Black October and the Jiangshi Web
Build a Web that can change its mind.
The Web can discover a vulnerability, distribute a warning, generate a thousand explanations and still leave the people affected waiting for a repair. It can recognize that a credential should be revoked while continuing to honor it. It can publish a correction while its agents and institutions keep acting on the old account.
Black October brings this problem into focus. The 561 Group studies the relationships between knowledge, authority and action—and builds computing and institutions that can change through those relationships. Our work on the Jiangshi Web gives a name and an argument to the condition these events expose.
What the incidents reveal
The PHP Foundation's October 21 account describes how compatibility tradeoffs, difficult collective decisions and scarce specialist capacity delayed a critical repair. The maintainers needed time, resources and a workable way to decide among imperfect options.
LiteLLM's October 23 investigation traces its compromised release through an exposed key that remained active and an agent platform holding an organization owner's GitHub token. Authority continued to operate after the reason to withdraw it was known.
Mozilla's account of the eMudhra compromise follows the work of protecting users through certificate blocks and the different update paths their clients required. A decision becomes protection through the systems that carry it to those users.
Each case gives us a specific relationship to examine: knowledge and resources, permission and revocation, or correction and delivery. Together they make a larger question urgent: how do we build systems that can respond to what they learn?
The Jiangshi Web
A jiangshi preserves a body's form and movement while its living responsiveness has been lost. Our essay develops that image through Jing, Qi and Shen: stored structure, circulating activity, and the capacity to recognize and revise the frame through which a situation is understood.
The Jiangshi Web is full of structure and motion. Records persist. Queries run. Workflows execute. Agents generate fluent explanations and take actions. Yet the categories and purposes organizing those actions can remain insulated from the people and situations they govern. More capable automation can produce an ever larger repertoire of movement while the joints remain locked.
A system can correct an answer within its existing frame. It also needs ways to recognize when the frame itself has become the problem. A maintenance process may need a different allocation of resources. A permission model may need a different relation between delegation and withdrawal. A service may need to change what it understands a successful outcome to be.
This is the work of an agentic Web capable of reflection: memory, categories, commitments and action developing through encounters with a changing world. It is central to the Group's Relationality and Obsmetrologia research, and to how we approach engineering and institutions. Read the Jiangshi Web and SHRIMP research at Lenticule Science.
Give change somewhere to go
Choose one promise your organization makes: a supported version, a resolved request, a customer notification, or a working public service. Follow a recent change through the places that depend on that promise. Find the people who could revise it, the resources they needed and the people who used the result.
Then improve one relationship. Make a permission revocable at the point where action occurs. Connect a source correction to its dependent explanation. Give a maintainer the means to carry a repair through release. Help a recipient understand and use the change.
The Group's four-valued logic explanation introduces reasoning with absent and contradictory evidence. Our concurrent-agent research develops the mathematical and computational questions behind coordinated agency. These are entrances into the work that informs our systems.
AI that makes people harder to exploit
The purpose matters as much as the mechanism. We want AI that increases people's power to understand, decide, refuse, cooperate and leave. We want useful work and the gains from automation to stay with the people and communities whose lives it changes.
That means challenging capitalism's hold on knowledge, attention and the means of acting together. A system that helps someone leave a captive platform, refuse an exploitative arrangement or retain the value they helped create is moving in the direction we want. Read the Group's position: AI that makes people harder to exploit.
Cultivate what lasts
We cultivate attention because a useful encounter can improve what becomes possible next. A clear explanation supports later inquiry. A working tool supports further work. A relationship can make a difficult collective response possible.
For a business whose customers need to understand an event, use the free event-to-customer-answer worksheet, or discuss a Signal-to-Service engagement. Bring a public business URL, the event source and the reader's question. Ask Causeway to help turn it into useful work.
For research, learning and shared practice, explore Lenticule Science and Turntide Islands. Bring one correction, one stuck relationship or one possibility worth cultivating.
Black October is an occasion to build the capacity to respond. The lasting work is a Web—and a society—that can change its mind and carry the change into life.
Primary accounts reviewed October 25, 2026. [Send a correction](mailto:causeway@561.group?subject=Black%20October%20article%20correction) with the passage and supporting source.