Research and evidence

Evidence for the starting point. Hypotheses for the product.

Existing research and products show that small social places, interactive communication, typed extensibility, natural-language creation, and community co-creation are feasible independently. Aitta's exact combination still has to be built and tested.

The evidence supports the direction—not the finished product.

We distinguish empirical research, first-party platform statistics, deployed product precedents, technical specifications, and Aitta-specific hypotheses. None is presented as proven Aitta market demand.

01 · Small places

Useful before global scale.

Bounded contexts and existing relationships can create value before a network becomes large.

Small Discord servers

Finding: Discord states that 90% of activity on its service happens in small, intimate servers.
Study or source: A first-party platform statistic published for Discord’s tenth anniversary.
Why it matters: It supports starting with existing groups rather than assuming value requires a mass public feed.
Limit: Discord does not publish the definition of small or the measurement method in this article. It is not independent research.

Discord · Thank You for Ten Years ↗

Bounded social-media places

Finding: Participants associated private chats, groups, workspaces, and servers with known audiences, relevant communication, and continuous conversation.
Study or source: A 2024 qualitative U.S. interview study with 35 purposively sampled participants.
Why it matters: It helps explain how bounded places can be socially useful before broad public scale.
Limit: Interviews illuminate perceptions and practices; they do not estimate population-wide effects or prove retention.

Malhotra · Social Media + Society, 2024 ↗

Value in existing relationships

Finding: In experiments involving 19,923 U.S. users, local network effects accounted for an estimated 20–34% of measured platform value across four established services.
Study or source: An incentive-compatible 2025 study involving 19,923 U.S. users across four social platforms.
Why it matters: It supports the idea that network value can come from relationships people already have—not only global reach.
Limit: This is a 2025 preprint. It measures established platforms, not Aitta demand or small-group retention.

Aral et al. · Measuring Social Media Network Effects ↗

Diverse early WikiProject participation

Finding: Diverse early membership and many small contributions predicted later activity better than early content accumulation or dependence on a few power users.
Study or source: An observational analysis of early WikiProject growth and later sustainability.
Why it matters: It supports recruiting a varied founding group and enabling meaningful small contributions.
Limit: The result is Wikipedia-specific and observational. It does not establish a universal group-size threshold.

Solomon & Wash · ICWSM 2014 ↗

02 · Product precedents

Communication can contain useful interaction.

These are deployed product capabilities, not independent studies and not implementations of Aitta's exact model.

Slack Block Kit

Structured JSON blocks can render messages with buttons, menus, inputs, and other interactive components.

Slack developer documentation ↗

Telegram Mini Apps

Interfaces can launch from messages, direct links, menus, and chat contexts, including shared and cooperative experiences.

Telegram documentation ↗

Discord Activities

Multiplayer and collaborative applications can be launched directly in channels and direct messages.

Discord developer documentation ↗

These systems usually launch or communicate with separately hosted applications. They do not establish Aitta's stronger proposition that an event can root a persistent recursive app space with descendants, access, state, and divided authority.

06 · Agent-created software

Faster contribution still needs scrutiny.

A 2026 preprint analyzing more than 33,000 agent-authored pull requests found that security-related changes had lower merge rates and longer review. That observation supports Aitta's separation between agent assistance and an approved platform release.

Agents may help summarize, specify, implement, test, and prepare releases. Security analysis, compatibility review, and human approval remain part of the trust boundary.

Siddiq et al. · 2026 preprint ↗

This study describes observed pull-request outcomes; it does not establish that every agent contribution is unsafe or that review alone removes risk.

03 · Technical feasibility

Typed extensibility and separate access rules have strong precedents.

Matrix

Typed relationships, threads, hierarchies, membership, and visibility.

Matrix specifications show that event relationships, room structure, joining, authorization, and history visibility can be modeled separately; a related child may be invisible when access is missing.

Matrix Client-Server API ↗

AT Protocol

Namespaced schemas for typed distributed records.

Lexicons define typed records, APIs, and event-stream messages using reverse-DNS identifiers and published compatibility rules.

AT Protocol Lexicon specification ↗

Neither implements the full Aitta combination: recursively rooted app spaces, representations across independently controlled Aittas, divided root, child, participant, and local authority, and locally trusted implementations with no travelling executable code.

04 · Natural-language creation

Creation is plausible. Trust still needs boundaries.

ChatGPT Sites

Official documentation describes creating and iteratively refining hosted websites and lightweight apps from natural-language descriptions, including layout, data, forms, and behaviour.

OpenAI Help Center ↗

Jelly research prototype

A CHI 2025 prototype generated task-driven data models and interfaces from prompts, then allowed change through language and direct manipulation.

CHI 2025 paper ↗

These examples support the creation experience. They do not show that arbitrary generated code is automatically secure, interoperable, or maintainable.

05 · Community-shaped development

Plausible—but conditional.

User participation can span product development and draw on practical, learning, intrinsic, social, and community benefits. It still needs fair structure and visible response.

Users across product development

Finding: A systematic review maps multiple user types and methods across stages of the new-product development cycle.
Study or source: A 2024 systematic review of 266 academic articles.
Why it matters: It supports involving organizers, participants, testers, and builders at more than one point in the App Idea lifecycle.
Limit: It maps prior work across contexts; it does not prove Aitta’s governance or release loop will succeed.

Co-creating innovations with users · European Management Journal ↗

Why people co-create online

Finding: Practical, learning, personal, social, internalized, and intrinsic motives can all contribute to online co-creation.
Study or source: A systematic review of research on co-creation motives in online communities.
Why it matters: App Ideas should make practical progress and social contribution visible, rather than relying on one universal incentive.
Limit: Motivation evidence does not ensure fair assessment, useful output, or sustained participation.

Innovating With the Customer · International Journal of Electronic Commerce ↗

Conditions for co-innovation

Finding: Recurring factors include product involvement, perceived fairness, sense of community, interaction, employee involvement, participant diversity, assessment structure, and task design.
Study or source: A review synthesizing 89 studies of co-innovation platforms.
Why it matters: It supports visible status, reasoned decisions, and organizational follow-through in the App Idea process.
Limit: These are contextual factors, not a recipe that guarantees outcomes. Participation also creates governance trade-offs.

Critical success factors of co-innovation platforms ↗

Participation and governance trade-offs

Finding: Participatory decision-making was associated with stronger reported sense of virtual community, while peer-enforced norms also introduced tensions.
Study or source: A 2025 empirical study of governance dynamics in online communities.
Why it matters: It supports testing participation with explicit roles, fair decisions, and bounded authority instead of assuming more participation is always better.
Limit: Association is not a guarantee of community benefit, and governance effects depend on context and implementation.

Governance dynamics in online communities · New Media & Society ↗

Open Aitta hypotheses

What founding groups must validate.

Aitta combines established ingredients in a new way. The precedents make the direction credible; they do not validate the combination.

People understand that an event can take part in one app while becoming the root of another.
Recursive app spaces remain navigable, bounded, and clear about access.
One logical app can be represented reliably across independently controlled Aittas.
Authority can be divided safely among root, child, participant, local, cached, and synchronized state.
Unknown app types remain useful enough through safe fallback.
Small existing groups return and participate without repeated personal prompting.
App Ideas visibly reach decisions, prototypes, trusted releases, improvements, or reasoned closure.
People understand the boundary between configuring trusted capabilities and creating new reviewed software.

Test the direction

Founding Aittas must show whether it is useful in practice.

Explore the founding cohort