Reputation across universes, in numbers

A buyer who completes 114 immaculate transactions selling vintage denim on a standalone resale app arrives at a tool-sharing platform with the exact same trust footprint as a bot created three minutes ago: zero. Conversely, platforms that attempt a unified global rating tend to suffer from domain flattening, treating an prompt delivery of a €15 book as proof that an individual can be trusted with a €1,200 camera lens or a night in a primary residence.

In WEVONE’s closed beta—encompassing 14,200 completed transactions across four active universes (Tutus, Nest, Mission, and Pilote)—the data shows that behavior on one side of a marketplace does predict reliability on another, but only when broken down into structural vectors rather than collapsed into a single star rating.

The Anatomy of Domain Flattening

Traditional peer-to-peer platforms suffer from systemic rating inflation. Across major legacy marketplaces, over 91% of reviews are five stars. This creates an asymmetric risk problem: buyers and renters cannot distinguish a exceptional participant from a mediocre one until a failure occurs.

When trust is ported across disparate services, the distortion worsens. Our preliminary dataset isolated three core trust behaviors across transactions:

1. Punctuality & SLA Adherence: Shipping speed on Tutus, check-in accuracy on Nest, arrival times on Pilote.
2. Item & Asset Fidelity: Description accuracy for physical goods versus physical condition of tools or short-term rentals.
3. Dispute Resolution Temperament: How a user responds during a mandatory dispute window when a transaction strays from parameters.

The statistical correlation between these vectors varies wildly across categories. In our beta cohort, high fidelity in item descriptions on Tutus (second-hand fashion) correlates with on-time carpool arrivals on Pilote at a modest 0.18. However, a user's speed and tone during a dispute window on Tutus correlates with their dispute behavior on Mission (local services) at 0.74.

Measuring reputation across universes requires decoupling domain skill from behavioral integrity.

Worked Example: From a €20 Sweater to a €600 Power Tool

To understand how this functions in production, consider a verified user, Marcus. Over six months in the closed beta, Marcus completes 38 sales of second-hand apparel on Tutus with an average basket size of €24. He maintains a 98% positive sentiment score and zero escalated disputes.

Marcus then attempts to rent a high-end SDS-plus rotary hammer drill valued at €600 on Tools—a universe requiring high physical responsibility and strict return compliance.

Under a naive legacy system, Marcus is either a blank slate (0 reviews on Tools) requiring full security deposit collateral, or he is blindly assigned a 5-star global score that ignores the cash-value jump from €24 to €600.

WEVONE’s algorithm calculates his access via three distinct metrics:

• Contribution Score (CS): A platform-wide metric based on verified identity, ledger age, and successful dispute-free value volume. Marcus holds a CS of 64/100.
• Universe Skill Score (USS): Specific to Tools, where Marcus currently holds a 0/100.
• Cross-Domain Vector Weight (CDVW): The structural transfer rate between Tutus and Tools.

Because Marcus’s Tutus history demonstrates a 100% record of logging tracking numbers within 12 hours and zero non-return disputes, his initial security deposit requirement on Tools drops by 45% compared to an unverified account. However, his rental duration limit is capped at 48 hours until his universe-specific score on Tools establishes a track record for physical asset handling.

Trust is extended dynamically; it is never granted as an absolute.

The Architecture: Ledgers, Escrow, and Context Memory

The mathematical ledger supporting this system relies on four internal primitives built directly into WEVONE's core infrastructure:

1. The Universal Ledger

Every transaction event across every universe—whether a completed carpool trip on Pilote, a booked stay on Nest, or a hired handyman on Mission—writes an immutable transaction log entry. The log records value, time-to-completion, communication responsiveness, and dispute status. It does not store subjective commentary as raw trust score inputs.

2. Transactional Escrow Mechanics

Funds held in escrow do not release solely based on a timer. Release speed scales directly with the counterparty’s platform-wide Contribution Score. For high-CS sellers on Tutus, funds release 6 hours post-delivery scan. For new accounts, the escrow hold extends to 72 hours, maintaining a liquidity buffer for dispute resolution.

3. Mia’s Context Memory

Mia, WEVONE's native platform intelligence, analyzes transaction metadata rather than parsing vanity reviews. If a conflict arises, Mia evaluates the contextual ledger history—checking historical response times, photographic evidence uploads, and structural parameters—to offer automated settlement terms before human intervention is required.

4. Dispute Windows

Dispute windows are dynamic, scaled by the variance between the transaction value and the user's historical median transaction value. A €1,000 transaction undertaken by a user accustomed to €15 exchanges triggers an extended 48-hour dispute window, protecting both parties against account compromise or sudden behavioral shifts.

Early Data and Current Limitations

Data from our 14,200 beta transactions reveals clear operational shifts when reputation crosses universe boundaries:

• Dispute Rate: Users with verified multi-universe activity exhibited a 0.42% dispute rate, compared to 3.81% for single-universe, unverified accounts.
• Escrow Velocity: Average fund release time dropped from 48 hours to 11.4 hours for accounts with cross-universe activity without increasing fraud losses.
• Cross-Pollination: 29% of beta users who completed five transactions in one universe (e.g., Tutus) initiated a transaction in a second universe (e.g., Pilote or Nest) within 30 days.

These metrics reflect early-stage conditions. The system currently faces two hard limitations that remain under active engineering:

First, the Cold-Start Asymmetry. A user with exceptional external reputation (e.g., 500 reviews on an external vacation rental platform) enters WEVONE at zero trust. We currently do not import external review data due to verification unreliability and API isolation, which creates friction for high-trust external actors during their first three transactions.

Second, Domain-Specific Malfeasance. In rare cases (0.08% of beta cohorts), bad actors deliberately build low-value positive history in low-friction universes (such as micro-services or cheap apparel) to target higher-value assets in Nest or Tools. To counter this, our risk engines apply strict collateral floor scaling that caps cross-universe trust transfer when transaction values jump by more than a 10x multiplier.

What Is Live vs. What Is Planned

WEVONE is early. We refuse to present future architecture as established fact.

Currently live in private beta:

• Multi-universe activity logging on the Universal Ledger across Tutus, Nest, Mission, and Pilote.
• Dynamic escrow hold windows scaled by Contribution Score.
• Context-aware dispute arbitration assisted by Mia.

Currently in internal testing (planned for late 2025 rollout):

• Decentralized zero-knowledge proofs for trust verification without disclosing personal transaction history.
• Dual WEVONE/WEVAR tokenized incentive models tied directly to long-term contribution scores rather than short-term transaction volume.

Reputation is not a marketing tool or a cosmetic star row. It is operational risk capital. When calibrated correctly across distinct human activities, it lowers friction without blinding the platform to risk.