There's a wallet out there with five years of spotless history: dozens of loans on Aave and Compound, every one closed on time, not a single liquidation, a steady balance, votes cast in three DAOs. All of it sits in the open on a public ledger — anyone can check it, anyone can verify it. A bank will never see any of it. To a bank, that person is a blank page, because they have no Social Security number, no credit card, no three years of filings with Equifax. That's the paradox a small industry has grown up around: the most detailed financial history a person can have might exist in full public view — and be completely useless to the system that decides whether to lend to them.
That's the promise of on-chain reputation: stop asking "who are you on paper" and start asking "what did you do when nobody was forcing you to." It's a good promise. The reality is a much rougher construction site than the marketing around it suggests.
What's broken about the old score
The classic credit score — FICO in the US, its equivalents elsewhere — was built in 1989 as a way to compress a person's financial behavior into a single three-digit number, 300 to 850. The idea is reasonable: a lender needs a risk signal. The problem is who holds the formula, and who never makes it into the formula at all.
The formula is held by a handful of private bureaus — in the US, effectively three companies: Equifax, Experian, and TransUnion — and the algorithm inside is opaque. You don't get told why your score dropped twenty points, and disputing the weight of any one factor is close to impossible. In 2017, Equifax lost the personal data of roughly 147 million people in a breach — Social Security numbers, birth dates, addresses — and almost nothing changed. The system was too embedded in the economy to actually be punished.
And the people who never enter the formula at all aren't a fringe case. By various estimates, tens of millions of adults in the US alone remain "credit invisible" — not enough history, no card, no mortgage — and the bank turns them down not because they're unreliable, but because there's nothing for the formula to read. A migrant with a spotless payment record back home arrives in a new country and starts at zero, as if they were twenty years old and had never paid a bill in their life.
What's already legible on-chain
The difference is that an on-chain history is never a blank page, and it's never bound to one jurisdiction. Every loan taken and repaid on Aave, Compound, or MakerDAO, every liquidation, every stake, every DAO vote — it's a record that can't be erased, backdated, or hidden behind a bureau's closed door. A wallet that survived the 2022 market crash without a single liquidation proved something real about its holder's discipline — and proved it publicly, with no intermediary deciding who gets to see it.
That's the space a handful of projects have started building in: read the data straight from the chain instead of locking it in a closed bureau. Spectral Finance built the "MACRO Score," a non-transferable NFT tied to a wallet that aggregates lending history and reputation into a single number. ARCx shipped an open DeFi Score back in 2020 — an attempt to rate wallet risk from public lending-protocol data. Cred Protocol and a few similar projects try to do the same thing at finer grain, breaking reputation into factors: wallet age, protocol diversity, repayment history, behavior during drawdowns.
A separate and more ambitious line of work is undercollateralized lending. A normal DeFi loan on Aave requires collateral worth more than the loan itself — meaning the credit is, structurally, useless to anyone who doesn't already have capital. Goldfinch and Maple Finance tried to break that pattern: pools that lend without full collateral, with risk assessed by "backers" — participants who put their own capital first in line to absorb losses, signaling to the market that a given borrower can be trusted. The goal was to get DeFi capital flowing to real businesses outside of crypto speculation — fintechs in emerging markets, trading firms, small operators.
Why it hasn't really taken off yet
This is the part that needs honesty rather than a brochure.
The first problem is Sybils: nothing stops someone with a bad history from simply opening a fresh wallet and starting clean. Unlike FICO, where your name and Social Security number keep you tethered to your past, a pseudonymous wallet is a costume you can take off and put back on. Without a tie to a durable identity (see proof of personhood), reputation is a reputation of the address, not the person — and the address is disposable.
The second is what happens when the borrower doesn't pay. A traditional bank has courts, collections agencies, a bad mark that follows you for a decade. An on-chain lender facing an anonymous wallet has almost nothing but public shaming on social media. That's exactly why undercollateralized DeFi lending has stumbled more than once on real defaults: in 2022, amid the collapse of major crypto counterparties, borrowers in Maple Finance's pools defaulted on tens of millions of dollars — money that backers and depositors actually lost. The model only really works when a pseudonymous address is backstopped by an institution with real legal liability behind it — at which point it's not quite "trust without a bank." It's a bank with an extra step.
The third is that a rich wallet looks trustworthy almost by default. If reputation is scored on transaction volume and history, a whale who already holds ten million dollars gets a high score automatically — while the person who needs credit most simply doesn't have enough history to build a score from in the first place. It's the same old trap in new packaging: the system rewards the people who need it least.
Our Record. There's something worth seeing past the technical scoring debate here. FICO doesn't really decide "what you did" — it decides "who's allowed to know about it." The formula is closed, the appeal is nearly impossible, and three companies own the right to decide millions of people's financial fate. That's Isfet's signature move: rent extracted from someone else's visibility, a monopoly on judgment with no accountability to the people being judged. Ma'at's Scales work differently — they weigh the deed, Ib against the feather, and the weighing has to be visible to everyone, not sealed in a bureau's archive. On-chain reputation, done honestly, points that direction: your Ren — your name, your record of action — stays with you and travels with you between protocols, and no private company gets to confiscate it. That doesn't solve the Sybil problem or the accountability gap by itself. But it's the right thing to be aiming at.
The dilemma is real: reputation without identity is trivial to fake, and identity without privacy is a leash. The live projects in this space worth watching — Gitcoin Passport, and work in the direction of zero-knowledge identity — are exactly the ones trying to find the middle: prove enough, reveal nothing extra.
Today this whole field is still under construction — the promise outruns the product you can actually put your hands on. But the direction is the right one, and it's worth tracking not as a speculative trend but as a genuine second attempt at a question the old system answered badly and unfairly: who deserves trust — and who gets to decide.