Trust, on both sides
A record you can build, and have to live with.
Renter rates owner. Owner rates renter. Each rating is signed by the wallet that signed the booking, written to a public chain, and bound to the account that built it. The protocol is patient about old mistakes and sensitive to recent ones — which is what the word "reputation" has always meant, before software.
The principle.
Most rental platforms ask one party to be trusted and the other to be vetted. The customer rates the seller; the seller, at best, gets to flag the customer to the support queue. The asymmetry is quiet but load-bearing — it determines who the platform's customer really is, and who the platform's problem really is.
DaiZen will not be that shape. The bike's owner is not the seller. The renter is not the customer. They are two neighbours who have agreed on a price, and the only fair way to write down what happened afterwards is for both of them to say so. So both of them do. Two ratings per rental, in the same record, neither contingent on the other being kind.
How the rating works.
When the rental settles — the bike returned, the deposit refunded, the fee delivered — a thirty-day window opens for each party to submit a rating. Five stars, one through five, and an optional short note. The rating is signed by the wallet that signed the original booking; the wallet's signature is what makes it count. There is no anonymous review and no review without a rental that actually happened.
A rating, once submitted, is final. It cannot be edited and it cannot be withdrawn. This is the small but important guard: a counterparty who reads your rating and dislikes it cannot reach back and rewrite their own in response. The record either stands on the day it was written, or it does not exist.
Two ratings, both signed, both for keeps.
What "soulbound" means in plain language.
The record is bound to the wallet that built it. Technically: the token that represents it cannot be transferred or sold; every attempt reverts. Practically: an account with a long careful history cannot lend that history to a new account, and a new account cannot inherit one. Each wallet is the witness of its own behaviour, and only its own.
The honest tradeoff is the other half of the same sentence. If you cannot transfer a good record, you also cannot escape a bad one by spinning up a new wallet — or rather, you can, but you start from zero, post the full deposit, and earn your way back. The asymmetry is the point. The record that vouches for you is the same record that holds you to it.
Why recent rentals weigh more.
A score that averages every rental you have ever done would give equal weight to the chain that came off in 2025 and the careful run of returns in 2027. That is not what trust looks like in practice. Trust is mostly about the recent past, partly about the slightly less recent past, and almost not at all about the distant past.
So the protocol weights the same way. A rating from last month counts the most; a rating from a year ago counts less; a rating from three years ago is a small echo. Concretely, the weight halves roughly every six months. A reckless month in your early twenties does not follow you into your thirties unless you keep repeating it. A careful run last summer means more than a careful run in 2024.
Reputation is local.
A single number that travels across every kind of rental is not reputation. It is a credit score. Credit scores compress a person into a digit so that strangers in distant offices can decide quickly. We are doing the opposite.
The protocol carries a small, neutral record of how each wallet has behaved as a renter and how it has behaved as an owner. What that record means — in this category, with this asset, for this kind of rental — is the work of the guild that hears the case. A strong record as a bike-renter is a starting position with the bike guild; the cello guild forms its own view. The numbers are a hint, never a verdict. The verdict belongs to the people who know the thing.
A hint, never a verdict.
The economic feedback.
A deposit is a hedge against the risk that something goes wrong. When there is no history to read, the hedge has to cover the full plausible cost — a first-time renter posts the full base deposit, plus light identity verification. When there is history, the hedge can be smaller in proportion to what the history says.
The protocol moves in three steps. A renter with a few good rentals posts somewhere around three-quarters of the base. A renter with a longer, cleaner record posts around half. A renter with a sustained record across many rentals posts around a quarter. These are not loyalty tiers. They are not a reward. They are the math of risk, run forward: the better the lender knows you, the less collateral the rental needs.
What this is not.
It is not a five-star feed for the public to brigade. The only accounts that can rate yours are the ones that completed a rental with you. It is not anonymous. It is not editable. It is not a credit score — no third party scores you on assumptions about your postcode or your bank account; only neighbours you traded with do.
And it is not the company's record of you. We do not keep a private file. The record is on the chain, in your wallet, the same shape for you as for the owner two streets away and the renter who borrowed her drill last month. If DaiZen disappears tomorrow, the record stays.
The substrate.
A sharing economy that works needs three things at the floor: a deposit that both parties can see, a guild that can decide the rare hard case, and a record that follows everyone, in both directions, in proportion to what they have actually done. The first two are written in other essays on this site. This one is the third. The contract holds, the guild decides, and the record — on both sides — is what makes tomorrow's rental cheaper than today's.