anthropolo.gyYour collection ↗
Field notes / edition 001

A network you can check.

Every published connection points to its evidence. The graph is a growing research index, with a clear boundary between documented relationships and leads awaiting review.

01 / Provenance

Follow the original source

Edges retain source references and dates where supported. Official biographies, institutional histories and company announcements anchor the initial graph.

02 / Identity

One person, one canonical node

When a person has a DevReal record, that identity is reused here. Name similarity alone is insufficient to merge two people.

03 / Interpretation

A mention is a research lead

People appearing in the same article are not automatically colleagues, investors, or friends. Archive candidates require review before becoming factual edges.

Collection proposal

Browse the scanned archives ↗ · Explore person–eigenvector histories ↗

These are proposed collection cadences. Collection jobs must obey source permissions and current robots rules, cache responses, identify themselves, and limit requests per host. Restricted sources require a permitted feed, existing archive, or license.

ResourceCadenceAccessPurpose
Eigenhacks / Hacker News archiveDaily incremental; weekly partition and duplicate auditExisting local/production archive through its maintained ingestion pipeline; official HN API metadata when configured.New article metadata; Name-match candidates; Co-mention review queue; Stable projection joins
The New York Times Archive APIMonthly previous-month archive; weekly current-month refresh within API quotaOfficial Archive API using the existing private API key; preserve typed keyword categories in new metadata.Article headline and URL; Publication date; Keyword name/value/rank; Archive month coverage
The Guardian Open PlatformDaily incremental metadata; monthly completeness auditExisting authorized upstream adapter and local archive; key-specific terms and quotas apply.Article metadata; Named-entity candidates
TechCrunchWeekly public search-metadata review; daily feed only after explicit access reviewLocally held archive metadata and public search discovery now; direct AI collection blocked by current robots policy.Source links; Funding/exit leads requiring primary confirmation
WIREDWeekly public search-metadata review; licensed feed after access reviewLocally held archive metadata and public search discovery; AI crawler restrictions apply.Historical profile leads; Source links
CrunchbaseMonthly licensed API/export reconciliation; weekly permitted public search metadataSearch/answer user agents including ChatGPT-User and OAI-SearchBot are allowed in the snapshot; GPTBot and other bulk crawlers are disallowed. Structured bulk collection needs licensed API/export.Person/company identities; Funding/acquisition dates for review
Official company, fund and university biographiesWeekly conditional check of curated allowlist; daily change queuePublic source pages only after per-host robots and terms review.Role changes; Founding and investment evidence; Dated career events
Computer History Museum, Smithsonian, NASA, Nobel Prize and institutional archivesMonthly source check; quarterly new oral-history discoveryPublic historical metadata and reviewed pages within each host policy.Historical identity review; Mentorship and collaboration evidence; Institutional lineage
SEC EDGAR company filingsDaily tracked-company filing metadata; weekly relationship reviewOfficial SEC interfaces under their current fair-access policy with a descriptive contact user agent.IPO and acquisition events; Named executives/directors; Ownership and investment evidence
DevReal canonical identity registryAt every publicationShared public entity databaseReuse reviewed canonical person and organization IDs and retain their existing record links.

Research coverage

Scanned 21,264,134 stored rows across 3,221 local partitions. Each corpus is deduplicated by article ID. The broader NYT metadata index contains 106,907 unreviewed name-shaped candidates; these are not verified people.

This edition contains 284 nodes, 380 documented relationships, 23 tribes, and 142 cited sources. It is not an exhaustive record of technology history. The locally available Eigen Hacks archive ends in September 2026; archive coverage and candidate matches are recorded separately from the curated graph.

Private collection notes and ontology tags are personal annotations. A profile claim is a request for identity review; paying for Pro does not confer control over the historical evidence.

Source register

Responsible access

Proposed collectors use at most 1 active connection per host and a default 10-second interval, refresh robots rules every 24 hours, and stop when policy cannot be checked. Publisher-specific licensed API limits take precedence. No recurring external collection is enabled yet.

Public records are open to browse. Authentication and personal-workspace APIs are excluded by robots.txt and protected by persistent request limits. We do not publish personal contact details or infer sensitive personal traits. To propose a correction, keep the original evidence URL with the record so that changes can be reviewed.