Retrieval fused a bi-encoder and BM25 by reciprocal rank. A bi-encoder embeds a
document long before the question exists, so the two never meet: it is good at
"same topic" and mediocre at "answers this". A cross-encoder reads the pair.
The proxy already serves three — `cohere-rerank-v4.0-pro` is the default and
measurably better than the fast variant. Query text goes exactly where the
embeddings already go, and nothing new was signed up for.
It found a defect nobody was looking for. In AI Mode each finder scored
`1/(1+rank)` *within its own corpus*, so the best article, section, question and
card all scored 1.0 and the shortlist was a meaningless round-robin. A
cross-encoder is the first thing in this system that can compare a question
with a section. Candidates per kind widened so it can select rather than merely
reorder.
Measured against labels neither ranker produced. Questions, 60 disease tags:
precision@3 0.394 → 0.483. Sections, 60 article titles: 0.772 → 0.833.
"Management of bronchiolitis" led with influenza transmission and a pregnancy
question; "when do you image a first febrile seizure" returned the definition
rather than the sentence saying imaging is unnecessary.
And the honest negative, in docs/reranking.md: board vignettes are written
*not* to name their diagnosis, so on "what causes croup" it prefers a question
that says the word in passing over the barking-cough vignette that never says
it. Some of the bi-encoder's strength is traded away.
Not on the typeahead. A page of results is a choice being made and worth a
third of a second; a typeahead is a word being finished, runs on every
keystroke, and has nothing to judge yet.
The three-state thresholds stay on cosine, argued at the constant: a reranker
only ever sees a shortlist and structurally cannot answer the corpus-wide
question those numbers ask, and whether an answer claims to come from the
library is a promise that must not depend on a network hop.
Every failure returns None and leaves the order alone — unconfigured, no proxy,
connect error, bare 502, timeout, non-JSON, a duplicate or out-of-range index,
a non-numeric score, a list the wrong length. Verified against the running site
with a bogus model name: same results, fused order, no error to the reader.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TqXevQJhxFrM7jJg82cgZN
**Prepared sessions.** Most of this existed: unanswered first, weakest topic
next, wrong-before-right after that, all scaled by what share of the real paper
each topic carries. What it could not do was change with time, say anything
about itself, or be reached without filling in a form.
Evidence now decays on a thirty-day half-life. Exponential rather than a fixed
window because memory has a slope, not a cliff — under a window, 29 days counts
fully and 31 counts for nothing — and because it is memoryless, so an answer's
weight does not shift when unrelated questions are answered, which is what lets
the preview stay a valid forecast. Spring is worth an eighth of last week. Two
things decay: a question's recall probability, drifting towards even rather
than past it, so an old right answer becomes eligible rather than wrong; and a
topic's accuracy, against a prior of two "no idea" answers, which fixes "right
once, known forever".
Strict unanswered-first meant that on a bank of 2,900 nothing was ever
recycled — spaced repetition existed and was unreachable. Review now takes up
to two fifths of a session. And the damping that spread the picks across topics
was applied only to seen material, so a learner with no history was handed the
heaviest domain entire instead of a spread; that was live.
The plan is the product. It is computed, shown, and then the session is built
from that plan's own ids and the plan returned with it, so the two cannot
differ; every figure in it is a tally over the chosen questions rather than a
forecast. No model touches the ranking — a learner asking "why these twenty"
has to get the same answer twice.
**Vision.** The proxy's own `/model/info` says which models can see, so nothing
is hard-coded: 77 report yes, 11 no, and 328 say nothing at all, which means
absent rather than incapable — so those are asked once with an 8px PNG and the
refusal cached. The deployment's main model turns out not to see, and questions
carry figures the learner is looking at, so the tutor was answering about an
image it had never been shown. It routes to a configured tool model now, folds
the description back in as text saying plainly where it came from, and caches
on the bytes because the same figure is re-sent every turn.
Also fixed on the way: `article` was missing from the admin's task list, so
article drafting always ran on the fallback model whatever an administrator
chose; and `.jpx` stem images were sent as JPEG because `mimetypes` guesses
that from the name, so the provider rejected them two hops later.
An administrator must pick a tool model in Settings → AI models. Until then the
tutor says a figure exists that nothing could read, rather than describing one
it cannot see.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TqXevQJhxFrM7jJg82cgZN
Five corpora were each searchable from their own page, which meant knowing which
of five pages held the thing you were looking for before you could look for it.
`GET /search` runs them together.
Visibility is never re-implemented here. Questions go through the same bank
predicate and exam scope as the question bank, articles through the same draft
rule, cards through deck ownership, images through library grants. A search page
with its own idea of who may see what is how private content leaks, so the tests
that matter are the boundary ones: a peer's search reaches neither another
user's unshared question nor their deck, and a draft is invisible to everyone
but the educator who wrote it.
A section hit is reported under its article, not beside it — ten sections of one
article are one result with ten places to start reading, not ten results burying
everything else. This is what the section index was backfilled for; each one
links straight to that section.
Results are grouped by kind rather than interleaved by score. A question and an
article are different kinds of answer, and a single ranked list makes you read
every row to work out which kind each one is. Snippets show the window around
the match rather than the opening of the document, because every document's
opening looks the same. A question found only by the semantic ranker says so.
The header box has two ways out: pick a suggestion and go straight to that
article, or press Enter and search everything. Suggestions are lexical and
prefix-first — a typeahead is finishing the word you are typing, and a semantic
neighbour of half a word is noise — and debounced 180ms so typing is not a
request per keystroke. One corpus failing is logged and returned as a gap in the
answer rather than a failed page.
154 backend, 163 frontend green.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XeFQJXJTfHKTfbfsdxv57Z
Daniel