The research library

Articles where every claim links to a source.

A collection of research articles published on this site. Each claim in an article links to the source it comes from, or is marked as having no source. Every earlier version of an article is kept, and a reader can dispute any single claim on the page where it appears.

1,211articles in this system’s research libraryEach article is a web page for people and, at the same time, a data file an AI model can read.

7,282claims across all the articlesA claim is one sentence that states a fact which can be checked against a source.

4,803sources the claims link toWhen a claim has no source, the article marks it “no source” next to the sentence.

Live counts, read from this site when this page loaded

Search the articles What one article contains

How an article is structured

The six parts of one article in the library.

Each article is made of claims, the source each claim links to, a “no source” label on claims without one, every earlier version, a way to dispute a claim, and a copy of the same article as JSON (a structured text format that programs and AI models read). Tap a step to highlight that part in the example. The switch sets how much detail the explanation gives.

The research library Articles with sources

Articles where every claim links to a source.

A library of research articles. Every claim in an article links to its source, or is marked as having no source. Every change to an article is kept.

Any claim can be disputed on the page where it appears. An AI model can read the same article as structured data.

Each article is JSON: its claims, their sources and its revisions. The library is stored like everything else in the system.

Example article3 claims · 2 sources

How this library records its claims

This library holds 1,211 articles.1 Every article is also served as structured data.2 Readers trust a claim more when its source is one tap away.no source

1This site’s public index of articles, read when this page loaded

2The same article as JSON, at /api/articles/<article-name>

4 versions, all keptDispute this claim

{"slug": "how-this-library-keeps-its-claims",
 "body": "This library holds … [s1] … [s2] …",
 "meta": {"claim_n": 3, "sources": [
   {"id": "s1", "quote": "…", "prev": "genesis", "hash": "…"},
   {"id": "s2", "quote": "…", "prev": "s1’s hash", "hash": "…"}]},
 "hash": "sha256:…"}

An illustration of one article’s parts, in the same format the site uses. The third claim is labelled “no source” because nothing on this site supports it. In the data, each source carries a SHA-256 hash (a fingerprint computed from its content, which changes if the content changes) of the source before it.

What is true of every article

Four things every article in the library does.

Each claim links to a source or is marked as having none. Every change is saved as a new version. Any claim can be disputed on its page. An AI model can read the same article as JSON.

1Sources

Every claim links to its source, or is marked as having none.

A claim either links to the page, document or data it comes from, or the article shows a “no source” label next to it. A reader can see which statements are backed and which are not.

This library holds 1,211 articles.source 1

Every article is also served as structured data.source 2

Readers trust a claim more when its source is one tap away.no source

2Versions

Every change to an article is saved as a new version.

Editing an article adds a new version; it never replaces the old one. The earlier wording stays available next to the new wording.

writtensource addedclaim disputedclaim corrected

3Disputes

Any claim can be disputed on the page where it appears.

A reader who thinks one sentence is wrong can dispute that sentence directly on the article page. The dispute, and the answer to it, are stored with the claim.

Readers trust a claim more when its source is one tap away.

Dispute this claimopen · waiting for a source

4For AI models

An AI model can read the same article as JSON.

The page a person reads and the data an AI model reads come from the same stored article, so both contain the same claims, sources and versions. Programs fetch it from this site’s API (web addresses built for programs to request data, rather than pages for people).

GET https://miscsubjects.com/api/articles/what-the-build-can-do-on-your-mac-iphone-and-messages

→ {"slug", "title", "body",
   "meta": {"claim_n", "sources": [{"url", "quote", "prev", "hash"}, …]},
   "hash"}

The newest article, as an AI model or a program receives it. Each source includes the SHA-256 hash (fingerprint) of the source before it, so a changed source is detectable.

Every article

Search the articles in the library.

3 claims

This page can search 1,159 of the 1,211 articles in the library. Each card opens the article on this site.

Other parts of the system

Four other parts of the system.

The research library is one of the system’s ten parts. Each card below opens the page that explains another part.

Text a question your business needs researched.

Send the question by text to the number below. The answer is written the same way as the library: every claim in it links to its source, or says it has none.

Open in Messages +1 (424) 504-0908