Repository navigation
fix(copy): align homepage, faq, docs and blog copy with protocol adrs - #251
Merged
Merged
Conversation
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Deploying website with
|
| Latest commit: |
675ec96
|
| Status: | ✅ Deploy successful! |
| Preview URL: | https://43159752.website-70y.pages.dev |
| Branch Preview URL: | https://claude-charming-swartz-77678.website-70y.pages.dev |
The docs and one blog post said clients rank candidates by a unified price, latency and reputation score, and that the client finds holders through a DHT lookup. Per ADR 039 § Source set and selection, ADR 037 § Client selection policy and ADR 012, the client takes candidates from its peer store or the CapacityBond registry, ranks by measured RTT alone, pays the signed rate subject to an optional ceiling, and keeps no reputation score. The unified score is the node's, on its cache-miss pull leg (ADR 001 § Node Selection Algorithm). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XutWi52yRav4gAzyrDBLZH
The 500 ms probe cap and its early stop belong to the node's cache-miss pull leg (ADR 001 § Content Discovery). The client's probe round waits on every shortlisted probe, so the cap misdescribed the client. The source count is the configurable max_sources (ADR 039), so the hero and swarm copy now say several nodes instead of up to four. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XutWi52yRav4gAzyrDBLZH
alpergundogdu
approved these changes
Sep 27, 2026
3 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Parts of the homepage copy contradicted the protocol ADRs (
decdn/adr/), which are the source of truth. This PR fixes the strings inlib/copy.tsandlib/faq.ts. The llms-full.txt mirror and the FAQPage JSON-LD read the same strings, so they update with no code changes. Role wording now follows the ADR terms: node, operator, client, publisher and pool, with "node" in place of "peer".Method section (
lib/copy.ts)max_sourcesis configurable (default 4).Hero and compare
HERO_LEAD: "three peers" becomes "several nodes" (max_sourcesis configurable, ADR 039), and the unsourced "under a second" Berlin→Tokyo claim is dropped.COMPARE_ROWSfailure row: "peer drops" becomes "node drops".FAQ (
lib/faq.ts)Docs and blog: client ranking (
docs/,content/blog/)The same client-ranking claim appeared in the Mintlify docs and one blog post. The ADRs were corrected in decdn/decdn#1991 (2026-09-11) to describe the shipped client: candidates come from the peer store or the registry (no DHT lookup), ranking is by measured RTT only, and there is no client reputation score. The unified price/latency/reputation score belongs to a node's cache-miss pull leg (ADR 001 § Node Selection Algorithm).
docs/overview/architecture.mdx: Discovery and Selection steps rewritten for the client path.docs/overview/how-it-works.mdx: the single-source pick and the multi-source admit order are now RTT-based; coverage is attributed to probes, not client discovery.docs/overview/glossary.mdx: Selection score is scoped to the node pull leg.docs/protocol/network.mdx: § Selection now covers clients and nodes separately; the region-mismatch reputation penalty is applied by the pulling node; the region paragraph says the client only uses region to shortlist probes.docs/protocol/reputation.mdx: reputation tunes a pulling node's choice; clients keep none.content/blog/06-show-me-the-money.mdx: "clients choose among price, latency, and reputation" becomes RTT ranking plus a client rate ceiling.content/blog/03-we-dont-store-we-serve.mdx's "unified selection score", which describes node-to-node propagation and is still accurate.Other
AGENTS.md: dropped the staleMETHOD_FIGURESreference. design: homepage critique pass — hero hierarchy, copy voice, panel consistency #250 already removed that export, along with its "chain-agnostic" claim.Reviewer notes
HERO_FIGURES"p50 latency 50–100 ms" has no ADR source and is flagged for a later decision.Test plan
pnpm test(465 passed, 2 skipped)pnpm typecheckpnpm lintpnpm format:checkmarkdownlint-cli2overdocs/mintlify broken-links(no broken links; covers the new/protocol/network#selectionlink)🤖 Generated with Claude Code
https://claude.ai/code/session_01XutWi52yRav4gAzyrDBLZH