pancakes

Non-Exploitative Infrastructure

Purpose

Pancakes and Pitchfork should be part of a broader effort to build non-exploitative versions of everyday digital infrastructure.

The problem is no longer only ads, social media feeds, or platform lock-in.

The problem is that ordinary life on phones and computers is increasingly captured, summarized, predicted, and repackaged by large AI-enabled platforms. Personal messages, photos, documents, location, work habits, social relationships, care work, health-adjacent behavior, creative work, and household life can all become raw material for automated interpretation.

The product should reject the assumption that every human action is platform input waiting to be monetized or summarized.

Core Thesis

People need local, cooperative, open, and non-extractive alternatives to the tools that mediate daily life.

This includes alternatives to:

• Search and knowledge tools.
• Social feeds.
• Short-form video.
• Photo sharing.
• Messaging.
• Calendars.
• Notes.
• Documents.
• Creative tools.
• Household management.
• Service exchange.
• Maps and location discovery.
• Cloud storage.
• Identity.
• Payments and credits.
• AI assistants.

The goal is not to clone every platform feature.

The goal is to rebuild the useful human functions without the exploitative defaults.

Exploitative Defaults to Avoid

Avoid systems that assume:

• Everything should be tracked.
• Everything should be uploaded.
• Everything should be summarized.
• Everything should train models.
• Everything should produce engagement.
• Everything should be ranked.
• Everything should be monetized.
• Everything should be shared by default.
• Users should trade privacy for convenience.
• Communities should depend on platforms they cannot govern.

These defaults are incompatible with Pancakes.

AI Boundary

AI can be useful, but it must not become ambient surveillance.

Do not build:

• A system that silently ingests a person’s life.
• A universal life summarizer.
• An assistant that reads everything by default.
• Household surveillance framed as helpful automation.
• AI-generated judgments of productivity, care, health, relationship quality, or worth.
• Model-training pipelines based on private logs without explicit consent.
• Dark patterns that pressure users to let AI process more data.

Build instead:

• Explicit AI tools.
• Local-first processing where possible.
• Narrow scopes.
• Temporary context.
• Clear consent.
• Clear deletion.
• Clear records of what was processed.
• Strong boundaries around health, children, relationships, household conflict, and financial data.
• Useful summaries only when the user asks for them.

The product should never feel like Clippy has returned with permission to summarize a person’s private life.

Replacement Pattern

For each platform category, ask:

What human need does this serve?
What does the exploitative version extract?
What would a local, cooperative, consent-based version look like?
What data can remain private?
What data must never be collected?
Can this run on a Pancakes node?
Can users export and leave?
Can a community govern it?

Examples:

Existing Pattern	Human Need	Non-Exploitative Direction
Instagram	Sharing life, beauty, memory, identity	Small-circle albums, consentful sharing, no engagement ranking
TikTok	Discovery, expression, performance, learning	Community video libraries, local moderation, no addictive infinite feed
Google Search	Finding knowledge	User-respecting search, local indexes, transparent sources
Google Photos	Memory and organization	Local-first photo archive, user-initiated AI labels, private albums
Google Docs / Microsoft 365	Collaborative work	Open documents, local-first sync, node-hosted workspaces
Adobe	Creative production	Open creative tools, community asset libraries, portable formats
TaskRabbit	Getting help	Pancakes service exchange, credits, covenants, local trust
Facebook Groups	Community coordination	Node-hosted groups, explicit membership, no ad targeting
Cloud storage	Continuity and backup	Node backup, user-held export, encrypted storage
AI assistant	Help thinking and organizing	Scoped assistant, local context, consentful summaries

Product Implications

Pancakes should become a container for humane alternatives, not a single monolithic app.

Possible modules:

• Pancakes Tasks.
• Pancakes Earth.
• Pancakes Groups.
• Pancakes Albums.
• Pancakes Notes.
• Pancakes Calendar.
• Pancakes Library.
• Pancakes Vault.
• Pancakes Identity.
• Pancakes Credits.
• Pancakes AI, if it exists, under strict boundaries.

Each module should be optional.

Each module should be able to run against a self-hosted node where practical.

Hosted Pancakes can exist, but it should not be structurally privileged over community-run nodes.

Design Rules

Use these rules when adding features:

• Default private.
• Explicit sharing.
• No hidden training.
• No hidden resale.
• No engagement-maximizing feeds by default.
• No public output leaderboards for life activity.
• No forced central account when a local account can work.
• Export before lock-in.
• Interoperable formats where possible.
• Open source core where possible.
• Community governance before marketplace scale.
• Human-readable permissions.
• Consent can be withdrawn.
• AI has a visible boundary.

Relationship to Pancakes Nodes

Pancakes nodes are the infrastructure answer to this doctrine.

A node can host:

• A household’s tasks.
• A guild’s covenants.
• A group’s albums.
• A local service exchange.
• A community library.
• A cooperative treasury.
• A private search or index.
• A shared calendar.
• A local AI tool with bounded access.

The node is not just a server. It is a way for people to own the digital structures that organize their lives.

Relationship to Pitchfork

Pitchfork supplies accounting primitives:

• Events.
• Resources.
• Credits.
• Covenants.
• Ledgers.
• Permissions.
• Caps.
• Pools.

Those primitives can support non-exploitative alternatives to platform systems, but they must remain subordinate to Pancakes’ human and social boundaries.

Pitchfork should never become the ledger of everything a person does.

Open Questions

• Which platform category should Pancakes replace first?
• Should Pancakes Tasks be the first non-exploitative module?
• Should Pancakes Groups or Albums come before Pancakes Earth?
• Can existing open-source tools be integrated instead of rebuilt?
• How should a node advertise available modules?
• What AI features are useful enough to justify their risk?
• How can users inspect what an AI tool has seen?
• How do small communities moderate content without reproducing platform abuse?
• What does export look like for albums, notes, covenants, tasks, credits, and identity?

Working Assumption

Start with the smallest useful replacement:

private tasks
-> household covenants
-> local service exchange
-> self-hosted node
-> optional community modules

Do not start with a giant anti-platform suite.

Build one trustworthy local tool, then let the ecosystem grow.