Context
Deferred was blurbed “in the corpus, not yet read closely enough to
place”. Two situations shared the word:
- Nobody has tried. These are works seeded from a feed, an abstract or a bibliography, which wait in the queue. On 2026-10-13, 68 of them came from the owner’s papers-feed alone.
- Someone tried, and no lawful copy was reached. The status note says where the reader looked: OpenAlex, Unpaywall, the publisher, author pages and repositories. It also says why each failed. Many of these notes end “It stays Deferred until a copy is supplied.”
The two call for different actions. The first is reading work any session can pick up. The second is blocked until the owner supplies a copy, through library access, a request to the authors or a browser that passes a bot check. Until now a reader could only tell them apart by opening each note. The owner asked for the difference to be a status: “differentiate ‘looked for a legal copy but couldn’t find one’ from ‘we haven’t even attempted reading this yet’”.
Luria allows this. A scheme’s status vocabulary is the project’s own: the
QUESTION scheme already uses Open, Answered and Dissolved (ADR-032). Any
status other than the scheme’s active one counts as not in force. So a new
word is treated exactly as Deferred is in the inactive-citation report.
Decision
- A new LIT status,
Unavailable, labelled “No copy reached”. It means a lawful full text was looked for and none could be obtained. That covers two cases: none was found, or one was found that this client could not fetch. The status note says where and when the reader looked. LikeDeferred, it is not a judgement on merit. A supplied copy returns the work to reading, and the reading sets its status. Deferrednarrows to a work with no failed search recorded: nobody has tried yet, or reading stopped at the abstract or a skim. Its label stays “Unreviewed”.- Both cases of
Unavailabletake the one word. A copy that exists but could not be fetched is a different blocker from a copy that does not exist. In both, though, the owner is what unblocks it. The note says which case it is (for example LIT-171’s “failed on access, not availability”). - Where the line falls. A note counts as recording a search if it names
at least one place a lawful copy was looked for, even only the publisher.
Three notes record no search, and those stay
Deferred:
Migration
All 345 Deferred LITs were classified by script from their status notes,
on 2026-10-13, and the borderline ones were read by eye. 177 moved:
- 169 record that no lawful copy was found. LIT-475 is one.
- 8 record a lawful copy this client could not fetch. These are LIT-137, LIT-171, LIT-180, LIT-181, LIT-186, LIT-387, LIT-536 and LIT-636. They are the quickest for the owner to clear by hand.
Each merged entry gets a version bump. Its history note gives the reason,
and says that prose written before the change may still say Deferred,
which then covered both cases. Two kinds of present-tense phrase were
changed in place: “stays Deferred” and “Deferred because no …” (or ”…
it could not …”). Prose that gives non-reading as the reason (“Deferred
because it was not read”) was left as written. The four entries filed
today and not yet merged were changed in place without a bump.
inactive-ok annotations elsewhere that cite a moved work still apply,
because Unavailable is not in force either. Where their reasons say
“Deferred”, they were true when written and are left alone.
Alternatives
- Keep
Deferredand rely on the status note. That is the state the owner asked to leave. The difference can only be found by reading each note. - A tag instead of a status. The LIT tag vocabulary is topics plus one curation flag (ADR-005). Copy availability is not a topic, and a flag would sit beside a status that contradicts it.
- Two new words, one for “none exists” and one for “exists, not fetched”. That is more precise. But the owner’s request named one distinction, both cases are cleared the same way, and eight entries do not justify a sixth status.
- Name it
Unobtainable. That overstates the case: a copy that exists behind a bot check can be obtained by hand.
Consequences
- The LIT scheme differs from the anthology’s. ADR-001 copied the
anthology’s LIT, NOTE and THEORY schemes so that a work could move between
the records as a file. An
UnavailableLIT moved to the anthology would fail its status vocabulary until it is set back toDeferredor the anthology adopts the word. The anthology may want the same split, since it has the same two populations. - The pending-decisions report no longer counts
UnavailableLITs. Luria countsProposedandDeferredas undecided. AnUnavailablework is blocked rather than undecided, so dropping out of that queue is the intended effect. - The rule for new filings. When a reader looks for a lawful copy and
fails, the LIT becomes
Unavailableand the note says where the reader looked. A LIT filed without a search staysDeferred. The LIT template says so.
Note of 2026-10-25: amended by ADR-053
ADR-053, accepted by the owner on 2026-10-25, revisits two parts of this
decision: the bullet “Both cases of Unavailable take the one word”, and
the alternative that rejected two words because “eight entries do not
justify a sixth status”. A triage of the Unavailable notes found 43 in the
second case, and the two cases are not cleared alike: a located free copy
takes the owner a browser and a minute, a paywalled one a library. A LIT
whose lawful free copy was located at an exact address, and which this
client could not fetch, is now Unfetched. Everything else here stands,
including the line between Deferred and Unavailable. Of the eight
entries named under Migration, seven moved to Unfetched. LIT-137 did not,
since only one chapter of that volume has a free copy.