Words and codes used on SocioLens

SocioLens pages use some words and codes that are not obvious. Each entry below gives the word or code as it appears on the site, then what it means for you as a reader.

Many codes come from the project’s earlier way of working and appear mostly in the collapsed provenance notes at the foot of a page. Where the project files do not say what a code means, the entry says so and gives no guess. For how to read a debate, claim or research opportunity, see How to use SocioLens.

What is on the site

Living Debate: One contested question about AI and society. The debate page gives the current answer, the evidence behind it and what is still unknown. A debate can get new versions as the evidence changes, and earlier versions stay readable. All debates.
Claim: A single, sourced statement that a debate rests on. Each claim says what kind of statement it is, how mature the evidence is, and what it does not license.
Source: A paper, report or official document behind a claim. Its page links to the original where a link has been checked, and lists the claims that cite it. All sources.
Concept: A theoretical idea the debates use, such as discretion or epistemic dependence. Its page gives a short SocioLens reading of the idea and the theory texts it draws on. The concept pages are early structural prototypes, not canonical public theory pages. All concepts.
Research opportunity: A research question that follows from what a debate does not yet settle, with the claims that motivate it and a possible design. It is a hypothesis drafted with AI, screened against the literature by AI web search, not a finding. All research opportunities.
Proposal pack: One printable page per research opportunity, for a supervisor or funder, with the references as BibTeX. It gathers what the opportunity page already holds; nothing in it is new.
Proposal outline: A worked example of how a research opportunity could become a research project: questions, hypotheses, design and risks. Unless the page says otherwise, it was drafted by AI. It is a proposal, not a finding and not a registered protocol.
Radar update: A dated note of new evidence that bears on a debate. It links to the claims it bears on. It never adds a second claim, and it never changes a claim’s standing without saying so.
Open questions: What each debate says we still do not know, gathered on one page. An open question is not a verified gap in the literature; that is checked for each research opportunity. Open questions.
Field Map: The project’s map of the field: its domains, tensions, lenses and planned debates. It is kept unchanged as part of the record (version v0.2, 25 September 2026). It names 17 debates; those that are published link to their pages. Field Map.
Start Here: The founding statement of SocioLens, written while the project was planned. Like the Field Map it is kept unchanged, so some details describe plans rather than the site today.

Versions and the claim record

Version numbers (v0.1, v0.2, v1.0, v1.1, v1, v2, v3): Each version is one fixed text of a debate, claim, source, research opportunity or outline; a higher number is a later version. Debates and outlines are numbered like v0.1 or v0.2, claims, sources and radar updates like v1.0 or v1.1, and research opportunities v1, v2, v3. A source’s version is the version of the SocioLens record, not the edition of the publication.
Current version: The newest version. The plain address of a page, without a version number, always shows it. Earlier versions stay available, and their records never change.
Pinned, version-pinned: A pinned page or link points to one fixed version, for example /claims/ld05-hq-c05/v1.0/. Its record never changes, even when a newer version appears; the site may add a note to the page or withhold a figure not cleared for public use. A debate pins the claim versions it cites, and a claim pins the source versions it cites.
--v1.0 after an ID: Names one version of a record, for example LD05-HQ-C05--v1.0 is version 1.0 of claim LD05-HQ-C05. A ClaimVersion is one such version of a claim.
Superseded, supersedes: A version is superseded when a newer version replaces it. The older version stays available with its record unchanged, and the newer one says what changed and why. This is not the same as the verification state "Superseded by newer evidence" below.
LOCKED, PUBLISHED (in a debate’s status line): The debate version was closed (locked) when it was finalised. It is not edited afterwards; a change comes as a new version. The first versions of the six debates of September 2026 say LOCKED; versions written since say PUBLISHED.
Evidence cut: The date up to which a debate’s evidence was gathered. Work published later is not covered until the debate is reviewed or a radar update records it. Where a debate states no cut, the page says so; the date it was last updated is not a substitute.
Does not license: Conclusions the evidence does not support, even though they are often drawn from it (field 16 of a claim). Read it before citing the claim.
What would change this claim: The finding that would overturn or narrow the claim (field 23 of a claim).
The 25 canonical claim fields, field 6, field 22: Every claim is recorded in the same 25 numbered fields, for example field 6 for its source, field 16 for what it does not license and field 22 for its Field Map codes. The table at the foot of each claim page shows all of them, exactly as recorded.
FACT, ESTIMATE, INTERPRETATION, UNKNOWN: The kind of statement. FACT is directly supported by a cited source; ESTIMATE is a number or range inferred from sources, with uncertainty; INTERPRETATION is SocioLens’s reading of the evidence; UNKNOWN is not established by any source found.
Maturity: emerging, rapidly maturing, mature, contested: How settled the body of evidence is, from emerging (few studies) through rapidly maturing to mature. Contested means credible studies disagree.
Research strand (field 4), for example A or C/F: The line of research inside a debate that a claim belongs to, for example B, Data and access. Each debate letters its own strands, so B in one debate is not B in another. Where only a letter is shown, the claim record gives no name for it.
Publication status: ELIGIBLE, ELIGIBLE (QUALIFIED): Field 24 of a claim: whether it was cleared for publication. A word in brackets, such as ESTIMAND or PREPRINT, is a qualifier recorded with that clearance.
HOLD: Held back. A figure or detail marked HOLD was not cleared for public use, so this site does not give it. Where a recorded line names such a figure, the page shows [figure withheld] in its place; the record itself keeps it. Once the figure is checked, a new version of the claim gives it and says what it was checked against.
Visibility: PUBLIC, PUBLIC_REFERENCE: Whether a record may appear on the public site. Only PUBLIC records, and PUBLIC_REFERENCE sources that a public claim or concept cites, are published; internal and held-back material never is.
Claim Value Vocabulary: The versioned list of standard values used for some claim fields. The claim table shows each field as written and, beside it, its standard value (for example CONTENT VERIFIED becomes CONTENT_VERIFIED_PRIMARY).

How far a claim was checked

Each claim page says in words how far the claim was checked against its source. The claim’s field table shows the same state as a code. Claim checks so far were AI-led; no person has checked claims against their sources.

CONTENT_VERIFIED_PRIMARY: Checked against the primary source.
BIBLIOGRAPHIC_RECORD_VERIFIED: The publication was confirmed to exist, but its content was not checked. No claim on the site has this state at present.
RELAYED_ORIGINAL_NOT_VERIFIED: Taken from a report about the original, which was not checked.
UNVERIFIED_LEAD: Not verified: a lead only.
CONTESTED: Contested.
SUPERSEDED: Superseded by newer evidence. No claim on the site has this state at present.
SUPERSEDED_WITHOUT_REPLACEMENT_ESTIMATE: Superseded, with no replacement estimate yet. No claim on the site has this state at present.
CONCERN_FLAGGED_OR_RETRACTED: A concern, correction or retraction is recorded for its source.
(ABSTRACT_ONLY) after a state: Checked against the abstract of the primary source only.
(PREPRINT) after a state: The source is a preprint.
(HQ_ADJUDICATED) after a state: The claim’s ledger marked it HQ ADJUDICATED: the earlier HQ review ruled on it (see HQ below). The claim vocabulary requires such a claim to keep a trace to the claims it rests on.
Other codes in brackets after a state: Codes such as SOURCE_SPECIFIC or PARTIAL_VERDICT are qualifiers from the Claim Value Vocabulary. They are shown as recorded; this page does not yet explain each one.

Labels on research opportunities

No study found: The literature screen found no study that does the core design. That does not mean none exists; check the queries and run your own search.
Partly addressed: Related studies cover parts of the question. The page sets out what they leave open.
First-pass screen only, checked in depth: A first-pass screen is the AI web search done before the opportunity was published. Opportunities checked in depth were searched again after publication; those marked first-pass screen only have not been.

IDs you will see

IDs name one object exactly, so they are safe to cite. Letters stand for digits below: nn is a two-digit number.

LD-nn: A Living Debate, numbered as in the Field Map (LD-01 to LD-17). Inside other IDs the hyphen is dropped: LD05 is LD-05. Field Map.
LDnn-HQ-Cnn: A claim from one of the six debates of September 2026 (LD-02, LD-05, LD-07, LD-10, LD-11, LD-12): claim number nn in that debate’s locked claim ledger, from the earlier HQ process (see HQ below). Its later versions keep the same ID.
LDnn-SL-Cnn: A claim written by the SocioLens project in its own files since 14 October 2026, for a new debate or a new debate version. SL stands for SocioLens.
Cnn on its own, for example C07: Claim number nn of the debate being discussed, in an outline or a note on a claim. In an outline it links to the claim version the outline was drafted from.
SRC-…: A Source. The rest of the ID usually names the author or institution and the year. The part right after SRC- says where the record came from (see the next four entries).
SRC-LDnn-…: A source recorded for one of the six debates of September 2026, named after that debate.
SRC-SL-…: A source added by the SocioLens project since 14 October 2026, for the debates and debate versions written since then.
SRC-Rnn-…, SRC-Rnn-HQ-…: A source recorded during research build Rnn (see R03 to R08). SRC-Rnn-HQ-… was recorded by the earlier HQ review: some are its own syntheses (source class SOCIOLENS_SYNTHESIS), not outside publications, and others group several outside documents under one ID (MULTIPLE_SOURCE_TYPES). The source page shows which.
SRC-THEORY-…: A theory text that a concept page draws on, for example a book by Foucault or Lipsky.
RO-LDnn-NN: Research opportunity number NN, listed under debate LD-nn. An opportunity that draws on several debates lives under one of them and appears on the others too.
RADAR-…: A radar update. The rest of the ID names the debate it bears on and its subject, for example RADAR-LD07-PEW-SYNTHETIC-SURVEYS-2026.
THEORY-…: A concept, for example THEORY-DISCRETION.
S1, S2 on a concept page: The concept’s numbered senses: distinct meanings of the same term, each with its own sources. Elsewhere, in notes on where a source record came from, S1 or S2 is an item number in a recorded list.
SC-LDnn-…: A stage, level or state on one of a debate’s own ordered scales (its analytic scaffold), used to place a claim within the debate, for example SC-LD02-T01 EXPOSURE.

Abbreviations

HQ, Central HQ, factory-hq: The review and decision function of the project’s earlier way of working. In September and October 2026 it audited the research, ruled on claims and recorded decisions. That way of working has been replaced; its records are kept as history and decide nothing now. The label HQ does not mean that a person checked a claim.
SL: SocioLens, in IDs such as LD04-SL-C01 or SRC-SL-…: written by the SocioLens project in its own files, not in the earlier HQ process.
R01: An early research stage of the project, cited in Field Map v0.1. Field Map v0.2 (25 September 2026) calls itself the first revision after R01. Field 22 of a claim, headed R01 / FIELD-MAP LINKS, lists the claim’s Field Map codes, such as D5 or L-B.
R03 to R08: The six research builds behind the debates of September 2026, one per debate: R03 for LD-07, R04 for LD-05, R05 for LD-02, R06 for LD-11, R07 for LD-10 and R08 for LD-12. Each ended in a locked claim ledger; their research was frozen on 25 September 2026.
P1 to P5: The positions a debate compares, numbered in its text. Each debate numbers its own positions, so P1 in one debate is not P1 in another. The full text of the debate describes each one.
D1 to D9, D4a, D6.1: The nine domains of the Field Map, for example D2 Organizations, work and the labour process. D4a, D4b and D6.1 to D6.4 are sub-areas of D4 and D6. Field Map.
T1 to T7: The seven tensions of the Field Map that run through the whole field, for example T4 What kind of thing? Field Map.
L-A to L-C, R-1, R-2: The Field Map’s lenses, applied across all domains: L-A distribution and stratification, L-B geography and external validity, L-C AI in use. R-1 (evidence architecture, the evidence rules for every claim) and R-2 (conceptual translation) are listed with them. R-1 is not R01. Field Map.
T01: The label of the document the six concept pages come from, HQ Concept Prototypes v0.1. It is not one of the Field Map tensions T1 to T7. Inside an ID such as SC-LD02-T01, T01 is a stage of that debate’s scaffold (see SC-LDnn-…).
RQ1, H1 in a proposal outline: Research question 1 and hypothesis 1 of that outline.
Claude (SocioLens Project Lead): Claude, an AI model made by Anthropic, which has acted as the project’s lead for its owner since 12 October 2026. Where a page says Claude drafted or checked something, an AI did it, not a person.

Names in the provenance notes

Collapsed sections at the foot of many pages record where the page’s data came from. They name documents and steps from the project’s earlier way of working. These are kept as history and decide nothing now.

Programme Status & Authority Register, Publication System Amendment, Implementation Architecture Selection, Gate Update, Proof Projection, Proof Source Registry: Planning and decision documents of the earlier way of working. Pages name them to show which rule or list a record followed.
AP-nn, Architecture Proof Protocol: A numbered requirement of the Architecture Proof Protocol, the set of tests the site’s design had to pass in an early engineering stage (the proof spike).
Canonical Claim Ledger: The locked list of a debate’s claims at the end of a research build, for example R07, the LD-10 ledger. LDnn-HQ-Cnn claims come from these ledgers.
Batch 01, Batch 02: One numbered batch of checks by the earlier HQ review within a research build, each on one group of claims and their primary sources.
Q9 Tranche 1 to 6: The six rulings in which the earlier HQ review selected the debates of September 2026 for the site, one debate per tranche, with exactly which claims and sources to publish.
Q9: An internal label from an earlier stage of the project. Its meaning is not recorded in the project files, so it is not explained here.
HQ-ESC-…: The reference of a recorded decision of the earlier HQ review.
HQ-Q6, HQ-Q7: A numbered item of a recorded HQ ruling, for example items HQ-Q1 to HQ-Q8 of the ruling that selected LD-11 for the site.
Issue #nn, PR #nn: An issue or pull request in the project’s GitHub repository, where the change was discussed or made. The repository is not public, so these are references, not links.
Drive ID, SHA-256, content seal, materialized: The technical record of a page’s data. A Drive ID names the stored file in the project’s Google Drive archive. A SHA-256 value is a fingerprint of the exact text: if the text changed, the fingerprint would no longer match and the site would not build. Materialized means turned from that stored text into the records the site is built from.