My Honest Review of Thonest (Updated 2026)

We spent an afternoon on the site and wrote this review from scratch.

What Thonest is

Thonest calls itself "the public record for predictions." The idea is simple: a public call is only worth something if it can be checked later. So you lock a prediction before the outcome is known - the exact words, the original source, the date, and the specific test that decides whether it came true - and the record sits there, unchangeable, until it settles.

You do not need an account to read anything, browse the feed, look at a forecaster's history, or verify a record yourself. That is a deliberate choice and it tells you what the product is really for.

How it works

Three steps, and the whole thing took us a few minutes.

1. Put it on the record. You can log your own prediction, or register one someone else made in public. That second path is the interesting one. Paste a post from X and Thonest fetches it from the platform and confirms the account. Paste a podcast or a YouTube video and it transcribes the source and confirms the words at the moment they were said, by the person who said them. Write it directly on Thonest and the timestamp itself is the proof.

2. Lock the test. This is the part most "prediction" sites skip. Thonest works out what would actually settle the statement - the condition, the deadline, and which source gets consulted - and shows it to you in plain English before anything is locked. You confirm it. After that, the test is fixed. On the record we looked at, the locked test read like a contract: which exams, graded by whom, on what scale, by what date.

3. See what happens. A record settles the moment its locked condition becomes conclusive. Some settle automatically from a data source with no human involved. Some settle on evidence, where a named reviewer applies the locked criteria in public and publishes the reasoning. And some are marked "preserved - not scored," which is our favourite detail: if a statement carries no objective test ("we will make things stronger"), the system refuses to invent one. It is kept with its source and never scored.

Outcomes are worded carefully too. Records read Open, Condition met, Condition not met, Could not be determined, or Unresolvable. There is no "liar" label and no honesty score attached to a person. The claim is about the evidence, not the character.

Why this beats a screenshot

Anyone can fake a screenshot, and the post behind it can be quietly edited or deleted. Every Thonest record carries four layers instead, each one checkable by somebody who assumes the site is lying.

• The words are checked against the source. Thonest does not take the submitter's word for a quote. The platform is asked directly and the wording has to appear in its own response, or the record is refused.
• An independent copy. The Internet Archive is asked to save the page at capture time, on a server Thonest does not control.
• Two dates, never mixed up. When it was said, and when Thonest registered it. Only a record registered before the outcome was known can claim it was locked in advance; anything added later is labelled retrospective and kept out of pre-outcome track records.
• Tamper-evident by design. Records are append-only, batches are anchored on-chain, and the latest anchor is published on the homepage with a link to a block explorer rather than to Thonest. Corrections are appended in the open; the original stays visible.

There is also a dispute lane with published commitments - acknowledged within 72 hours, resolved or given a status within 14 days, outcome appended to the record either way - and the person a record is about gets priority.

Pros

• No account needed to read, browse or verify. The proof is genuinely public.
• Registering calls made elsewhere is the killer feature - your best prediction is almost never on the site where the receipts live.
• The locked test removes the usual argument. Nobody gets to reinterpret what they meant after the fact.
• Refusing to score vague statements is a rare piece of restraint. Most accountability projects manufacture precision they do not have.
• The corrections and disputes policy is published up front, not buried.

Cons

• It is narrow. This is a record for predictions, not a general reputation or review tool.
• Verification depends on the source still being fetchable. A deleted post or a paywalled video is harder to lock.
• The record is only as good as the test you confirm, and writing a genuinely tight test takes thought.
• The volume is still small. A public record gets more useful the more people are on it.

What it deliberately does not claim

Thonest is upfront that none of this proves a statement is true. Provenance shows where something came from; anchoring shows it has not changed since. As the site puts it, anyone telling you a blockchain makes something true is selling you something. We appreciated reading that on the homepage rather than having to work it out.

Summary

Thonest's instinct is simple - make claims checkable. If you make public calls, whether that is a market forecast, a product ship date, or a take on a podcast, this is the cheapest way to give them weight, and it costs nothing to be verified by someone who thinks you are wrong. Worth a look.