Strava declares war on scrapers ahead of IPO Strava 在 IPO 前打击数据爬虫

Another day, another walled garden going up. A company—name not important, the pattern is—is slamming the gates on its public data, yanking away guest access to user profiles and business listings under the banner of "security." The real, unspoken reason? To build a moat against the relentless, all-consuming AI data-scraping machines. This isn't a security update; it’s a white flag in the war for data, and we all lose when it's raised.

Let’s be blunt about the facts first. Previously, if you wanted to look up a gym or see a public profile, you could. It was a basic, almost foundational feature of the modern web. Now, you have to create an account, surrender your email, and become an authenticated user just to peek at what was once freely available. The company frames this as protecting its "assets" from "unauthorized AI scraping." In reality, it's punishing every legitimate human user for the sins of the bots. This is the digital equivalent of a bookstore deciding the only way to stop shoplifters is to charge everyone a $5 entry fee and demand ID at the door.

Here’s the thing about the "AI scraping" justification that stinks to high heaven. These models are trained on the vast, glorious, chaotic public internet. The entire ecosystem of search engines, social media, and niche forums we built over 25 years was fueled by this open ethos. The problem isn't that AI companies are scraping; it's that they're scraping too much, too fast, and with zero reciprocity. But the industry’s solution isn’t to build better, more ethical data pipelines or to establish fair compensation models. No, that’s too complex. The lazy, brute-force solution is to just make everything private. It’s scorched-earth policy for data.

This move fundamentally misunderstands—or perhaps willfully ignores—the purpose of public data. A public business listing isn’t a "scarce asset" to be guarded like Fort Knox gold. It’s an advertisement. It’s a tool for discoverability. Its entire value is derived from being seen by potential customers, not hoarded in a database behind a login wall. By hiding it, the company isn’t protecting its asset; it’s actively destroying its utility. It’s like a chef deciding the best way to protect his secret recipe is to stop serving food altogether.

And what about the users who willingly put their data there? The fitness club owner who listed their hours, the individual with a public profile—they did so under an implicit contract: "You, the platform, will make this data findable for the world, and in return, I participate in your ecosystem." Authentication shatters that contract. Suddenly, my data isn’t for people to find; it’s for our users to find, a much smaller, captive audience. The platform has unilaterally changed the deal, diminishing the reach and value I signed up for.

I can hear the tech executives' rebuttal: "But we have to protect our users’ data from being sucked into a large language model to be regurgitated!" This conflates two wildly different things. A public business listing with a name and address is not the same as private messages or health data. They’re lumping everything together to create a false sense of danger. It’s a smokescreen. The primary motivation isn't user protection; it’s data control. In the age of AI, the entity that controls the best datasets controls the future. This is a pre-emptive land grab, fencing off previously open territory to build a proprietary advantage for their own future AI products.

This is the most corrosive trend in tech right now: the end of the open web as a default state. We’ve seen it with the death of the "link economy" as platforms walled off their content, and now we see it with data itself. Every company is becoming a fortress. The result? A balkanized internet where information silos proliferate, search becomes less useful, and the serendipity of stumbling upon a new idea or a local business through open links withers and dies. Innovation thrives on accessible, connective tissue. We’re busy cutting it all.

The great irony is that this fortress-building is likely futile. Determined scrapers will find a way in. They always do. But the friction will be enough to stop casual developers, journalists, small businesses, and researchers who might have built something useful on this data. It’s the digital version of “if you can’t beat them, join them” gone horribly wrong: “If you can’t beat them, lock the door, turn off the lights, and hope no one notices you’re home.”

What this company has done is a canary in the coal mine. When the foundational act of viewing public information online becomes a logged, tracked, and permission-based activity, the nature of the web changes. It ceases to be a library and becomes a series of private showrooms. We’re trading the chaotic, vibrant, and sometimes messy public square for a series of sterile, authenticated shopping malls. And every time we click "Create Account" just to see a gym’s opening hours, we’re consenting to that bleak future. The bots aren’t the biggest threat to the internet’s soul. It’s this cowardly, enclosure-driven reaction to them.

互联网的公共广场，又竖起了一道需要门禁卡的玻璃墙。这一次，挥舞大锤的是某家健身数据公司，理由是“防范AI爬虫”。他们宣称，为了让AI无法再随意抓取其公开页面的用户资料、健身俱乐部列表等信息，现在所有这些数据都必须登录后才能查看。逻辑链条清晰得令人感动：为了防止被AI“偷”，所以先给所有人类访客设卡。

这动作本身并不新鲜，但时机和借口却透着一股熟悉的焦虑。过去几年，从新闻出版到社交媒体，从代码库到学术论文，无数网站都在悄然或激烈地对未经授权的自动化访问说“不”。核心矛盾始终如一：你的数据养大了谁的AI模型？谁该为此付费？但这次案例的特别之处在于，它指向了一个更具体的恐惧——不是担心巨头偷你的新闻版权，而是担心隔壁新冒出来的、专门健身的AI创业公司，用你的用户列表和门店信息，训练出一个能直接干掉你的垂直模型。

于是，我们看到了一个经典的技术性防御动作：数据上锁，仅对“认证用户”开放。表面上看，这合情合理，是保护用户隐私、防止数据滥用的正当之举。但如果我们把时间线拉长，嗅到的味道就复杂了。想当年，正是这些公司，高举“连接”与“开放”大旗，鼓励用户贡献数据，鼓励开发者通过API构建生态，用数据的流动性和可及性来证明自己的价值和平台潜力。数据越多，估值越高。如今，当数据的洪流可能流向一个无法控制的、潜在的颠覆者——生成式AI时，开放的闸门瞬间关上了。嘴上说的是“防范爬虫”，心里想的恐怕是“数据主权”和“竞争壁垒”。这并非简单的安全升级，而是一场以“安全”为名的资源封锁战。

这种“关门”行为，本质上是用工业时代的物理思维，来应对数字时代的信息流动。爬虫不过是自动化访问的工具，你加了登录墙，难道就防住了有心人注册账号进行规模化抓取吗？更深层的讽刺在于，这些公司如今视若珍宝、严防死守的数据，其价值恰恰是建立在过去的开放生态之上的。用户愿意留下信息，部分原因正是这些信息的半公开属性带来了某种社区感、比较性或简单的展示欲。突然上锁，等于单方面修改了社区契约。那些只是想分享自己健身记录的普通用户，和那些意图不轨的AI爬虫，在公司眼里被划上了等号，统统需要先亮明身份。这是一种典型的“宁可错杀，不可放过”的防御心态，其代价是用户体验的倒退和创新活力的抑制。

而真正值得玩味的是，这种“关门”姿态暴露了传统互联网公司面对AI浪潮时的一种深层无力感。他们无法阻止AI公司用其他公开数据或合成数据来训练模型，甚至无法阻止对手公司以合作之名堂而皇之地获取数据。他们唯一能控制的，就是自己这个“数据水源地”的水龙头。于是，一个看似主动的防御动作，底色却是一种被动的、基于恐惧的收缩。他们害怕成为AI时代的“数据奶牛”，却也想不出如何在开放中建立新的护城河。

这绝非个案。我们正目睹一场静默的数据内战：一边是亟需海量高质量数据来迭代能力的AI模型公司，另一边是坐拥垂直领域富矿、却日益担心被“吸干”的数据所有者。传统的“开放网络”共识正在瓦解，取而代之的是一个个需要登录、授权、付费的“数据花园”。互联网最初那个“任何人在任何地方都能自由访问信息”的朴素理想，在AI的阴影下显得越来越像个乌托邦笑话。公司们保护数据的动机可以理解，但用一刀切地降低可访问性作为解决方案，无异于为了防止有人偷菜，就把自家菜园子整个用电网围起来，连路过闻香的邻居一并电倒。短视，且懒惰。

未来，我们或许会怀念那个只需在搜索框输入关键词，就能看到一个公开页面上整齐列表的时代。数据将越来越珍贵，也将越来越封闭。而AI，这个靠数据喂养的巨婴，正逼迫着它的供养者们重新思考一切的归属和边界。这场关于“抓取”的战争，才刚刚开始。只是别忘了，当所有人都忙着锁门的时候，互联网那个曾经开着门的客厅，也就此熄灯了。