Skip to main content

Design Principles: Atomic Granularity

The Power of Flat Data

A crucial best practice is keeping information modular and atomic. You are economically incentivized to create "flatter" Atoms – each representing a single, minimal concept – rather than packing composite information into one Atom.

Why Granularity Matters​

Consider representing the statement: "Tiger Research was founded in 2021"

// One Atom containing all information
const statementAtom = {
  data: "Tiger Research was founded in 2021"
}

Problems with this approach:

  • Hard to verify which part might be incorrect if disputed
  • Cannot reuse individual components
  • Difficult to update or correct specific elements
  • Less composable with other data

Atomic Approach (Best Practice)​

// Three separate Atoms
const tigerResearchAtom = { data: "Tiger Research" }
const foundedInAtom = { data: "founded in" }
const year2021Atom = { data: "2021" }

// Connected via a Triple
const foundingTriple = {
  subject: tigerResearchAtom,
  predicate: foundedInAtom,
  object: year2021Atom
}

Benefits of this approach:

  • Each piece can be independently verified and traced
  • Individual components are reusable (the [2021] Atom can be used in countless other triples)
  • Corrections can be made to specific elements without discarding everything
  • Trust accrued by each Atom benefits all its usages

Economic Incentives for Granularity​

Intuition's economic model naturally guides users toward optimal granularity:

  • Reusable Atoms accumulate more Signal as they're referenced in multiple Triples
  • High-Signal Atoms attract more $TRUST stakes, increasing their network gravity
  • Composite Atoms with embedded data are less likely to be reused or staked
  • Modular Design maximizes potential for community adoption and value accrual

Flexibility and Scope​

Atoms are not limited to static concepts. They can represent:

  • Dynamic entities: User-generated content, evolving documents
  • Abstract concepts: Tags, categories, classifications
  • Data structures: JSON schemas, configuration templates
  • Content chunks: Text snippets, code fragments (via IPFS CID)

Regardless of what an Atom represents, the principles remain consistent:

  1. Give it a clear, unique identity
  2. Include relevant data in the atomData field
  3. Keep its scope narrow for maximum verifiability and reusability