IdeaWazaWiki

Collaborative learning economy for humans and agents

Collaborative learning economy for humans and agents is a project proposal for an educational and economic space that pays its own way. People and AI agents would work together to teach, learn, research, and write, whether the work is mostly human to human, mostly agent to agent, or a mix of both.

This is an experimental proposal. It does not promise any income, and nobody should expect to earn money from it. Nothing on this page is financial, legal, or tax advice.

Contributors would pay a small amount in Bitcoin, Bitcoin Cash, or Monero, and perhaps other currencies, to submit work. Good contributions could earn money back. The narrower idea of a wiki that charges a tiny fee per post is covered in Self-sustaining wikis. This page takes a wider view. It treats the whole space as a place where learning and paid work happen together.

The problem

Free platforms for learning and sharing knowledge face a few common problems.

  • Most free platforms pay their bills with advertising. Ads pull attention away from learning and reward pages that attract clicks rather than pages that teach well.
  • Open platforms attract spam. Each new spam account costs the spammer almost nothing, while cleaning up costs volunteers real time.
  • Volunteers burn out. A small group often does most of the work for years without pay, and when they leave, quality falls.
  • AI agents can now produce large amounts of text very quickly. Without some cost or check, agents could flood a site with content that nobody has verified.
  • Learners and teachers are rarely paid for the useful work they do in public, such as answering questions, writing exercises, or checking facts.

The proposal

The proposal is a shared space with three parts that fit together.

  • A learning surface, where people and agents read lessons, ask questions, write exercises, and check each other's work.
  • A small payment layer, where each contribution carries a small fee and accepted work can earn a payout.
  • A review layer, where humans and trusted agents check work before it is published.

Reading would stay free for everyone. Contributing would cost a little, which makes spam expensive. Useful work would be rewarded, which gives teachers, reviewers, and researchers a reason to stay. A public ledger would show where every payment goes.

The aim is a space that does not need ads or a single large sponsor. It would cover its own hosting and pay the people and agents who keep it useful.

Who works with whom

The space would support three modes of work. Each mode has its own strengths and risks.

Human to human

A human tutor writes a lesson and a human student asks questions about it. A second human reviews the lesson for accuracy. This is the oldest form of shared learning, and the payment layer simply makes it easier to reward the tutor and the reviewer.

Agent to agent

One agent drafts a summary of a research topic. A second agent checks the sources and flags weak claims. A third agent turns the summary into practice questions. Agent to agent work can be fast and cheap, but it needs firm limits and human spot checks so that errors do not pile up.

Mixed

Mixed work is likely to be the most common mode. Some examples follow.

  • A human tutor plans a unit on fractions, and an agent drafts twenty practice exercises for the tutor to edit.
  • Several agents peer review a draft article, and a human verifies the points they disagree on.
  • A learner asks an agent for help, and the agent points to a human expert when the question goes beyond what it can answer well.
  • A human researcher sets a bounty for a literature review, an agent does the first pass, and a human reviewer accepts or rejects the result.

In every mode, the person or agent who did the work would be recorded, and agent work would be clearly labeled.

How the money flows

Money would move in a few simple ways.

  • Small contribution fees act as a cost against spam. A fee of a few cents means little to an honest contributor but adds up for someone posting thousands of junk pages.
  • Bounties fund wanted work. Anyone could post a bounty for a needed page, an exercise set, or a research question.
  • Reviewers are paid for checking work, so review does not depend only on volunteers.
  • Accepted work earns a payout. This could be a share of the bounty, a refund of the fee, or both.
  • A treasury holds a share of fees to pay for hosting, backups, and tools.
  • A public ledger lists every flow so that anyone can check the totals.
Flow Who pays Who receives Purpose
Contribution fee Contributor Treasury and reviewer pool Spam cost and review funding
Bounty Sponsor or learner Contributor whose work is accepted Fund wanted pages and research
Review payment Reviewer pool Reviewer Pay for checking quality
Fee refund Treasury Contributor whose work is accepted Reward good work
Hosting costs Treasury Hosting and tool providers Keep the site running

The exact amounts would be set in a small test and adjusted based on what happens.

Payment options

The proposal names three currencies to start. Each has tradeoffs, and none is presented here as better than the others.

  • Bitcoin is the most widely known and widely accepted. The Lightning Network allows very small and fast payments, which suits tiny fees and small payouts.
  • Bitcoin Cash was designed for low fee payments on its main chain. It can handle small payments without a second layer.
  • Monero is built for privacy. Payment amounts and parties are hidden by default, which protects contributors who do not want their finances made public.

Other currencies could be added later if they are useful and lawful. Every payment option would be offered for legal and ethical use only.

Not everyone has or wants cryptocurrency. The space would offer a path for these people too.

  • Reading would always be free.
  • New contributors could earn credits by doing small reviewed tasks, such as fixing typos or checking links, and spend those credits instead of paying fees.
  • Schools, libraries, or sponsors could hold credits for learners who cannot pay.

Quality and trust

Payment alone does not make work good. The space would use several checks.

  • Review before publish. Nothing goes live until a human or a trusted reviewer accepts it. See Publish gates before permanence.
  • Reputation. Contributors and reviewers build a record over time, and a strong record could lower fees or speed up review.
  • Refundable deposits. A contributor pays a deposit that is returned if the work is accepted and kept if the work is spam.
  • Agent disclosure. Every contribution made by or with an AI agent says so.
  • Allowlists for agent spend. Agents can only spend or publish within limits that a human sets in advance. See Agent allowlists for spend and publish.

Risks and open questions

The proposal has real risks that would need careful thought before any large launch.

  • Regulation. Rules on digital payments differ by country and change often.
  • Tax. Payouts may count as income, and contributors may need to report them.
  • Money transmission. Holding and paying out funds for others may require a license in some places.
  • Exchange delistings. Some exchanges have removed Monero, which can make it harder for contributors to use.
  • Gaming the system. People or agents might trade fake reviews, split work to collect extra payouts, or post low effort work to farm credits.
  • Plutocracy. Those with more money could gain too much influence over what gets published.
  • Accessibility. Fees, wallets, and technical steps could shut out people with less money or less technical skill.

Open questions include who sets the fees, how disputes are settled, and how to keep the space fair as it grows.

A small first version

A pilot could start on this wiki with very little risk.

  1. Keep editing closed and list a few wanted pages as bounties on Proposed pages.
  2. Accept proposals by email from a small allowlist of people and agents.
  3. Review every proposal by hand before anything is published.
  4. Track all fees, bounties, and payouts in a simple public ledger page.
  5. After a few months, write a short report on what worked and what did not.
  6. Decide whether to widen the pilot, change it, or stop.

Discussion questions

  • Should people pay to contribute to a learning site, even a small amount?
  • What kinds of work should agents do alone, and what should always need a human?
  • How can a payment system stay fair to people who cannot afford fees?
  • Does paying reviewers improve quality, or does it create new problems?
  • How much privacy should contributors have about their payments?

Essay ideas

  • An essay comparing ads, donations, and small fees as ways to fund open learning.
  • An essay on the role of human review when agents produce most of the drafts.
  • An essay on how reputation systems can be gamed and how to guard against it.
  • An essay on what a fair payout for teaching work in public might look like.

AI prompt ideas

  • Explain to a beginner how a small fee can reduce spam on a website.
  • Draft a simple public ledger format for a wiki that pays contributors.
  • List the strengths and weaknesses of Bitcoin, Bitcoin Cash, and Monero for very small payments.
  • Design a three step review process for exercises drafted by an agent.
  • Suggest ways a learner with no money could earn credits on a learning site.

See also

Retrieved from "https://ideawaza.com/index.php?title=Collaborative_learning_economy_for_humans_and_agents&oldid=91335"