Two cyclists were killed in hit-and-run collisions in San Francisco in the summer of 2016. A group calling itself the San Francisco Transformation Agency — the SFMTrA, a deliberate riff on the city's official Municipal Transportation Agency — responded by illegally installing traffic cones and, later, soft-hit delineator posts along Golden Gate Park's John F. Kennedy Drive. The city's reaction was the remarkable part: instead of tearing them out, officials announced plans to make the installation permanent. The SFMTrA operates as a collective of anonymous tactical urbanists whose stated objectives are threefold: provide immediate safety improvements for cyclists and pedestrians, draw public attention to dangerous road segments, and pressure the SFMTA to act faster on street safety. Most guerrilla urbanism follows a predictable arc — install, photograph, get removed, generate a news cycle, repeat. The Golden Gate Park intervention broke that pattern because the group escalated from easily-damaged orange cones to more robust safe-hit posts, which raised the implicit cost-to-remove and handed the city a ready-made pilot to ratify rather than replicate. The SFMTA acknowledged the legal bind directly. "Generally, we have no choice but to remove cones and posts that do not go through an official process," the agency wrote, citing code violations and traffic conflicts. But rather than enforce removal, officials said they would install their own permanent posts and leave the guerrilla ones in place during the review process. The framing matters: the city treated the illegal installation as a prototype, not a provocation. San Francisco is not alone in this dynamic. In Seattle, a group called the Reasonably Polite Seattleites installed similar delineators, which the city initially removed. City traffic engineer Dongho Chang then wrote an unusually gracious letter acknowledging the activists' investment of "time, money and risk," offered to return the confiscated posts, and ultimately coordinated with the Washington State Department of Transportation to reinstall them officially. In New York, the group Right of Way went further — painting an entire unsanctioned bike lane after a community board rejected an official extension, citing a taxi-pedestrian accident as the catalyst. The structural pattern across all three cities is the same: a safety failure (deaths, injuries, near-misses) exposes a gap between what the official process can deliver and what street-level conditions demand. Activists fill the gap with physical infrastructure rather than petitions or public comment. The city then faces a choice between the political cost of removing a visible safety improvement and the bureaucratic cost of legitimizing an unauthorized one. In San Francisco and Seattle, officials chose legitimization. This matters because it inverts the normal flow of urban infrastructure decisions. Typically, cities plan, fund, review, build, and then communities respond. Here, communities build and cities respond. The 27 miles of protected bike lanes San Francisco had already constructed provided political cover — the guerrilla posts aligned with an existing policy direction. That alignment is load-bearing. In jurisdictions hostile to cycling infrastructure, identical tactics have led to fines, arrests, and no policy change. The twenty-year question is whether tactical urbanism becomes a recognized feedback mechanism in urban planning — a way for cities to crowdsource small-scale pilot projects at zero public cost — or whether it remains a rare, context-dependent exception that works only when activists happen to push in a direction the city already wanted to go.