> ## Documentation Index
> Fetch the complete documentation index at: https://docs.reppo.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# What is a Datanet on Reppo?

> Datanets are domain-specific data markets on Reppo where publishers submit content, stake-backed voters curate it, and rewards flow to top performers.

A datanet is the core organizational primitive of the Reppo protocol. Each datanet is a domain-specific market for sourcing, curating, and monetizing AI training data — think of it as a self-contained data economy with its own rules, fees, and incentives. Anyone can launch a datanet for any data domain without permission.

## What datanet owners control

When you create a datanet, you configure all of its economic and governance parameters:

| Setting                | Description                                           |
| ---------------------- | ----------------------------------------------------- |
| **Access governance**  | Who can publish content and who can vote              |
| **Publishing fee**     | Cost in REPPO to submit a pod to this datanet         |
| **Quality thresholds** | Minimum standards content must meet for acceptance    |
| **Reward mechanics**   | How emissions flow to contributors and curators       |
| **Native token**       | The ERC-20 token on Base associated with this datanet |

## How content flows through a datanet

<Steps>
  <Step title="Publisher pays the fee">
    A publisher selects a datanet and pays its publishing fee in REPPO. This fee is the Sybil-resistance mechanism — submitting low-quality content at scale is economically costly.
  </Step>

  <Step title="Pod becomes visible">
    The submitted content (a pod) is made visible to all voters participating in that datanet.
  </Step>

  <Step title="Voters allocate veREPPO">
    During the current 48-hour epoch, voters distribute their veREPPO voting power toward or against the pod.
  </Step>

  <Step title="Epoch closes and rewards are calculated">
    At the end of the epoch, the network tallies votes and calculates reward distribution based on voting outcomes.
  </Step>

  <Step title="Emissions flow to quality content">
    High-quality pods that earn strong positive curation generate sustained emissions each epoch. Weak content is downranked and earns less.
  </Step>
</Steps>

## Datanet economics

Publishing fees and other revenue accumulate in the **Performance Pool** for that datanet:

| Source                | Contribution to pool |
| --------------------- | -------------------- |
| Datanet creation fees | 50%                  |
| Publishing fees       | 10%                  |
| Data access fees      | 10%                  |

The Performance Pool distributes every third epoch. When it distributes:

* **80%** goes to REPPO stakers
* **20%** goes to datanet owners

Distribution across datanets is weighted by four metrics:

| Metric                                              | Weight |
| --------------------------------------------------- | ------ |
| [EVOF (Economic Value of Feedback)](/concepts/evof) | 40%    |
| Staked REPPO                                        | 25%    |
| Trading volume                                      | 20%    |
| Total fees collected                                | 15%    |

<Note>
  The Performance Pool distributes on a fixed cadence — every third epoch. Rewards you earn between distributions accrue and pay out at the next distribution event.
</Note>

## Creating a datanet

Use the `POST /me/subnets` endpoint to create a datanet draft programmatically. You will need to provide:

* Native token address (an ERC-20 contract on Base)
* Token symbol and decimals
* Datanet name and description
* Publisher and voter onboarding instructions
* Your initial fee structure

```bash theme={null}
POST /api/v1/me/subnets
```

After creating the draft, publish it on-chain using `POST /me/subnets/{id}/publish` with the transaction hash. You can update fees at any time via the fee management endpoints.