Detecting humans in the age of machines

An internet identity and reputation protocol. TRACE reads durable, hard to fake signals across the networks people already use, fuses them into a single score, and issues a portable, verifiable passport that other applications can trust.

v1.0 · Last updated May 24, 2026

Abstract

Generative AI has made fake accounts, synthetic personas, and automated activity effectively free to produce at scale. The cost of faking a person collapsed; the cost of verifying a real one did not. Every community, market, and platform now faces the same question: who is actually real?

TRACE answers it. The protocol reads public, time-tested signals across a person's social, on-chain, and community presence, fuses them with a non-linear model into a single TRACE Score from 0 to 1000, and issues a clearance-tiered Passport that travels with the person across applications. A single signal can be bought; the correlation of many independent signals, sustained over years, cannot. TRACE turns that correlation into portable, defensible reputation.

TRACE was built for one of crypto's most expensive problems, and one felt most acutely in the NFT industry: the sybil attack. With TRACE, any project can tell whether a participant is a real, credible user before a mint, whitelist, airdrop, or allocation, and filter out the wallet farms and bot accounts that would otherwise capture it.

1. The Problem: the internet cannot tell who is real

Bots, sybil clusters, airdrop farmers, and AI-generated identities now outnumber genuine participants in many online spaces. The damage is concrete: NFT mints, token allocations, and whitelists drain to farmers, governance is captured by sock puppets, social proof is manufactured, and honest users are crowded out.

The tools meant to stop this no longer work in isolation:

• CAPTCHAs are routinely solved by automation and cheap human farms.
• KYC is privacy-hostile, excludes pseudonymous users, and proves an identity document, not authentic behavior.
• Follower counts and engagement are openly bought and sold.

None of these measure what actually matters: a durable, consistent, multi-network footprint that is expensive to fabricate. That is the gap TRACE fills.

2. The TRACE Thesis

Real people leave a wide footprint that is costly to fake at scale. They hold aged wallets with real economic history, post with human irregularity, collect and trade across years, and participate in communities over time. Any one of these can be simulated. The joint distribution of all of them, weighted by how hard each is to forge and discounted for bot-like patterns, is a strong proxy for authenticity.

Three principles follow:

• Breadth over depth. Many correlated signals beat any single one.
• Pseudonymous by default. Reputation should not require surrendering legal identity.
• Owned and portable. A person's standing should follow them across applications, not be locked inside one platform.

3. Protocol Overview

The TRACE flow is simple from the outside and rigorous underneath:

• A person signs in with X (their primary identity) and optionally links wallets, Discord, and Telegram.
• TRACE reads public signals from each connected network.
• A server-side model fuses them into the TRACE Score (0 to 1000).
• The score maps to a clearance tier and a rendered Passport.
• Holders verify at the Passport Office; access codes gate entry; invitations grow the network with real people.

4. The TRACE Score

The score is the heart of the protocol. It is computed entirely server-side; the exact weights, curve constants, and thresholds are never sent to the client, which keeps the model resistant to reverse-engineering and gaming.

4.1 Signal categories

Several categories of signal contribute to the score:

• On-chain wallet and NFT activity. NFT participation is the dominant on-chain signal: real holdings, bluechip exposure, collection breadth, and genuine trading on NFT marketplaces. Long-term wallet age and anti-sybil checks support it. Raw ETH balance, generic transaction count, token diversity, and DeFi activity are deliberately minor inputs, because they are trivially inflated by self-transfers and airdrop farming.
• X reputation. Account age, posting consistency, and engagement quality, with raw follower count treated as a minor input.
• Community. Discord and Telegram tenure and participation quality.
• Identity consistency. A unified, cross-linked identity across real networks.
• Activity quality. Recent, genuine activity across the present sources.

The relative weighting of these categories, the curve constants, and all thresholds are kept private so the model stays resistant to gaming.

4.2 From signals to a score

Every connected source produces a set of normalized metrics, each a value between 0 and 1: presence, age, activity, quality, reach, and authenticity. From these, TRACE derives a reputation value cᵢ for each category, also between 0 and 1. The categories are then fused into a single raw value:

where wᵢ are the category weights (kept private) that sum to 1. Absent categories contribute 0 and are not renormalized, so a high raw value requires genuine breadth across networks rather than one maxed-out signal. Authenticity penalties and a small set of hidden bonuses (for example, durable on-chain history) then adjust raw before it is mapped to the final score.

1. 1
   Signals

   Each connected source yields normalized metrics (presence, age, activity, quality, reach, authenticity), every value from 0 to 1.
2. 2
   Category reputations

   Those metrics roll up into five category scores cᵢ, each from 0 to 1.
3. 3
   Weighted fusion

   Categories combine into a single raw value. Weights are private, and breadth is required: absent categories count as 0.
4. 4
   Integrity adjustment

   Authenticity penalties and a few hidden bonuses adjust the raw value.
5. 5
   Curve and decay

   A concave curve and an inactivity factor map raw onto the final 0 to 1000 range.
6. 6
   Result

   A TRACE Score, mapped to a clearance tier and a verifiable Passport.

4.3 The non-linear curve

The adjusted raw value, between 0 and 1, is mapped onto the 0 to 1000 range through a concave power curve and an inactivity factor:

with k a fixed exponent between 0 and 1 (kept private) that makes each additional point harder to earn than the last, and decay a time factor that gently reduces a score after a period of inactivity. The curve below shows the effect: the score rises quickly at first, then flattens, so the top tiers stay rare by design.

4.4 Authenticity and anti-sybil

Every source also produces an authenticity read that penalizes bot-like cadence, templated or duplicated posting, airdrop spam, sybil minting, and dormant or empty wallets. Crucially, breadth is required: absent categories contribute zero and are not renormalized, so a high score demands a real, multi-network footprint rather than one maxed-out signal.

4.5 Decay

Reputation reflects the present. After a grace period of inactivity, a score decays toward a floor. Standing must be maintained, not earned once and abandoned.

5. Identity Sources

X (primary identity)

TRACE reads account age, posting consistency and cadence, and engagement quality, and runs a behavioral classifier over recent public posts that labels the account along a spectrum from human to inactive, spam, automated, or farming. Reach is treated as a minor input, because followers can be bought but consistent human behavior over time cannot.

Wallet and NFT (the backbone)

On-chain history is read through public providers, but not all on-chain activity counts the same. NFT participation carries the wallet category: NFT holdings (with bluechip exposure and collection breadth) and genuine NFT marketplace trading are the dominant levers. Real collectors hold real collections and trade on real venues, and that pattern is costly and slow to fabricate. Long-term wallet age and anti-sybil signals support it.

Inputs that are easy to inflate are deliberately minor: raw ETH balance can be moved in by a single transfer; transaction count can be padded with self-sends; token diversity is a trivial outcome of airdrop farming; DeFi activity does not by itself prove a person. A wallet that never actually holds or trades NFTs cannot reach the upper tiers on the on-chain category alone. Conversely, a pure flipper with high trade volume but almost no holdings is treated as flipping, not collecting: real participation requires real retention, so holdings gate how much credit trading volume can earn. Linking a wallet uses a free signature to prove ownership and never moves funds.

Discord and Telegram

Community tenure and participation quality contribute a smaller share, and spam is never rewarded.

6. The Passport and Clearance Tiers

The TRACE Passport is an official, shareable identity document rendered for each holder, with a machine-readable zone, official stamps, and an authority signature. It is the human-facing expression of the score. Five clearance tiers band the range:

• Civilian · entry standing.
• Prime · an established, credible footprint.
• Origin · strong, multi-network reputation.
• Elite · exceptional standing, auto-verified, carries access codes.
• Sovereign · the rarest tier, reserved for the most authentic, highest-reputation identities.

Elite and Sovereign holders are the eligible mint cohort for the TRACE NFT collection. Holders below those tiers can elevate by linking additional networks, completing verification at the Passport Office, or redeeming an access code.

7. Verification and Access Codes

Verification is human-curated at the Passport Office. Holders submit their passport for officer review; approval applies a signature and stamps. Elite and Sovereign tiers are auto-verified by the system.

Access codes let verification spread through trusted hands rather than open the gates to everyone at once. Elite and Sovereign holders receive an initial batch on sign-up, and further codes are distributed by operators over time. A code is one-shot: redeemed once when its recipient submits at the Office. To keep access aligned with current standing, if a holder's reputation falls below the issuing tier, their unredeemed codes are revoked.

8. Network Growth: Invitations

Anyone can earn access codes by bringing in real members. An invitation only counts when the invited account is plausibly human:

• It signs up with a real X account.
• It has a profile picture (an empty avatar is treated as a likely bot).
• It meets a minimum follower threshold.
• It is unique: duplicate sign-ups from the same network address do not count.

Every set number of valid invitations earns one code. Growth is therefore tied to authenticity, not raw volume.

9. For Projects: Sybil-Resistant Allocation

TRACE was built for a specific, costly problem in crypto, and one felt most sharply in the NFT industry: the sybil attack. When a mint, whitelist, airdrop, or allocation goes live, a single operator can spin up hundreds or thousands of fresh wallets and fake accounts to capture a disproportionate share. Real collectors and genuine community members lose out, and the distribution is corrupted before it begins.

TRACE gives any project a way to tell real participants from farmed ones. Because every identity is scored on a wide footprint that is hard to fake (real on-chain history, genuine NFT collecting and trading, and an aged social presence), a project can:

• Gate a mint or whitelist to a minimum TRACE tier, filtering out fresh sybil wallets and bot accounts.
• Export a tier-filtered snapshot of real, connected wallets (with tier and score) for allocations and airdrops. X handles are kept private and are not included in the export.
• Look up any participant's public passport to see, at a glance, whether they are a credible, real user.
• Integrate directly: the TRACE API lets applications check a tier or score live and gate access automatically, with no manual review (see Section 10).

The result is distribution that reaches humans who are genuinely part of the ecosystem, not airdrop farmers running wallet factories. TRACE turns the question "are these real users?" from a guess into a verifiable, on-record answer.

Sybil-resistant allocation is the first product. Over time the identity layer is intended to anchor a broader toolkit of Web3 and NFT utilities that projects, collectors, and traders use day to day. The Roadmap describes the direction in more detail.

10. The TRACE API

The TRACE API exposes the reputation layer as infrastructure. A single endpoint accepts a wallet address or an X handle and returns whether that identity meets an optional minimum tier and an optional minimum score. The response is live: each call reads the same record that the Top Passports leaderboard and the Passport Office both read.

Projects use it to gate mint allocations, Discord roles, application sign-ups, and any other access decision that should distinguish humans from bots. Each key carries a configurable monthly request quota and can be rotated or revoked without code changes on the calling side. The full specification, request and response shape, authentication model, and integration examples are published at traceeco.io/api-docs.

The API is intentionally minimal. It does not stream private fields, does not return wallet history, and does not expose anything that is not already public on the holder's passport. The asymmetric value comes from the network it is reading, not from the size of the response.

11. TRACE NFT: The On-Chain Collection

TRACE NFT is the cultural layer of the protocol: original pixel identities, generated and classified across the five clearance tiers, each on record. The artwork is original and tier-themed, with rare traits including the TRACE sigil. It is a collectible expression of the network classifying its own community. Final supply will be announced ahead of mint.

12. Privacy and Data Sovereignty

TRACE is pseudonymous by default. It never asks for passwords, seed phrases, or private keys; wallet linking is a free off-chain signature. Wallet addresses are never displayed on public pages. People connect only the networks they choose and can disconnect at any time. Full detail is in the Privacy Policy.

13. Security Model

TRACE is built with security and abuse-resistance as first principles. Authentication follows industry-standard practices, wallet links are verified by cryptographic signature, and the reputation model runs entirely server-side, so its internals are never exposed to the browser. Access is granted through single-use codes, and the protocol applies anti-sybil protections across sign-ups, invitations, and linked wallets. We deliberately do not publish implementation details that would help an attacker.

14. Roadmap

Phase 1 (live)

X login; wallet, NFT, and community signals; the TRACE Score and Passport; the Passport Office; access codes and invitations; the TRACE NFT collection; public passports and the Top Passports leaderboard; the partner-facing TRACE API for live eligibility checks; and tier-filtered allocation exports.

Phase 2

Broader on-chain coverage across more chains and protocols and richer behavioral models. The partner-facing reputation API moved into Phase 1 and is live now (see Section 10).

Phase 2 also extends the identity layer into an expanding suite of Web3 and NFT tools built on top of TRACE reputation. The direction is to give projects, collectors, and traders the practical utilities they reach for every day, including:

• Sybil-resistant mint and whitelist tooling with one-click TRACE-tier gating.
• Allocation snapshots and airdrop list builders that drop the wallet farms before they reach the queue.
• Holder and trader analytics that combine on-chain history with TRACE reputation signals.
• Project dashboards that turn the community's TRACE footprint into actionable distribution decisions.
• Embeddable widgets so any site can verify a holder's TRACE tier in place.
• Additional collector and trader utilities released as the network grows.

The intent is that TRACE becomes the identity-aware base layer for the NFT ecosystem, with the toolkit growing alongside it. Specific products and release order remain subject to change.

Phase 3

Progressive decentralization of verification and full portability of the passport across ecosystems.

15. Closing

As machines learn to imitate people perfectly, proof of humanity becomes the scarcest thing online. TRACE turns the footprint that real people already leave into reputation that is portable, defensible, and theirs. Detecting humans in the age of machines.

This document is informational. It describes the protocol as designed and may evolve. It is not investment advice and is not an offer or solicitation of any security or token.