Why is per-citation settlement fairer than a flat per-fetch toll?
9/18/2026, 6:47:57 PM · llm:deepseek:deepseek-v4-flash + llm:mimo:mimo-v2.5 on 1 step
The dispatch, itemised.
Breaking down: "Why is per-citation settlement fairer than a flat per-fetch toll?"
Identified 3 research target(s) to investigate; these are not established facts
Deep mode: up to 4 paid/cached reads plus one bounded gap-expansion pass when needed.
Discovered 21 verified source(s)
Recalled 25 past runs on this subject — how these sources performed when they were available.
ERC-8004 reputation loaded — composite scores on this subject.
Claim-aware portfolio selected 2/2 positive proposal(s): 2 cached + 0 fresh, predicting 2/3 claim(s) above the evidence floor with $0.000000/$0.025000 fetch USDC reserved.
Free-preview pre-check covers 2/3 sub-claims (67%). The agent may buy only claim-targeted sources and will label the answer provisional if paid evidence stays thin.
This source is the most directly relevant and has the highest historical citation rate (80%) on this subject. The preview explicitly mentions 'Per-citation payments weighted by contribution' and 'Reward sources in proportion to how much they grounded an answer,' which directly addresses the fairness comparison at the core of subClaim 0 and the properties attributed to per-citation settlement (subClaim 2). It is cached, so we can reuse it for free. — selected for the claim-aware evidence portfolio (targets claims 1, 3; 0 fetch USDC, 1 attention slot).
This source provides technical benchmarks for x402 settlement on Arc, which is a specific per-citation settlement system. The preview mentions 'Benchmark methodology and results for x402 batched-settlement finality,' which could provide concrete evidence for the fairness and efficiency of per-citation settlement (subClaim 2). It has a solid historical citation rate (53%) and is cached. — selected for the claim-aware evidence portfolio (targets claim 3; 0 fetch USDC, 1 attention slot).
This source is about Russian crypto law, not about payment settlement fairness. The preview ('Trading Is Legal, Payments Stay Banned') is irrelevant to per-citation vs. per-fetch toll fairness. No targets are supported.
This source discusses stablecoin adoption gaps, not payment model fairness. The preview mentions '300-to-1 onchain gap between the dollar and euro,' which is about currency usage, not the fairness of per-citation vs. per-fetch tolls.
This source is about fraud rates at AI startups, not payment settlement models. The preview discusses 'fraud attempts' and 'customer abuse patterns,' which is unrelated to the fairness of per-citation vs. per-fetch tolls.
This is Keryx's own engineering notes about buyer recovery, not an independent analysis of payment model fairness. The preview describes a technical process ('recovering a paid research job'), not the fairness comparison. It has a low historical citation rate (20%) on this subject.
This source is about settlement timing, not the fairness of payment models. The preview asks 'How long do x402 payments take to finalize?' which is about latency, not the fairness of per-citation vs. per-fetch tolls. It has a low historical citation rate (11%) on this subject.
This is a Coinbase blog post responding to a Wall Street Journal article about proprietary trading, not about payment model fairness. The preview discusses 'proprietary trading' and 'market maker' activities, which is irrelevant to the question.
This is a personal blog post titled 'Feeling sad about AI' with no preview content about payment fairness. It is metadata_only and not relevant.
This source is about ontologies and the semantic web, not payment settlement models. The preview mentions 'AI engineers are rediscovering ontologies,' which is unrelated to the fairness of per-citation vs. per-fetch tolls.
This source is about rogue AI agents and security hacks, not payment fairness. The preview discusses 'serious security vulnerabilities,' which is irrelevant.
This is a blog post about building an AI agent called Shippy, with no preview content about payment fairness. It is metadata_only and not relevant.
This is a mystical/esoteric article about the soul, completely irrelevant to payment models. No targets supported.
This is about low-risk DeFi for Ethereum, not about payment settlement fairness. The preview is metadata_only and not relevant.
This is about malware and crypto theft, not payment fairness. The preview discusses 'malware that redirected about $150,000 in crypto,' which is unrelated.
This is about recapping retro game consoles, completely irrelevant to payment models.
This is about gardening and building raised beds, completely irrelevant to payment models.
This source is about idempotency keys to prevent double-spends, which is a technical detail but not directly about the fairness comparison of payment models. The preview discusses 'Use a unique key per operation to make retries safe,' which is about reliability, not fairness.
This source is about stablecoin settlement speed, not payment model fairness. The preview mentions 'USDC transfers settle in seconds,' which is about latency, not the fairness of per-citation vs. per-fetch tolls. It has a 0% citation rate on this subject.
This source is about the x402 standard for agent payments, which is tangentially related but the preview focuses on the technical mechanism ('lets a server demand payment and an agent pay inline') rather than fairness. It has a 0% citation rate on this subject.
This is about running AI agents against Ethereum's protocol code for security, not payment fairness. The preview discusses 'Protocol Security team' and 'coordinated AI agents,' which is unrelated.
Agent spend wallet ready: 0x72449d71ea6a955aD4d4d22C2fDAcdFCA622b9BE (balance sufficient)
Reused cached Onchain Micropayments Digest — Per-citation payments weighted by contribution (free) — S1
Reused cached Arc Settlement Benchmarks — Measuring x402 settlement latency on Arc (free) — S2
Sub-claim "Why is per-citation settlement considered fairer than a flat…": 30% covered by S1 — S1 asserts a fair model pays each cited source in proportion to its contribution, with heavily-relied-upon sources earning more, but it does not explicitly compare this to a flat per-fetch toll or explain why the flat toll is less fair.
Sub-claim "How does a flat per-fetch toll work, and what fairness conce…": 0% covered — No supplied passage describes a flat per-fetch toll mechanism or states fairness concerns about it. S1 only contrasts weighted per-citation payment with lighter/heavier usage, not with a flat per-fetch toll.
Sub-claim "How does per-citation settlement work, and what fairness pro…": 60% covered by S1 — S1 directly describes per-citation settlement: pay each cited source in proportion to contribution, weighted nanopayments for granular settlement, and splitting rewards across multi-author works. It attributes fairness (proportionality) but does not fully explain the settlement mechanism or compare it to alternatives.
Coverage is partial but the main fairness rationale is present in S1. The skipped sources are mostly irrelevant to per-citation vs flat per-fetch fairness; none clearly fills the missing flat-toll mechanism/fairness gap. Given the small remaining budget and lack of a directly relevant affordable source, no purchase is recommended.
Final check — "Why is per-citation settlement considered fairer than a flat…": 40% assessed by S1
Final check — "How does a flat per-fetch toll work, and what fairness conce…": 0% assessed
Final check — "How does per-citation settlement work, and what fairness pro…": 90% assessed by S1
Final coverage assessment — The supplied passages only address per-citation/weighted payment fairness in S1. S1 states that a fair model pays each cited source in proportion to its contribution, with heavily-relied-upon sources earning more and lightly-used ones earning less, and that weighted nanopayments make granular settlement practical. This directly answers the per-citation fairness sub-claim and the per-citation mechanics/fairness-properties sub-claim. However, no passage explains how a flat per-fetch toll works or what fairness concerns are raised about it; S2 is about settlement latency on Arc and does not address toll mechanics or fairness. Therefore the comparison requested in the main question is only partially supported: the per-citation side is covered, but the flat per-fetch toll side and the explicit comparative fairness argument are missing. The assessment does not establish a complete supported answer for every requested part.
Synthesizing a grounded answer from 2 source(s)…
Relevance review returned; only checked excerpts can retain support, and review cannot raise it.
Verified — S1 supports claim 1 at 60%: “A fair model pays each cited source in proportion to its contribution to the final answer.”
Verified — S1 supports claim 1 at 50%: “Heavily-relied-upon sources earn more; lightly-used ones earn less.”
Verified — S1 supports claim 3 at 50%: “Weighted nanopayments make this granular settlement practical, and multi-author works can split a single reward across contributors automati…”
Below reward gate — S2 supports claim 2 at 0%: “Across thousands of submitBatch calls on Arc testnet, x402 batched settlements finalize in roughly 180 milliseconds (measured median 178ms, …”
Rejected 0 invalid evidence span(s) and 1 unsupported citation marker(s); rejected markers cannot receive citation rewards.
Drafted answer citing 1 source(s)
Confidence: Low — 1 sub-claim remains below the evidence threshold.
Onchain Micropayments Digest contributed 100% → reward $0.025
Paid $0.015 citation reward → Mara Okoye; Circle confirmed settlement even though the paid route acknowledgement failed.
Paid $0.01 citation reward → Devin Park; Circle confirmed settlement even though the paid route acknowledgement failed.
Done. Spent $0.025 across 2 confirmed/simulated payment(s) to creators.
Onchain Micropayments Digest
batched
Onchain Micropayments Digest
batched
> ⚠ Low confidence — 1 sub-claim remains below the evidence threshold within budget. Treat this as provisional.
The sources do not explain why per-citation settlement is fairer than a flat per-fetch toll. describes a fair per-citation model: it "pays each cited source in proportion to its contribution to the final answer," so "[h]eavily-relied-upon sources earn more; lightly-used ones earn less," and notes that "[w]eighted nanopayments make this granular settlement practical" . However, none of the supplied passages describe how a flat per-fetch toll works, nor do they raise any fairness concerns about such a toll, so the comparative fairness question cannot be answered from these sources. is unrelated to fairness, covering only x402 batched settlement latency on Arc testnet.
Evidence ledger — quotes verified before rewards
Why is per-citation settlement considered fairer than a flat per-fetch toll?
40%“A fair model pays each cited source in proportion to its contribution to the final answer.” [S1] Per-citation payments weighted by contribution
“Heavily-relied-upon sources earn more; lightly-used ones earn less.” [S1] Per-citation payments weighted by contribution
How does a flat per-fetch toll work, and what fairness concerns are raised about it?
0%No reward-qualifying evidence
How does per-citation settlement work, and what fairness properties are attributed to it?
50%“Weighted nanopayments make this granular settlement practical, and multi-author works can split a single reward across contributors automatically.” [S1] Per-citation payments weighted by contribution
Footnotes — each one pays its author
- 1Per-citation payments weighted by contributionOnchain Micropayments Digest100%+$0.025
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.
Carries this dispatch’s question as context — never its answer. The next dispatch is read from sources bought for it.