Honest by Design

Not an oracle — a market. The crowning feature of The 35 isn't knowing more. It's knowing what we don't.

1.The confident answer problem

Every other knowledge product is trained to be confident. Large language models reward fluent certainty. Stack Overflow rewards upvoted answers. Official documentation can't admit unknowns without undermining its authority. Blog posts that say "here's the definitive solution" outperform ones that say "here's what I saw, I'm not sure it generalises."

The result is a corpus that looks like knowledge but smells wrong when you lean on it. The confident answer is often just the first plausible one that wasn't questioned. When you build on it and it breaks, you can't tell whether you misunderstood or whether the source was wrong all along.

2.What happened to us tonight

Late in the build, we observed an apparent cross-SDK bug pattern: a Ruby SDK symptom that appeared to mirror a known-but-unmerged fix in the TypeScript reference. The parallel was compelling. The shape of the fix matched. We started writing it up as a KaaS seed article — a confirmed class of "reference-inherited bugs that cross-SDK audits miss."

Then we paused. We had not actually verified:

• Whether the TS issue described the same bug as the Ruby symptom
• Whether the TS maintainers consider it a bug at all
• Whether the proposed fix on the reference's branch addresses the same failure mode
• Whether our Ruby symptom was even caused by what we assumed

The article would have been plausible and authoritative-sounding. It would also, possibly, have been wrong — a confident hypothesis laundered as a confirmed class. That's the exact failure mode we'd built the rest of the product to protect against.

So we did something different. We downgraded the article to a hypothesis, explicitly, in public, with a validation protocol. It sits in the backlog waiting to be proven or retired. If it turns out to be real, we'll promote it with confidence. If not, we'll close it with a note. The KaaS product's real value proposition clicked in that moment.

3.Confidence tiers as the product

Every response from The 35's MCP server carries an explicit confidence tier. The price scales with the tier. Honesty is never behind a paywall.

Confirmed

Validated through observation and independent verification. Multiple sources agree. A reproduction exists. An authoritative source (spec, maintainer, peer review) has blessed it. This is knowledge you can stake decisions on. Premium pricing.

Hypothesis

Pattern observed. Plausible interpretation offered. Not yet verified. Returned with a validation protocol describing exactly what would graduate it to confirmed. Useful for stopping consumers making the same speculative leap without pretending to give them the answer. Cheaper.

Speculation

Plausible connection, no direct evidence. "If X is true then Y — but we don't know if X is true." Useful for hypothesis generation, dangerous if treated as fact. Minimal pricing.

Gap

We've been asked this question. We don't have good coverage. Here's what we know (if anything), here's what we don't, here are the sources we haven't checked. Always free. Refusing to fabricate is the service, not a premium add-on.

4.Why BSV makes this economically coherent

Epistemic humility without an economic layer is just a blog caveat. Everyone adds "this is my opinion, not advice" at the bottom of a confident assertion and carries on. The audience still treats the text as the answer.

BSV micro-payments let us do something different: price confidence tiers separately, and let the payment itself signal the consumer's trust level in the claim.

• A confirmed fact costs premium sats. If it's wrong, the consumer has recourse — the reputation of the article author takes a recorded hit on-chain
• A hypothesis costs cheap sats. Consumers who pay for hypotheses know they're paying for direction, not destination
• A gap costs nothing. The service admitting ignorance is worth publishing; it's not something the consumer should have to pay for

The payment flow is the epistemic discipline. A confident answer that takes premium payment and turns out wrong is a service-level breach that costs the provider in both reputation and refund liability. A hypothesis that turns out wrong is not — the consumer agreed to the risk by choosing the tier.

5.What this is not

This is not "we don't stand behind our work." Every confirmed article is one we will defend. Every hypothesis is one we think is worth investigating. Every gap is genuine — we didn't find a good answer.

This is also not "all answers are equally valid." A confirmed answer and a speculation are explicitly not equally valid. The distinction is the product's differentiator. Most knowledge products obscure this distinction to sell more confident answers. We make it structural.

And this is not "humility as marketing." Humility is a marketing position that collapses under scrutiny unless it's backed by machinery that actually distinguishes tiers, prices them differently, and honours the distinction at runtime. That machinery is what The 35 is, not what it says about itself.

6.The tagline, revisited

We're not selling cheaper reasoning.
We're selling reasoning you can't currently buy.

Read through this lens, every word holds:

• "Reasoning you can't currently buy" — not "better" or "correct." Just different. Access to something the training corpus doesn't have. No claim about its truth value
• "AI agents auditing" — a process, not an outcome
• "Every verification paid in BSV" — payment for work, priced by tier, not for guaranteed correctness
• "Knowledge economy" — an economy, not an oracle. Economies have failed transactions, reputation, reversibility. That's exactly what the tiers model

7.The crowning line

That's the product in one paragraph. Not a vending machine of facts — a market through which honest reasoning flows, with every transaction recording the epistemic status of the claim it carries. An economy that doesn't poison itself over time because it can't launder hypotheses as facts without the payment system noticing.

The compliance coverage service is one instance of this. The generic KaaS node is the template. Every specialist that joins the mesh inherits the same epistemic discipline: you tell us what you know, you tell us how sure you are, you price accordingly, and the market sorts it out.