一起,整个来算可能是最大的,超过美国。相当一部分中国开发者不仅有GitHub也有Gitee账户,所以Gitee第二名地位是说的过去的。

Source GitHub Octoverse 2019: https://octoverse.github.com/

从业务角度来看,Gitee是家经典的企业服务公司,这与GitHub、GitLab等其他类似产品没有太大区别。华为、中国电信和少数几家银行都是企业级客户,所有这些公司也都具有国家战略重要性。有了工信部的官方支持,Gitee在中国企业业内的吸引力将会增长。这也意味着,它在国外不会有什么吸引力和生意,但在全球开展业务从来不是Gitee雄心壮志的一部分,尽管这与它宣称要促进发展开源是相矛盾的(稍后会详细解释这一点)。

当我在5月份撰写《中国的开源世界》的三篇文章时,我只简单地提了一下Gitee是整个生态里的一个小玩家。工信部的任命将提升Gitee的形象和地位,因为它已经被指定为是“国家队”了。政府将对这个平台提供更多的支持。公司将很可能从那些喜欢与政府导向对齐的基金获得投资。我甚至可以看到有一天,Gitee会在上海科创板IPO,类似于其他更大的“国家队”队员,比如中芯国际(SMIC)。

Gitee作为一款产品将如何发展,还有很多未知数。但很明显的是,Gitee会是一家“政策驱动”的公司,将依赖政府的导向,做为保护主义的对象,这与投了它的百度非常相似。虽然许多中国科技巨头过去多多少少都曾从某种程度的保护主义中获益,但阿里、腾讯、华为和字节跳动都已将自己的核心技术和产品提升到可以与其他海外巨头完全以能力竞争的水平了。百度还不行。Gitee更不行。

工信部的角色

工信部是中国科技领域最需要关注的中央政府监管部门。它在2016年的一份官方政策规划文件中拥抱了开源,该文件的标题是:《软件和信息技术服务业发展规划(2016-2020年)》。本文件列出了一些以开源为中心的倡议(更像是宣言):

"…发挥开源社区对创新的支撑促进作用,强化开源技术成果在创新中的应用,构建有利于创新的开放式、协作化、国际化开源生态..."

支持Gitee的决定是5月份一个官方项目采购的结果,该项目采购从业界寻求“开源托管平台”的解决方案。这应该是工信部具体参与开源的第一步。

工信部招标公告

最终赢得该项目的是一个有10个组织组成的联合体,其中包括:华为、几所科技大学和智库、其他一些公司,还有OSChina,因此Gitee脱颖而出。

Gitee为什么会赢呢?因为没其他玩家。GitHub或GitLab应该都没有提交任何方案。

如果工信部只是想拥抱开源,给目前所有的中国开源开发者提供支持,那么GitHub应该是必然的选择。绝大部分开发者已经在GitHub上公开的协作和互动。在知乎上,有超过8.4万人关注GitHub。Gitee呢?243人(在撰写本文时)。

将来会怎么样?

目前还没有足够的政策细节可以推测未来。由于当前的四年规划期已经在2020年结束,工信部应该在制定未来四年的规划之中,Gitee也将在其中有更重要的作用。

可以推测的是,今后工信部或其各分局使用任何开源项目的前提应该是代码必须托管在Gitee上。如果项目已经在GitHub上,管理该项目的公司或团队可能必须在Gitee同步一个副本,从而需要维护两个代码裤。这个要求可能会延伸到工信部管辖的许多有战略意义的科技行业和领域里。

所有这些举动都是为了中国走向科技自力更生之路中向至关重要的开源技术施加一定程度的“控制”,以防万一由于华盛顿更多的制裁或政策改变而无法访问GitHub。

这种担心并非没有根据。目前,GitHub是可以访问的,尽管性能有些问题,但没有人能保证在目前中美关系的环境下,这种状态会继续下去。作为微软的子公司,GitHub可能还会卷入微软买TikTok的各种问题中。

未来很复杂,也很不明朗。因此,做一些应急准备都是合理的。

但这种把开源“国有化”的举动会奏效吗?

开源不能“国有化”

您可能已经猜到了,这篇文章的标题是一个反问句。开源当然不能“国有化”。

但更重要的一点是:认为能把开源”国有化“是危险的。

对于一个政府来说,拥抱和发挥开源的创新能量来推动自己国家的科技进步是明智之举。中国正在这样做。我在这篇Wired杂志的评论中也敦促美国这样做。

但是,没有一个政府能够,也不应该试图,将开源“国有化”,因为这与能让开源变成一股全球科技创新力量的本质和文化是矛盾的。

自由软件及开源运动的创始成员之一,Eric Raymond,在90年代写了一篇深刻的文章《黑客境界里的“贡献文化”》。他在文章中解释说,开源黑客文化是建立在一种“贡献文化””的基础上,在“贡献文化”中,人们通过无常赠送自己的工作成果和创造来获得地位和尊重。“贡献文化”与另外两种模式形成鲜明的对比:“命令式阶层”(或自上而下的控制)和“交换经济”(或自由市场)。

通过控制或“国有化”来对待开源,甚至试图这样做,将会把一个基于“贡献文化”的东西转变为“命令式阶层”或“交换经济”。如果这种情况发生,那开源就不再是开源了。

“贡献文化”能实现的基础是默认丰富资源而没有稀缺的这个前提。Raymond的文章引用了一些有意思的人类学例子,在一些土著文化中,他们生活在温和的气候中,食物丰富,因此演变出了自己的“贡献文化”。

我不怪中国政府没有觉得当前的世界不是个有“温和的气候,食物丰富”的环境。尽管从GDP的意义上说,中国已经变得很富有,但它在核心产业技术方面严重缺乏,无法不依赖美国和其他国家,还不能自立。它是在这种稀缺而非富足的环境下运作的。

因此,以交易方式处理眼前的挑战也是合理的。然而,把交易式或自上而下的模式强加于开源,将扼杀中国政府希望开源所能推动的东西,也是工信部政策文件明确提倡的东西:创新

公开认可Gitee以及推动其他“国有化”举措来分拆开源,只会减缓创新,而不会加速创新。

中国的顶级工程师(很多很多)只会继续在GitHub或其他全球性平台上工作和创新,需要翻墙就翻墙,因为只有在那种平台上才会找到最好的信息,与来自世界各地最nb的人才合作。(事实上,有些顶级的开源公司已经故意职位描述放在Google Doc上,从而把“能翻墙”作为招聘的要求之一。如果你找不到,看不了Google Doc,就不是合适人选。)

如果没有顶级工程师的参与,技术转让再多也没什么用。坦率地说,如果没有足够的合格开发人员来理解、改进和创新现有的科技,“控制”的代码再多也没有多大价值。

分拆开源的后果则是,中国的创新会受到影响,全球的创新也会受到影响,因为中国最优秀,最能干的人才在人工智能、云计算、气候科学等许多领域中一直在全球做出重要贡献。为了中国和世界的未来,我希望工信部明白这一点:它不能“国有化”开源。

中国对开源的姿态和做法也许是目前最重要的没有人关注的事情。

在大家开始把眼光转向这件事情之前,我也只能继续练习我的模糊字母阅读能力,争取有一天能顺利通过Gitee繁重的验证码测试,搞到个账户了。

如果您喜欢所读的内容,请用email订阅加入“互联”。新文章将会直接送到您的收箱(每周两次)。请在TwitterLinkedIn上给个follow,跟我互动交流!