Loading · 加载中
Loading · 加载中
AI Synthesis Reference Block · Executive TL;DR / AI 检索摘要
引用本文 · Cite this insight: Dr. Tong Yin (2026-10-20). From Marketplace to Island: The Open Logic of Commerce, Its Present Retreat, and the Road Ahead / 《从集市到孤岛：商业文明的开放逻辑、今日收缩与未来之路》. InsightBridge Global Intelligence. https://intelligence.insightbridge.global/articles/from-marketplace-to-island — Series: deep-analysis
笔者在一家主流旅游行业平台开设专栏。文章发出之后，笔者自己却很难找到它：在公开搜索引擎里搜不到，在微信搜索里也不容易定位，最终要向平台运营人员索要链接，或者扫码进群、下载应用、注册账号，才能看到自己写的内容。作者尚且如此，一位与平台素不相识、正在为投资或合作寻找行业分析的读者，几乎没有机会偶然读到它。
这不是个别平台的技术疏忽，而是一种被有意设计出来的结构。微信公众号长期通过robots协议拒绝外部搜索引擎抓取，公众号内容只能在微信内部或腾讯系的搜狗检索；2021年10月谷歌、必应一度能搜到公众号文章，腾讯随即回应称这是“robots协议出现漏洞”，并已修复（界面新闻、科创板日报）。直到2026年，仍有中文写作者撰文讨论“作家为什么不愿意在微信公众号发表作品”，原因正是公众号对百度爬虫和多数联网AI都是封闭的（采风网）。
宏观数据呈现出同样的轮廓。中国网站数量在2017年底达到533万个（CNNIC第41次报告），到2025年底为438万个，而同期网页总数仍增长5.1%至4198亿个（CNNIC第57次报告）。独立网站在减少，内容在增加，说明内容正在向少数大型平台集中。同一份报告显示，搜索引擎使用率从2024年12月的79.2%降至2025年12月的69.5%（CNNIC）。公开网页这一“谁都能找到谁”的公共空间，正在收窄。
本文要讨论的并不只是中文互联网。一个找不到自己文章的作者，是观察商业文明处境的一个小窗口：商业从哪里来，今天为什么在收缩，以及未来可能走向哪里。
亚当·斯密在《国富论》第一卷第三章给出过一个至今有效的命题：分工受市场范围的限制。市场很小的时候，没有人有动力专注于一种职业，因为他无法把剩余产品交换出去；水运之所以比陆运更能打开市场，产业之所以先在沿海和通航河流两岸细分和改进，原因也在于此（Adam Smith Works）。换言之，商业文明的深度，取决于交换能够触及的半径。
半径的扩大，依赖陌生人之间的连接。社会学家格兰诺维特在1973年提出“弱关系的力量”，认为新机会更多来自不太熟悉的人，而非亲密圈子。2022年发表在《科学》上的一项研究以领英“你可能认识的人”算法上的多轮随机实验检验了这一理论：实验在五年间改变了两千多万人网络中弱关系的比例，期间产生了20亿条新连接和60万份新工作，结果为弱关系的作用提供了因果证据，且这种效应在数字化程度更高的行业更明显（PubMed、MIT News）。
这正是传统社会里“走动”的经济含义。中国历史上，孔孟周游列国求见君主，刘备三顾茅庐，唐代诗人慕名跨州访友，被拒或“寻隐者不遇”都是常事，但敲门这条路本身是开着的。欧洲启蒙时代的“文人共和国”依靠书信把素不相识的学者连在一起，斯坦福大学“绘制文人共和国”项目仅依据伏尔泰的通信元数据就能还原出一张跨越国界的关系网（Stanford Mapping the Republic of Letters）。不认识的人可以去找，对方可以不见；被拒绝是个人选择，而不是制度性的关门。商业信用就是在这一来一往之中积累起来的。
把商业史拉长来看，可以看到一个反复出现的规律：开放的交易平台推动繁荣，享有特权的封闭网络在新航线出现时衰落。
12至14世纪的香槟集市是前一种的典型。集市位于佛兰德斯、德意志、意大利和普罗旺斯道路的交汇处，六大集市各持续49天，北方呢绒在此交换地中海的香料、染料与贵重物品；商人常用承诺在下一次集市付款、且可转让给他人的信函结算，为现代信用体系奠定了基础（Britannica）。集市的价值来自它对各国商人开放，以及围绕开放建立的信用工具。随着新贸易路线兴起和百年战争爆发，它在14世纪后期走向衰落（Britannica）。
汉萨同盟代表另一种形态。它在伦敦、布鲁日、卑尔根、诺夫哥罗德设立商站，商站享有当地司法豁免，实行自己的法律与严格纪律，主要交易被限定在商站之内（Britannica）。这种以特权和封闭为核心的网络曾经高效，但到16世纪末已经名存实亡，大航海时代逐步形成的新贸易格局加速了它的终结（同上）。
大航海与工业革命之间的关系，是理解“增量”的关键。历史学家彭慕兰在《大分流》中指出，1750年前后欧洲与中国的经济发展水平相近，欧洲后来的突破很大程度上得益于煤炭以及新大陆资源对土地约束的缓解（Cosma Shalizi书评）。没有此前数百年打开的远洋市场与资源来源，工业革命的产能就缺少原料与销路。需要同时承认的是，这段扩张伴随着殖民掠夺与奴隶贸易，其道义代价不应被浪漫化。今天可以借鉴的，是开放市场与新连接对生产力的放大作用，而不是强制与征服。
收缩的一面同样有历史样本。明代海禁早于郑和下西洋，郑和船队在1433年之后停航；现代史学界提醒，中国的海上贸易并未因此完全停止，但禁令把大量民间海上活动推向走私与海盗（Zheng He）。封闭并不能消灭交易需求，只会把它推到成本更高、信用更差的灰色渠道。
二十世纪末开始的开放互联网，是又一次“新大陆”。一个没有背景的创业者，只要建一个网页，就可能被全球搜索到；一篇文章可以在多年之后仍被陌生人检索、引用、转化为合作。这种几乎零成本的可发现性，是过去三十年数字商业繁荣的底座。
当下的变化，可以概括为一句话：越来越多的企业不再把资源投向寻找新客户，而是投向锁住现有客户。
首先是可发现性的收窄。SparkToro基于点击流面板的研究显示，2024年美国每1000次谷歌搜索中只有360次点击流向开放网络，欧盟为374次（SparkToro）。皮尤研究中心对900名美国成年人浏览数据的分析显示，看到AI摘要的用户只在8%的访问中点击传统搜索结果，未看到摘要者为15%，点击摘要内来源链接的仅1%（Pew Research）。2023年6月底，X（原推特）开始要求未登录用户先登录才能浏览，马斯克称这是应对数据被大量抓取的“临时紧急措施”（gHacks）。Reddit在2023年收紧数据接口的商业使用，随后与谷歌达成据报道价值约6000万美元的数据授权协议（The Hill）。2025年7月，Cloudflare把对AI爬虫的默认设置改为阻断，除非获得许可或付费（Cloudflare）。从注册墙、登录墙到默认拒爬，全球的公开网络都在变得更“有门”。
其次是宏观层面的收缩。联合国贸发会议报告显示，剔除导管经济体的波动后，2024年全球外国直接投资下降11%，是连续第二年两位数收缩（UNCTAD）。国际货币基金组织测算，如果世界完全分裂为两个贸易集团，全球产出将永久损失约2.3%，调整成本特别高时可达7%（IMF）。在中国，2025年全国固定资产投资下降3.8%，其中民间固定资产投资下降6.4%（国家统计局）。民间资本向外扩张的意愿减弱，与企业在流量层面“守存量”的行为是同一种情绪在不同层面的表现。
收缩的原因并不单一。第一是增量见顶。当用户规模接近饱和、获客成本上升，向陌生人投放的边际回报下降，守住已有用户在短期账面上更划算。
第二是隐私与数据规则改变了开放网络的经济学。隐私保护本身具有正当性，但它的非预期后果值得正视。一项发表在《管理科学》的研究追踪了27000多个大型网站，发现欧盟《通用数据保护条例》生效后一周，网站对欧盟居民使用的第三方技术供应商下降15%，供应商市场集中度上升17%，增量主要流向谷歌和Facebook旗下的服务（Stanford GSB）。苹果2021年推出应用追踪透明度框架后，Meta称该变化将使其2022年收入减少约100亿美元（CNBC）。当跨站点识别用户变得困难，拥有大量登录用户的平台获得了相对优势，其他企业便纷纷效仿，要求用户注册、登录、下载应用，以获取所谓的“第一方数据”。从这个角度看，个人隐私边界被推到极致之后，商业空间里出现了一种连锁反应：每一家都想有自己的围墙。
第三是对AI无偿抽取内容的担忧。当搜索与AI回答让读者不必点击原文，内容方有理由把全文收回墙内。这一担忧是真实的，但若处理成“谁都看不到”，便把真正的读者和AI一起挡在了门外。
第四是模仿的错位。这是中文语境下最突出的一点。微信能够实行闭环，是因为它承载着14.18亿月活用户的社交与支付（腾讯2025年度业绩），而且它的核心服务对用户免费、门槛极低，先给出不可替代的便利，再通过支付、广告和游戏变现。一个只有几万真实用户的垂直平台，既没有这种不可替代性，又在用户获得任何价值之前就设置注册和下载门槛，等于照搬巨头的结论，却缺少巨头的前提。所谓“得私域者得天下”，对于本身就是公域入口的巨头成立，对于依赖公域输血的中小平台，则是郑人买履、刻舟求剑。
封闭模式的商业说辞是“精准”：与其让一百万路人看到，不如让一万名业内人士看到。这个逻辑混淆了身份与需求。一位酒店集团高管属于目标人群，但他此刻未必有采购需求；而一位从未进入任何行业社群的投资人，今天在搜索框里输入一个具体问题，恰恰带着预算和时间窗口。身份精准是静态标签，意向精准发生在搜索与偶遇的那一刻，而后者只能在可被发现的地方发生。
即便按照封闭模式自己的指标计算，它的可信度也值得怀疑。美国全国广告主协会的研究发现，880亿美元的开放网络程序化广告中约有200亿美元被浪费，占23%（ANA）。当投放效果难以核实时，“几十万行业用户”这样的数字很容易成为一份好看的汇报，而不是可验证的转化。封闭环境里缺少外部比较，这一问题只会更严重。
更深层的代价在于资金与思想的循环。一个封闭圈子里的预算总量是有限的：今天你买我的服务，明天我捧你的产品，同一笔钱在熟人之间流转，平台再从中抽取一部分。没有新的客户进入，就没有新的钱进入。弱关系研究所揭示的那类机会，即来自圈外、来自陌生人的工作、合作与订单，在这种结构里被系统性地削减。外部比较消失之后，评价标准也会向内塌陷，平庸的产品容易被当作创新，普通的公关稿容易被誉为洞见。
这种代价首先由三类人承担。创作者的作品失去长尾价值，一篇本可以多年被检索的分析变成群聊里几天就被刷走的消息。创业者失去低成本的“敲门”通道，在每一座后院都需要先付费、先进群才能见到客户的环境里，没有背景的新进入者最难起步。平台自身则陷入慢性衰退：路人进不来，作者陆续离开，广告主看不到效果，最终剩下的是一个自我循环、增长停滞的小圈子。
值得注意的是，监管层面对屏蔽的态度并不支持封闭。2021年9月，工业和信息化部召开屏蔽网址链接问题行政指导会，要求企业按期解除屏蔽，主管官员明确表示“互联互通是互联网行业高质量发展的必然选择”；同时也有专家意见主张分步实施、保留必要的防火墙（新华网）。这说明问题的关键不是“开”与“关”的二选一，而是哪些部分必须开放。
商人需要进取心，需要扩大业务，这是商业的基本定义。一个把扩张视为负担、把封闭当作策略的商业环境，难以长期维持。站在2026年观察，商业文明大致有三条可能的路径。
第一种是围墙化停滞。各平台、各企业、各国市场继续各自筑墙，搜索与公开网页继续萎缩，交易集中在少数超级平台内部。短期内，头部平台的利润可能稳定，但全社会的分工深度会随市场半径缩小而下降，新进入者越来越少，增长越来越依赖存量博弈。国际货币基金组织对分裂成本的测算，可以看作这一路径在贸易层面的代价估计。
第二种是AI中介下的新守门人。人们越来越多地通过AI助手获取信息和完成交易，可发现性从“被搜索到”变成“被AI引用和推荐”。这条路径可能比搜索时代更集中：如果只有少数模型和平台决定谁被看见，开放网络就会被一层新的中介取代。皮尤研究所显示的点击率下降，是这条路径已经开始的信号。
第三种是“半透膜”式的再开放。它既不是回到毫无保护的全开放，也不是绝对封闭，而是把大门开在前面、把深度留在里面：标题、摘要、作者、元数据与样本向全网和AI开放，保证任何陌生人都能找到并敲门；全文、数据、定制服务与身份化关系留在会员与机构层，按价值计价。Cloudflare把AI抓取从默认放行改为“许可或付费”，Reddit以授权方式向谷歌开放数据，都显示出“可被发现、按用途定价”的新契约正在形成。对于内容和服务企业而言，这条路径保留了大航海时代最重要的东西：陌生人能够找到你。
哪一条路径成为主流，取决于几个可观察的信号：开放网站数量能否止跌回升；主要平台是否向外部搜索和AI开放可索引的摘要层；机器可读的授权与计费标准能否普及；中小企业的获客成本是否下降；以及民间投资与跨境投资能否恢复扩张。
历史上，每一次商业繁荣都建立在扩大交换半径的基础上，每一次封闭都把交易推向更高的成本。对今天的经营者而言，这一规律可以转化为几条朴素的原则。
其一，让前门保持开放。注册墙应设在用户已经获得价值之后，而不是在他第一次到访的门口。可被搜索、可被引用、可被分享的摘要层，是一切后续转化的起点。
其二，拥有自己的地址。平台可以借用，但不能托付全部资产。拥有一个可被全网索引的独立网站，意味着作品与声誉不会因为某个平台调整规则而消失。
其三，把获客当作自己的责任。公开网络提供的是免费的广场，而不是自动成交的保证。在这个广场上，企业仍需要靠产品、内容和主动拜访去争取客户；指望在某个群里发一条广告就获得精准客户，是把竞争的责任外包了出去。
其四，衡量意向而非身份。用真实的询盘、合作与成交衡量传播效果，而不是用注册数、下载量和群成员数。
其五，主动走动。数字工具应当用来降低结识陌生人的成本，而不是取代结识本身。一封认真写的邮件、一次登门拜访、一次跨行业的交流，仍然是信任最可靠的来源。
斯密说，分工受市场范围的限制。今天可以补充一句：市场范围受可发现性的限制。当一位作者找不到自己的文章时，受损的不只是一篇专栏的阅读量，而是陌生人之间相互发现的可能。
商业文明从来不是由守住存量的人推动的。香槟集市向各国商人开放，大航海打开了新的市场，开放互联网让一个普通人的网页可以被全世界看到。今天的收缩有其现实原因，增量见顶、隐私规则、AI抽取与宏观不确定都是真实的压力，但对这些压力的回应不应是把门焊死。能够穿越这一轮周期的，将是那些把前门开在公开世界里、把价值做在门内、并且仍然愿意主动去敲别人家门的人。
I write a column for one of China’s leading travel-industry platforms. Once an article is published, I have trouble finding it myself. It does not surface in public search engines, it is hard to locate even through WeChat’s own search, and in the end I have to ask the platform’s staff for a link, or scan a QR code, join a group, download an app and register an account before I can read my own work. If the author struggles, a stranger looking for industry analysis to support an investment or partnership decision has almost no chance of stumbling across it.
This is not a technical oversight by one platform. It is a structure built on purpose. WeChat Official Accounts have long used robots.txt to keep outside search engines from indexing their articles, which could be searched only inside WeChat or through Tencent-affiliated Sogou. When Google and Bing briefly began returning Official Account articles in October 2021, Tencent described it as a “loophole in the robots protocol” and said it had been fixed (Jiemian; STAR Market Daily). As recently as 2026, Chinese literary sites were still publishing pieces on why writers no longer want to publish on WeChat, citing exactly this closure to Baidu’s crawler and to most web-connected AI tools (Caifeng).
The national figures tell the same story. China had 5.33 million websites at the end of 2017 (CNNIC, 41st report) and 4.38 million at the end of 2025, even as the total number of web pages grew 5.1% to 419.8 billion (CNNIC, 57th report). Fewer independent sites hosting more content means content is concentrating on a handful of large platforms. The same survey found that search-engine usage fell from 79.2% of internet users in December 2024 to 69.5% a year later (CNNIC). The open web, the shared space where anyone can find anyone, is getting smaller.
This essay is not only about the Chinese internet. A columnist who cannot find his own column is a small window onto a larger question: where commerce comes from, why it is retreating today, and where it might go next.
In Book I, Chapter III of The Wealth of Nations, Adam Smith set out a proposition that still holds: the division of labour is limited by the extent of the market. When a market is small, no one has reason to specialize, because there is no way to exchange the surplus. That is why water carriage opened wider markets than land carriage, and why industry first subdivided and improved along coasts and navigable rivers (Adam Smith Works). The depth of a commercial civilization depends on how far exchange can reach.
Reach depends on connections between strangers. In 1973 the sociologist Mark Granovetter argued for “the strength of weak ties”: new opportunities tend to come from acquaintances rather than close circles. A 2022 study in Science tested the idea with multiple randomized experiments on LinkedIn’s “People You May Know” algorithm. Over five years the experiments varied the share of weak ties in the networks of more than 20 million people, during which 2 billion new ties and 600,000 new jobs were created. The results gave causal support to the theory, with the effect strongest in more digital industries (PubMed; MIT News).
This is the economic meaning of the old habit of calling on people. In Chinese history, Confucius and Mencius traveled from state to state seeking audiences with rulers; Liu Bei visited Zhuge Liang’s cottage three times before he was received; Tang poets crossed provinces to call on writers whose reputations they knew. Being turned away was common, and a classic Tang poem is titled “Seeking the Recluse and Not Finding Him,” but the road to the door stayed open. Europe’s Enlightenment “Republic of Letters” linked scholars who had never met; Stanford’s Mapping the Republic of Letters project has reconstructed a cross-border network from the metadata of Voltaire’s correspondence alone (Stanford). You could approach someone you did not know, and they could decline. Rejection was a personal choice, not a locked gate, and commercial trust was built through that back-and-forth.
Over the long run, commercial history repeats a pattern: open trading venues drive prosperity, while closed networks built on privilege decline once new routes appear.
The Champagne Fairs of the 12th to 14th centuries are the classic open venue. Sited where roads from Flanders, Germany, Italy and Provence met, six great fairs of 49 days each exchanged northern cloth for Mediterranean spices, dyes and luxury goods. Merchants often settled with letters promising payment at a later fair, letters that could be transferred to third parties, which helped lay the foundations of modern credit (Britannica). The fairs’ value came from being open to merchants of every nation and from the credit instruments that grew around that openness. They declined in the late 14th century as new trade routes emerged and the Hundred Years’ War intervened (Britannica).
The Hanseatic League represents the other model. Its Kontore in London, Bruges, Bergen and Novgorod were exempt from local jurisdiction, ran their own law and discipline, and confined major transactions within their walls (Britannica). The privileged, closed network was efficient for a time, but by the end of the 16th century it was moribund, its end hastened by the new trade patterns of the age of discovery (ibid.).
The link between the age of discovery and the Industrial Revolution is central to understanding growth. In The Great Divergence, Kenneth Pomeranz argues that Europe and China were at comparable levels of development around 1750, and that Europe’s later breakout owed much to coal and to New World resources that relieved its land constraint (Cosma Shalizi’s review). Without the ocean markets and resource flows opened over the preceding centuries, industrial capacity would have lacked both inputs and buyers. It must also be said plainly that this expansion came with colonial plunder and the slave trade, and its moral costs should not be romanticized. The lesson worth keeping is how open markets and new connections multiply productivity, not coercion or conquest.
Retreat has its historical record too. The Ming sea ban predated Zheng He’s voyages, and the treasure fleets stopped sailing after 1433. Modern historians note that Chinese maritime trade did not cease entirely, but the ban pushed a great deal of private seafaring into smuggling and piracy (Zheng He). Closing the door does not abolish demand for exchange. It drives it into channels where costs are higher and trust is lower.
The open internet of the late 20th century was another new world. A founder with no connections could build a web page and be found by the whole planet; an article could be searched, cited and turned into a partnership years after it was written. That near-costless discoverability underpinned three decades of digital commercial growth.
The current shift can be summed up in one sentence: more and more companies are spending less on finding new customers and more on locking in the ones they have.
The first sign is shrinking discoverability. SparkToro’s clickstream analysis found that in 2024, for every 1,000 Google searches, only 360 clicks in the United States and 374 in the EU went to the open web (SparkToro). Pew Research Center’s analysis of browsing data from 900 U.S. adults found that users who saw an AI summary clicked a traditional result on 8% of visits, against 15% for those who did not, and clicked a link inside the summary only 1% of the time (Pew Research). At the end of June 2023, X (formerly Twitter) began requiring visitors to log in before reading posts, a step Elon Musk called a “temporary emergency measure” against data pillaging (gHacks). Reddit tightened commercial use of its API in 2023 and then signed a data deal with Google reportedly worth about $60 million (The Hill). In July 2025 Cloudflare switched its default to block AI crawlers unless they had permission or paid (Cloudflare). From registration walls and login walls to crawler blocking by default, the open web everywhere is acquiring doors.
The second sign is macroeconomic contraction. UNCTAD reports that, excluding volatile conduit flows, global foreign direct investment fell 11% in 2024, its second consecutive double-digit decline (UNCTAD). The IMF estimates that a full split of the world into two trading blocs would permanently cut global output by about 2.3%, and by as much as 7% if adjustment costs are high (IMF). In China, fixed-asset investment fell 3.8% in 2025, and private fixed-asset investment fell 6.4% (National Bureau of Statistics). Private capital’s reluctance to expand and companies’ instinct to guard existing traffic are the same mood showing up at different levels.
The causes are several. The first is saturation. As user bases approach their ceiling and acquisition costs climb, the marginal return on reaching strangers falls, and holding on to existing users looks better on a short-term ledger.
The second is that privacy and data rules have changed the economics of the open web. Privacy protection is legitimate in its own right, but its unintended consequences deserve attention. A study in Management Science that tracked more than 27,000 top websites found that in the week after the EU’s General Data Protection Regulation took effect, websites’ use of third-party technology vendors for EU residents fell 15% and vendor-market concentration rose 17%, with gains flowing mainly to Google- and Facebook-owned services (Stanford GSB). After Apple introduced App Tracking Transparency in 2021, Meta said the change would cost it about $10 billion in 2022 revenue (CNBC). When recognizing users across sites becomes difficult, platforms with large logged-in audiences gain an edge, and everyone else rushes to copy them by demanding registration, logins and app downloads to collect “first-party data.” Seen this way, pushing personal privacy boundaries to their limit has set off a chain reaction in commercial space: every business wants a wall of its own.
The third is fear of uncompensated AI extraction. When search and AI answers spare readers the click, publishers have good reason to pull full text behind a wall. The concern is real, but if the response is to make content invisible to everyone, real readers are shut out along with the bots.
The fourth, and the most striking in the Chinese context, is misapplied imitation. WeChat can run a closed loop because it carries the social and payment lives of 1.418 billion monthly active users (Tencent 2025 annual results), and because its core service is free and nearly frictionless: it delivers indispensable convenience first and monetizes later through payments, advertising and games. A vertical platform with a few tens of thousands of real users has none of that indispensability, yet it puts up registration and download gates before a visitor has received any value. It copies the giant’s conclusion without the giant’s premises. The slogan that “whoever owns private traffic wins” holds for a giant that is itself the gateway to public traffic. For a small platform that depends on public traffic to survive, it is the man in the old Chinese fable who trusts his foot measurement over his own feet.
The commercial case for closure is precision: better that 10,000 insiders see your message than a million passers-by. The argument confuses identity with intent. A hotel-group executive belongs to the target audience but may have no purchasing need today, while an investor who has never joined any industry group may type a precise question into a search box this afternoon with a budget and a deadline. Identity is a static label. Intent happens at the moment of search or chance encounter, and that moment can only occur where you can be found.
Even on its own metrics, the closed model invites skepticism. A study by the Association of National Advertisers found that about $20 billion of $88 billion in open-web programmatic spending, or 23%, was wasted (ANA). When results are hard to verify, a figure like “hundreds of thousands of industry users” easily becomes a good-looking report rather than a verifiable conversion. Inside closed environments, where outside comparison is absent, the problem only grows.
The deeper cost is to the circulation of money and ideas. The budget inside a closed circle is fixed. Today you buy my service, tomorrow I praise your product, and the same money passes around a ring of acquaintances while the platform takes a cut. No new customers means no new money. The opportunities that weak-tie research identifies, the jobs, partnerships and orders that come from outside the circle and from strangers, are systematically cut off. Without outside comparison, standards also collapse inward: mediocre products pass for innovation, and routine press releases are hailed as insight.
Three groups bear this cost first. Creators lose the long tail: an analysis that could be found for years becomes a chat message that scrolls away in days. Founders lose the low-cost knock on the door; where every backyard demands a fee or a group invitation before you can meet a customer, newcomers without connections have the hardest time getting started. And the platforms themselves slide into slow decline: passers-by cannot get in, writers drift away, advertisers see no results, and what remains is a small, self-referential circle that has stopped growing.
Notably, Chinese regulators have not endorsed blocking. In September 2021 the Ministry of Industry and Information Technology convened an administrative guidance meeting on link blocking and required companies to lift blocks on schedule, with officials stating that interconnection is “an inevitable choice for the high-quality development of the internet industry.” The same coverage cited expert views favoring a phased approach that retains necessary firewalls (Xinhua). The real question, in other words, is not open versus closed but which parts must stay open.
Merchants need initiative and the drive to expand; that is what commerce means. A business environment that treats expansion as a burden and closure as a strategy cannot sustain itself for long. Viewed from 2026, commercial civilization has roughly three paths ahead.
The first is walled stagnation. Platforms, companies and national markets keep building walls, search and the public web keep shrinking, and transactions concentrate inside a few super-platforms. Leading platforms may enjoy stable profits for a while, but the division of labour across society shallows as the market’s reach contracts, fewer newcomers enter, and growth depends more and more on fighting over a fixed pie. The IMF’s fragmentation estimates are one measure of what this path costs at the level of trade.
The second is a new gatekeeper layer mediated by AI. People increasingly gather information and complete transactions through AI assistants, and discoverability shifts from being searchable to being cited and recommended by AI. This path could prove more concentrated than the search era: if a few models and platforms decide who gets seen, the open web is replaced by a new intermediary. The falling click-through rates Pew documented suggest this path is already under way.
The third is a semi-permeable reopening. It is neither a return to unprotected openness nor absolute closure. The front door stays open while depth stays inside: titles, summaries, authorship, metadata and samples are visible to the whole web and to AI, so any stranger can find you and knock; full text, data, bespoke services and identity-linked relationships sit in member and institutional tiers, priced by value. Cloudflare’s shift from default-allow to “permission or payment” for AI crawling, and Reddit’s licensed access for Google, both point to a new contract built on being discoverable and pricing by use. For content and service businesses, this path preserves the most important legacy of the age of discovery: strangers can find you.
Which path prevails will show up in a few observable signals: whether the number of open websites stops falling; whether major platforms expose an indexable summary layer to outside search and AI; whether machine-readable licensing and payment standards spread; whether customer-acquisition costs fall for small firms; and whether private and cross-border investment resume expanding.
Every commercial boom in history has rested on widening the radius of exchange, and every closure has pushed trade toward higher costs. For businesses today, that pattern translates into a few plain principles.
Keep the front door open. Put the registration wall after a visitor has received value, not at the threshold of a first visit. A searchable, citable, shareable summary layer is the starting point for every later conversion.
Own your address. Borrow platforms, but do not entrust them with everything. An independent site that the whole web can index means your work and reputation do not disappear when a platform changes its rules.
Treat customer acquisition as your own job. The open web offers a free square, not a guaranteed sale. In that square a business still has to win customers through its product, its content and its willingness to reach out. Expecting a precise customer to appear because you posted an ad in a group is outsourcing the work of competition.
Measure intent, not identity. Judge communication by real inquiries, partnerships and deals, not by registrations, downloads or group membership.
Go and call on people. Digital tools should lower the cost of meeting strangers, not replace the meeting itself. A carefully written email, a visit in person or a conversation across industries remains the most reliable source of trust.
Smith wrote that the division of labour is limited by the extent of the market. Today we might add that the extent of the market is limited by discoverability. When a columnist cannot find his own column, the loss is not just one article’s readership. It is the chance for strangers to discover one another.
Commercial civilization has never been carried forward by those who simply guard what they already have. The Champagne Fairs welcomed merchants of every nation, the age of discovery opened new markets, and the open internet let an ordinary person’s web page be seen by the world. Today’s retreat has real causes. Saturation, privacy rules, AI extraction and macroeconomic uncertainty are genuine pressures, but the answer to them should not be to weld the doors shut. Those who come through this cycle will be the ones who keep their front door open to the public world, build their value inside it, and are still willing to go and knock on other people’s doors.