Archived dispatch

What is easy to misunderstand about Keryx access tolls versus citation rewards, and what does the evidence actually show?

Moderateconfidence— 1 evidence-verified source cover every sub-claim, but corroboration or support strength is limited

9/30/2026, 1:17:56 PM · llm:deepseek:deepseek-v4-flash

§ IIThe reading1 cited
Moderateconfidence— 1 evidence-verified source cover every sub-claim, but corroboration or support strength is limiteddeep researchpreview plan 2/2 claimsportfolio 2/2 · evidence 50%

A common misunderstanding is that paying to read a source and being rewarded for citing it are the same thing. In Keryx, they are explicitly separate: the system "separates the cost of opening source content from the reward for citing it" , where an access toll "purchases a read through the source's x402 endpoint" while a citation reward is "a separate payment considered after the answer is written and its evidence checked" . Consequently, "a paid read does not guarantee a citation reward" , and cached content "can avoid a new access toll while still contributing to an answer and qualifying for a citation reward" .

What the evidence actually shows is that the two flows settle independently. When no citation qualifies, "the citation pool stays unspent, while access tolls already settled remain paid" , and the receipt "distinguishes settled and pending payment records" . Only sources accepted by the evidence ledger enter the citation-pool allocation , and creator payment calls use the source's authoritative payment terms rather than a model-proposed weight . Note that this is first-party engineering documentation, "not independent reporting or customer evidence" . The second source discusses nanopayment economics generally and does not address Keryx access tolls versus citation rewards, so it provides no evidence on this question.

Evidence ledger — quotes verified before rewards

  1. What is easy to misunderstand about Keryx access tolls versus citation rewards?

    90%
    “Keryx separates the cost of opening source content from the reward for citing it.” [S1] How Keryx pays cited creators
    “A paid read does not guarantee a citation reward.” [S1] How Keryx pays cited creators
  2. What does the evidence actually show about Keryx access tolls versus citation rewards?

    90%
    “When no citation qualifies, the citation pool stays unspent, while access tolls already settled remain paid.” [S1] How Keryx pays cited creators

Cited sources and planned rewards

Helpful?
Spent$0.015
To creators100%
Decisions0 bought · 2 cached · 17 skipped
llm:deepseek:deepseek-v4-flashlive on Arc testnet
Decision log · 48 steps
§ IThe decision$0.015 settled / $0.03
50%$0.015 under cap
Decompose

Breaking down: "What is easy to misunderstand about Keryx access tolls versus citation rewards, and what does the evidence actually show?"

Decompose

Identified 2 research target(s) to investigate; these are not established facts

Decompose

Deep mode: up to 4 paid/cached reads plus one bounded gap-expansion pass when needed.

Discover

Discovered 21 verified source(s)

Discover

Recalled 53 past runs on this subject — how these sources performed when they were available.

Discover

ERC-8004 reputation loaded — composite scores on this subject.

Pre-check

Claim-aware portfolio selected 2/2 positive proposal(s): 2 cached + 0 fresh, predicting 2/2 claim(s) above the evidence floor with $0.000000/$0.015000 fetch USDC reserved.

Pre-check

Free-preview pre-check maps an actionable source to every sub-claim (2/2); paid reading may proceed within the budget.

DecideCACHE
Keryx Engineering (first-party) — How Keryx pays cited creators$0.002 · EV 90%

First-party Keryx engineering doc explicitly covers 'Keryx access tolls, citation rewards, exact-quote evidence checks, contribution weights and settlement limits' — directly the distinction the question asks about, and it's full_text (3473 bytes) so it can actually answer both the misconception and the evidence. Already cached, so reuse free. Highest topical fit despite mid reputation (49/100, 17 citations). — selected for the claim-aware evidence portfolio (targets claims 1, 2; 0 fetch USDC, 1 attention slot).

DecideCACHE
Onchain Micropayments Digest — Nanopayments and the $0.000001 floor$0.005 · EV 45%

Onchain Micropayments Digest has the best citation record here (7/13, 54%) and its nanopayment/batching floor preview bears on how tolls and reward economics actually work at sub-cent scale — relevant to what the evidence shows. Cached, so free to reuse. — selected for the claim-aware evidence portfolio (targets claim 2; 0 fetch USDC, 1 attention slot).

DecideSKIP
Arc Settlement Benchmarks — Measuring x402 settlement latency on Arc$0.003 · EV 35%

Arc Settlement Benchmarks measures x402 batched-settlement finality on Arc — the rail Keryx tolls settle on, so it can ground the 'what the evidence actually shows' half of the question. Cached and cheap; moderate reputation (17/100). — cached bytes are free, but this read does not clear the attention gate (EV 0.35, minimum 0.45, with a required claim target).

DecideSKIP
Web Payments Review — How long do x402 payments take to finalize?$0.002 · EV 30%

Web Payments Review covers end-to-end x402 settlement timing, useful corroboration for the evidence claim about how tolls settle versus when citation rewards are paid. Cached; low reputation (20/100) so secondary. — cached bytes are free, but this read does not clear the attention gate (EV 0.30, minimum 0.45, with a required claim target).

DecideSKIP
Agent Economy Weekly — x402 turns HTTP 402 into an agent payment rail$0.004 · EV 30%

Agent Economy Weekly explains x402 turning HTTP 402 into an inline agent payment rail — background for why access tolls are paid up front, which frames the toll-vs-reward misunderstanding. Cached; weak citation record (25%). — cached bytes are free, but this read does not clear the attention gate (EV 0.30, minimum 0.45, with a required claim target).

DecideSKIP
Stablecoin Ledger — Why USDC settles instantly onchain$0.003 · EV 10%

Stablecoin Ledger's preview is only about USDC settling in seconds on L2s — generic settlement speed, nothing about Keryx tolls, citation rewards, or the misconception. Abstract-only and low reputation (24/100); not worth even a cached read for these targets.

DecideSKIP
Distributed Systems Notes — Idempotency keys prevent double-spends$0.003 · EV 8%

Distributed Systems Notes preview is about idempotency keys preventing double-spends — a generic reliability pattern, not Keryx tolls or citation rewards. Weakest reputation tier (10/100) and off-topic for both sub-claims.

DecideSKIP
Stripe Blog — Analyzing the evidence that helps businesses win “product not received” disputes$0.002 · EV 10%

Stripe Blog has the best reputation (57/100) but this specific article is about winning 'product not received' chargeback disputes — payments disputes, not Keryx access tolls or citation rewards. No target support.

DecideSKIP
Ethereum Foundation Blog — The triage is the product: running AI agents against Ethereum's protocol code$0.002 · EV 8%

Ethereum Foundation Blog piece is about running AI agents against Ethereum protocol code — unrelated to Keryx toll/reward mechanics, and this source has never been cited on this subject.

DecideSKIP
Cointelegraph.com News — Ethena launches USDe payments app, offers 6% rewards$0.002 · EV 10%

Cointelegraph piece is about Ethena's USDe payments app and 6% rewards — 'rewards' here means yield, not Keryx citation rewards, so it risks exactly the confusion the question warns about without resolving it. Not cached and only 131 bytes of abstract.

DecideSKIP
Latent.Space — Ontologies Are So Back: Why AI Agents Are Reviving the Semantic Web$0.004 · EV 10%

Latent.Space on ontologies and the semantic web is about agent reasoning boundaries, not payment tolls or citation rewards. No overlap with either sub-claim.

DecideSKIP
Simon Willison's Weblog — Feeling sad about AI$0.003 · EV 5%

Simon Willison 'Feeling sad about AI' is metadata_only with zero plaintext bytes and no topical link to Keryx tolls or citation rewards.

DecideSKIP
Hugging Face - Blog — BenchMIRT: What are LLM benchmarks actually measuring?$0.003 · EV 5%

Hugging Face BenchMIRT is about what LLM benchmarks measure — metadata_only, no content, and unrelated to payment tolls or citation rewards.

DecideSKIP
Vitalik Buterin's website — Memory access is O(N^[1/3])$0.004 · EV 5%

Vitalik's 'Memory access is O(N^1/3)' is a systems-performance note, metadata_only, with no bearing on Keryx toll/reward semantics.

DecideSKIP
The Coinbase Blog - Medium — In response to the Wall Street Journal$0.003 · EV 8%

Coinbase's response to the WSJ is about proprietary trading allegations — corporate PR, not Keryx access tolls or citation rewards.

DecideSKIP
Decrypt — Putin Signs Russia's First Crypto Law: Trading Is Legal, Payments Stay Banned$0.002 · EV 8%

Decrypt piece on Russia's crypto law concerns trading legality vs payment bans — regulatory news, no connection to Keryx tolls or citation rewards. Decrypt has never been cited on this subject.

DecideSKIP
CoinDesk: Bitcoin, Ethereum, Crypto News and Price Data — Crypto Long & Short: Inside the 300-to-1 onchain gap between the dollar and euro$0.002 · EV 10%

CoinDesk piece is about the dollar/euro onchain stablecoin gap — macro stablecoin adoption, not Keryx toll-vs-reward mechanics. Only 1 citation in 5 runs here.

DecideSKIP
Inner Axiom — The Codex — The Journey of the Soul$0.002 · EV 2%

Inner Axiom esoteric cosmology piece — wholly unrelated to payments, tolls, or citation rewards.

DecideSKIP
Conzit Labs — St. Vincent's Hollywood Bowl Show: A Theatrical Musical Triumph$0.002 · EV 2%

Conzit Labs concert review of St. Vincent — no relevance to Keryx tolls or citation rewards.

Fetch

Agent spend wallet ready: 0x29028Fe1122E17Fe7863A22701e863FE4DaE1aFB (balance sufficient)

Fetch

Reused cached Keryx Engineering (first-party) — How Keryx pays cited creators (free) — S1

Fetch

Reused cached Onchain Micropayments Digest — Nanopayments and the $0.000001 floor (free) — S2

Re-evaluate

Sub-claim "What is easy to misunderstand about Keryx access tolls versu…": 80% covered by S1 — S1 directly identifies the core misconception: access tolls and citation rewards are separate, a paid read does not guarantee a citation reward, cached content can avoid a new access toll while still qualifying for a citation reward, and model-proposed weights do not choose recipients or change payout authority. It does not explicitly frame these as 'easy to misunderstand' or enumerate common misreadings, so a small requested gap remains.

Re-evaluate

Sub-claim "What does the evidence actually show about Keryx access toll…": 70% covered by S1 — S1 provides first-party engineering evidence: tolls purchase reads via x402, rewards are considered after the answer and evidence check, only ledger-accepted sources enter the citation-pool allocation, unspent citation pool remains when no citation qualifies, and settled access tolls stay paid. It is explicitly first-party documentation, not independent reporting or customer evidence, so the 'evidence actually shows' part is answered only for first-party engineering evidence.

Re-evaluate

Coverage is adequate for both sub-claims via S1, which directly addresses the toll-versus-reward distinction and the first-party evidence. The skipped sources concern x402 settlement latency, USDC finality, idempotency, disputes, and unrelated topics; none would materially improve coverage of the specific Keryx toll-versus-citation-reward misunderstanding. No purchase is warranted within the remaining budget.

Sufficiency

Final check — "What is easy to misunderstand about Keryx access tolls versu…": 100% assessed by S1

Sufficiency

Final check — "What does the evidence actually show about Keryx access toll…": 100% assessed by S1

Sufficiency

Final coverage assessment — S1 directly addresses both sub-claims: it explains the distinction between access tolls and citation rewards, notes that a paid read does not guarantee a citation reward, and describes what the evidence ledger and payment records actually show. S2 is only topical context about nanopayments and does not address Keryx access tolls versus citation rewards.

Synthesize

Synthesizing a grounded answer from 2 source(s)…

Evidence

Relevance review returned; only checked excerpts can retain support, and review cannot raise it.

Evidence

Verified — S1 supports claim 1 at 90%: “Keryx separates the cost of opening source content from the reward for citing it.”

Evidence

Verified — S1 supports claim 1 at 90%: “A paid read does not guarantee a citation reward.”

Evidence

Verified — S1 supports claim 2 at 90%: “When no citation qualifies, the citation pool stays unspent, while access tolls already settled remain paid.”

Evidence

Below reward gate — S1 supports claim 2 at 20%: “Only sources accepted by the evidence ledger enter this allocation.”

Evidence

Rejected 0 invalid evidence span(s) and 1 unsupported citation marker(s); rejected markers cannot receive citation rewards.

Synthesize

Drafted answer citing 1 source(s)

Verdict

Confidence: Moderate — 1 evidence-verified source cover every sub-claim, but corroboration or support strength is limited.

Attribute

Keryx Engineering (first-party) contributed 100% → reward $0.015

Settle

Settled $0.015 citation reward → 0x6644A7C63C559454e77D5834554DCa3a60fcFDA2 (b209b27d-2…)

Done

Done. Spent $0.015 across 1 confirmed/simulated payment(s) to creators.

Portable research receipt

Take the evidence trail with you

One deterministic JSON bundle binds the answer, visible decisions, exact article versions, claim evidence and a Circle-settlement snapshot under SHA-256. Retain the digest to detect later changes; the self-check is not a publisher or Keryx signature.

Exact receipt still current

1 exact cited article version still match Keryx's current index. The source cited here has published nothing new since this dispatch settled.

Inspect machine-readable audit

Ask a follow-upNew dispatch · creators paid again

Carries this dispatch’s question as context — never its answer. The next dispatch is read from sources bought for it.

From the archive

Related dispatches