In May 2023, a startup called Spawning proposed ai.txt, a machine-readable file that website owners could place in their root directory to signal whether AI companies could use their content for training. Two companies signed on: Stability AI and HuggingFace. OpenAI, Google, and Anthropic did not.
The proposal was well-designed and genuinely intended. It was also a coordination mechanism proposed for what was, at its root, a power asymmetry. robots.txt had worked for two decades because search engines and publishers shared a reciprocal economy: crawl my pages, send me traffic. AI crawlers broke that reciprocity structurally. Twenty thousand pages crawled for every referral returned. But the coordination frame won anyway. It generated a GitHub repo, a specification, a blog post. The power-asymmetry frame, which would have meant something like mandatory revenue sharing, generated the prospect of a decade of lobbying.
By December 2025, OpenAI had quietly revised its crawler documentation, removing language indicating compliance with robots.txt directives. Research from Rutgers and Wharton found that publishers who blocked AI crawlers experienced a 23.1% decline in total traffic with no reliable reduction in AI citation rates. The voluntary mechanism penalized the people who tried to use it.
When RPA implementations began failing at scale in the late 2010s, the consulting ecosystem converged on a remarkably similar diagnosis: organizations needed a Center of Excellence. Better process selection. Stronger governance. UiPath stated it directly: failure statistics "are not a reflection of RPA as a technology" but of poor planning and oversight. The problem was coordination. The solution was more consulting.
What this frame excluded was who bore the costs. Forrester found that maintenance consumed up to 60% of total RPA implementation expenses. Those costs fell on IT teams maintaining brittle bots through every UI change, every security patch, every business reorg requiring cost center updates. Vendor license revenue was already collected. Executive sponsors had moved on.
Deloitte was named a global leader in RPA services by Forrester in 2019, praised for helping clients establish "a governance meta-framework." It was simultaneously one of the largest RPA services businesses in the world. Its revenue grew when clients needed repeated governance interventions. The coordination frame described the problem and, in doing so, generated the market for its own solution.
In both cases, the coordination frame captures something real. Coordination failures exist. Process selection matters. Opt-out mechanisms are better than nothing. But the frame persists because it locates the problem in implementation, leaving structure untouched. It asks for better tools, leaving power arrangements alone. And it is reliably held by people who will never check a robots.txt file to see if it's being honored. People who will never maintain a bot through a quarterly UI refresh, or watch their traffic decline for following the rules.