Reputation Should Have a Half-Life

One thing that always feels off in most reputation systems:

Trust never expires.

Someone can receive strong endorsements years ago and those signals still dominate their reputation today.

But that’s not how trust works in real life.

People naturally weigh recent behavior more than old signals when judging credibility. Research on online reputation systems shows that time-aware models significantly improve trust accuracy, because user behavior and reliability change over time.

Another body of work on trust modeling also highlights that reputation systems should incorporate temporal decay, allowing outdated information to gradually lose influence as new signals appear.

In other words:

Trust is temporal.

Not permanent.


The Idea: Reputation Half-Life

Borrow a concept from physics.

Radioactive elements decay over time.

Reputation could behave the same way.

A simple model might look like:

Reputation(t) = InitialTrust × e^(-λt)

Where λ represents how fast trust decays.

Older attestations slowly lose influence unless they are reaffirmed by new signals.

This keeps the graph aligned with reality.


Exp 1: The “Zombie Reputation” Problem

Imagine a founder who built a respected protocol in 2019.

They received hundreds of positive attestations.

Fast forward to 2026.

They launch a new project that behaves suspiciously.

If old attestations never decay, the trust graph still shows:

High trust
Strong reputation
Many endorsements

But those signals are stale.

With a reputation half-life model:

  • older attestations gradually weaken

  • new behavior becomes more influential

The trust graph reflects who the person is now, not who they were years ago.


Exp 2: Early-Stage Projects

New protocols often start with very little trust signal.

But reputation decay can actually help them.

If older incumbents lose influence over time, newer projects can gain credibility faster through recent performance.

The graph becomes more dynamic and merit-driven.


Exp 3: Continuous Credibility

Instead of reputation being something you earn once, it becomes something you maintain.

Builders, researchers, and contributors would naturally need to keep reinforcing their credibility through ongoing activity.

Think of it like:

Proof of Consistent Behavior.


What I Think This Could Unlock

lets say if attestations carried temporal weight, the graph could surface new signals like:

ReputationFreshness
AttestationDecayRate
CredibilityMomentum
ReaffirmationFrequency

These signals would be extremely useful for:

  • wallets evaluating counterparties

  • AI agents deciding which data to trust

  • DeFi protocols assessing risk

  • governance systems weighting influence


My Proposed Build Directions

If reputation half-life became a primitive in the Intuition ecosystem, a few interesting build paths could emerge.

a. Temporal Trust Engines

Services that compute reputation using decay models.

Possible outputs:

CurrentTrustScore
TrustDecayRate
RecentCredibilityIndex

Apps could query these signals before interacting with an address or protocol.


b. Reputation Renewal Mechanisms

Tools that allow users to reaffirm trust signals periodically.

For example:

  • periodic attestations

  • credibility checkpoints

  • community re-validation of past endorsements

This keeps trust signals fresh.


c. Trust Momentum Dashboards

Analytics platforms tracking how reputation evolves over time.

Example insights:

  • which protocols are gaining trust fastest

  • where credibility is declining

  • when reputation suddenly spikes or collapses

These could act as early signals for ecosystem health or risk.


My Final Thought

If Intuition is mapping the internet’s trust relationships, it might also need to model how trust fades and renews over time.

Because credibility isn’t static.

It’s something that constantly evolves with behavior.


My Question

Should attestations on Intuition gradually lose influence over time unless they’re reaffirmed?

Or should reputation remain permanent once it’s recorded?

2 Likes

Very valid concern as reputation is very dynamic.

Reputation should definitely not be permanent once recorded because that could and would go wrong in so many ways, and is not an accurate representation of irl reputation

Not exactly sure about all attestations losing influence over time would work though

Also, the whitepaper kinda touches on this

  • reputation is gained or lost through behavior
  • curators who regularly create accurate and valuable claims are more trusted
  • they can lose credibility if behavior changes and become less reputable and trusted by the system
1 Like