Tell Talarion

April 6, 2026

Manifestos are popular, and this is ours. Welcome to the marketplace for information.

First Principles

In our previous post, we outlined three core challenges for information markets: Novelty, Credibility, and Routing. Add in Liquidity, and you have the Big Four Problems.

  • Novelty: Information that’s being sold can’t be freely available. Anywhere.
  • Credibility: Sellers must be incentivized to tell the truth.
  • Routing: Buyers need to be matched with sellers.
  • Liquidity: The moment that a buyer needs information is almost never the same exact moment that a seller makes it available.

For a marketplace to function, you need to solve for all of them. For most of human history, this was an insurmountable task. But language models might just make it possible.

Balliol College

My grandfather, Nicholas Hope the elder, was a Rhodes Scholar of Balliol College. He taught me a little rhyme which feels surprisingly apt, despite being more than 140 years old:

First come I. My name is Jowett.
There’s no knowledge but I know it.
I am Master of this College,
What I don’t know isn’t knowledge.

Oh, what I would do to have Jowett on retainer! Since his passing in 1893, nobody has been available to act as an arbiter of novelty. Until Claude, that is. Modern LLMs combined with powerful information retrieval are an efficient and effective benchmark for knowability. If Claude doesn’t know it, it isn’t knowledge.

Put Your Money Where Your Mouth Is

Falling outside of an LLM’s world model may be an adequate test of Novelty, but to ensure Credibility, we’ll need two things: verifiability and incentive alignment. The verifiability constraint is simple but painful: the information you sell needs to be novel today, knowledge tomorrow. On Talarion, you can choose when your claim is resolved, but we will resolve it. In practice, that limits the information you can sell to a rather specific class: things which an LLM can’t find out today but can find out in the future.

In the ‘trust now, verify later’ paradigm, incentive alignment is easy. When you make an assertion, you post a bond. If you told the truth, you receive a reward. If you didn’t, well, your bond is gone. Depending on your frame of mind, this might sound like a prediction market, but we’d prefer an oracle. We’re in the market for information — not speculation.

Tell Talarion

Talarion solves the Routing and Liquidity problems by serving as a broker. We’ll pay you for what you know, when you want, if you want. There’s no need to seek out the right buyer, or to wait for their entry into the market. Nobody knows your mind better than you do, which is why we generate questions based on whatever you tell us. Market creation is meant to be a collaborative exercise in fact finding. In my first post I highlighted the need to answer questions that were worth answering. We’ll pay you for whatever you know that we don’t so long as it matters, it’s credible, and in the end, it’s verifiable. If nobody ended up needing the piece of information that you sold us, that’s our problem, not yours. So if you know something, come Tell Talarion.

Ask Talarion

Information sellers Tell Talarion, but buyers Ask Talarion. Value is already concentrating in the difference between what humans know about the world as it is and will be, and what language models can find out. That’s what we pay for by construction. So if you’re wondering why we can hand out money to our users, here’s the answer: because anyone who wants to empower AI to make decisions needs the answers that only we have.