EIP-14 extends Ethos's existing review fraud detection to catch coordinated review rings that route endorsements through intermediaries — making it significantly harder to artificially inflate reputation scores.
What's changing
Ethos already detects and down-weights direct review-for-review swaps (A reviews B, B reviews A). EIP-14 extends this detection to longer coordination patterns — triangles and rectangles — where users cycle reviews through other accounts to avoid detection.
Pattern | Path | Description |
Direct swap (already detected) |
| Two users reviewing each other directly |
Triangle (new) |
| Reviews routed through one intermediary |
Rectangle (new) |
| Reviews routed through two intermediaries |
Only positive reviews are eligible for cycle classification. Negative and neutral reviews are never affected — this prevents the system from discounting legitimate critical reviews.
What stays the same
The unlock model that governs how reciprocated reviews are counted is unchanged:
Your first 10 reciprocated reviews always count toward your score
Each external (non-reciprocated) review you receive unlocks one additional reciprocated review slot
Reviews beyond your unlock threshold receive zero weight
All existing scoring modifiers — ELO-based weighting and community vote sentiment — continue to apply on top of cycle detection.
Score weight rebalancing
To reflect the reduced signal quality of reviews in high-cycle environments, the maximum score contribution from reviews is reduced. The reclaimed weight moves to vouch-based signals, which require economic commitment and are harder to coordinate.
Signal | Before | After | Direction |
Review score ceiling | 540 | 400 | ↓ Decreased |
Vouched ETH ceiling | 280 | 390 | ↑ Increased |
Voucher count ceiling | 270 | 300 | ↑ Increased |
Who is affected
User type | Impact |
No cycle involvement | No change to score |
Some reciprocated reviews within unlock threshold | Minimal or no change |
Extensive reciprocated review networks | Score correction downward |
No Ethos profile | Unaffected — production behavior retained |
Scores can only decrease or stay the same relative to current production — EIP-14 does not artificially boost any scores.
Why this matters for reputation quality
Cycle detection addresses a specific failure mode: users who have built high scores primarily through coordinated review rings with a single social cluster. These scores may not reflect genuine community trust — they offer no economic security and can represent purely local reputation that hasn't been independently validated.
Extending detection to triangles and rectangles significantly raises the cost of coordinated inflation. To deliver 100 undetected positive reviews under EIP-14, a colluding group would need roughly 500 participants versus 200 under the current system — a 5x increase in coordination cost.
Rollout
EIP-14 is being deployed gradually over a 60-day rollout period. This window may be shortened if the team determines a faster rollout is appropriate. Score changes will apply progressively as the rollout advances.
Questions about your score? Reach out through the Ethos support channel. For technical detail on the scoring model, see the full EIP-14 specification.