Glossary

A family of plain-text and Markdown conventions located at well-known paths — robots.txt, sitemap.xml, llms.txt, AGENTS.md, design.md — that specify how automated readers (crawlers, indexers, language models, coding agents) may consume a site or repository. Each convention is advisory, operator-authored, and located at a fixed path; collectively they constitute the protocol surface of the Independent Internet.

An approach in which understanding accumulates: each source is read once and integrated into a durable, interlinked artifact, rather than re-derived at query time. Contrasted with retrieval-augmented generation (RAG), which reassembles an answer from raw fragments on each request and retains nothing between queries.

A design principle under which a system has no central owner, gatekeeper, or single point of failure; authority and data are distributed across independent participants that cooperate through open protocols. It is the structural basis of the Independent Internet and the principal counter to surveillance capitalism.

A resilient, self-hosted internet owned and operated by its users rather than by intermediary platforms: private by design, governed by the communities it serves, and able to function under degraded or absent connectivity. Also termed Web 4.0.

The use of filesystem layout — directories and plain Markdown files — as the orchestration mechanism for a multi-stage language-model workflow, in place of a code-level framework. Because each intermediate artifact is a human-readable file, interpretability is a property of the architecture rather than an added feature. Introduced by Van Clief and McDermott (2026).

A persistent, interlinked Markdown knowledge base authored and maintained by a language model: each source is integrated once so that knowledge compounds, rather than answers being re-derived from raw documents on every query. The pattern follows Karpathy’s LLM Wiki; this site’s wiki is an instance of it.

Peer-to-peer connectivity (e.g., LoRaWAN, Meshtastic) in which nodes relay traffic for one another, forming a network with no central point of control or failure. A key enabling technology for community-operated infrastructure.

The operation of one’s own services and storage of one’s own data in place of provisioning from a third-party provider. It is the most direct expression of control in the Independent Internet, removing the intermediary otherwise positioned to surveil or revoke access.

The proposal that web content carry structured, machine-readable meaning so that software agents can reason and act on it rather than merely render it for human readers. Articulated by Berners-Lee et al. (2001), it forms the bridge between the read-write web and the Independent Internet.

Zuboff’s term (2019) for an economic order that unilaterally claims private human experience as free raw material for extraction, prediction, and behavioral influence. It is the condition the Independent Internet is formulated to counter.

Synonym for the Independent Internet: the fourth era of the web, defined by ownership and control of the underlying compute and data, following the static (1.0), dynamic (2.0), and semantic (3.0) webs.