Measured on the operator's library (26,005 tracks, pool scale 2.28): the
arms the library scale grows all have far more rows than it asks for, and
two arms collapsed whatever their limits were.
- taste_overlap ordered by weight then track id, so the strongest taste
artist's catalogue filled it: 20 tracks of saib. (197 tracks) out of 174
taste artists, the same lowest-id tracks every time. It now takes one
track per taste artist, strongest first, md5-seeded, as similar_artists
does since #5297.
- tag_overlap broke ties in no order. A one-tag seed ties 7,804 tracks,
and the LIMIT took whichever Postgres reached first, which the order
seed did not control. Ties now follow md5(id || seed).
- library_scale.go records the per-arm measurement beside the reference
size it leaves as is.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
LB's similarity scores are counts (16 to 1445 for recordings, to 10,875
for artists), and they reached a scorer that reads similarity as 0 to 1.
One edge added 64 to 5,800 points under Songs-like, so likes, skips and
recency stopped mattering for any candidate with an edge.
- ResolveListenBrainzTrackEdges divides each edge by the seed's best
score in its whole cached answer. The cache keeps the raw answer.
- The worker stores matched and unmatched artist rows over the seed's
top artist.
- Migration 0096 rescales the rows already stored.
- The candidate arms hold sim_score bands so tier 1 stays above tier 2:
lb_similar 0.5 + 0.5 x edge, tag_overlap halved, likes_overlap 0.3.
Otherwise a seed's weaker LB edges fell below a shared genre tag.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The similar_artists and coplay_artists arms ordered by artist score, so the
closest related artist's catalogue filled the whole LIMIT. On the operator's
library the 30-row similar_artists arm held exactly one artist for all 17
seeds measured (#3879), though each seed had 7-33 similar artists in the
library. Both arms now rank tracks within each artist and take every
artist's first track, best artist first, before anyone's second.
Both also skip missing tracks: the outer select already dropped them, but
only after they had taken places in the arm's LIMIT.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Four arms of the candidate query ended in a bare `ORDER BY random()` with no
seed: similar_artists, likes_overlap, coplay_artists and random_fill.
Such an arm returns a STABLE set only while its LIMIT exceeds the rows
eligible for it — at that point it returns all of them and the order stops
mattering, because scoreAndSortCandidates sorts by track id before drawing
jitter. Below that threshold it returns a random SUBSET, and two builds on
the same day draw different ones.
So daily determinism held BY ACCIDENT, and only for libraries smaller than
the limits. Any real library is larger, which means same-day rebuilds have
been producing different mixes since those arms were written — invisible,
because a mix that changes after a refresh looks like a feature rather than
a broken promise.
Found by breaking it: cutting RandomFill to 10 while tuning Songs-like
turned TestBuildSystemPlaylists_DailyNonceDeterminism red. That test seeds
~20 tracks against a default RandomFill of 30, so its determinism came from
the limit exceeding the library, not from the code being right. It is now a
real guard.
The arms order by md5(id || $12) instead. The CALLER decides what that
means, which is the point: system mixes pass a per-(user, day) seed and get
the determinism they promise, radio passes a fresh value per request and
keeps varying, which is what a radio should do. Same shape the browse
queries in this file already use (`md5(id::text || current_date::text)`) —
existing idiom, not a new one.
This also unblocks the trim that #3881 wanted and could not have. Shrinking
a randomly-ordered arm was what broke membership; a seeded one takes a
smaller but REPRODUCIBLE slice. Songs-like's seed-independent share drops
from 29% to 12%, which was the original intent before determinism forced it
back to 20%.
TestSongsLikeLimits_DoNotShrinkTheUnseededRandomArms is DELETED rather than
kept passing. It existed to stop anyone trimming those arms while the
ordering was broken; the ordering is fixed, so the constraint is gone and a
guard enforcing it would now forbid correct code.
Was filed as blocked on tooling. It was not: `make generate-go` runs sqlc as
a pinned Go tool and is the same path CI takes.
One thing worth knowing for next time: three files in internal/db/dbq are
owned by root, left by `make generate` running sqlc in Docker. sqlc errored
on the first it could not write. They are untouched by this change and the
regeneration of recommendation.sql.go completed, but `make generate` will
keep failing until they are chowned.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01SQ31KQpYbStyK5y58UmPLH
2a re-ranks the existing pool by TasteMatch; this ensures taste-relevant tracks
ARE in the pool. Adds a 6th arm to LoadRadioCandidatesV2: in-library tracks by
the user's top positively-weighted taste-profile artists ($10 K, weight > 0,
deterministic weight-DESC,id order so it doesn't reintroduce same-day
nondeterminism). Pool-inclusion only (sim_score 0) — TasteMatch already scores
the fit. Empty for cold-start users (no profile).
- CandidateSourceLimits.TasteOverlap; default 20 (radio), 80 for For-You via
systemForYouSourceLimits.
- You-might-like deliberately sets TasteOverlap=0: it surfaces NOT-actively-
engaged artists, so flooding its pool with top-taste (mostly already-played)
artists would just feed the read-time dedup.
- Test: positive-weight artist's track enters via the arm; negative-weight one
is excluded (weight > 0). Existing pool tests unaffected (no profile seeded).
Deferred within 2b: profile-seeded For-You — marginal given the arm + TasteMatch
already inject taste broadly (top-played seed ≈ top-taste artist).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The combined seed_info CTE filtered the seed out when its genre was
NULL/empty. Restored the spec's two-CTE design (seed_artist + seed_tags)
so similar-artist matching works regardless of the seed's genre coverage.
Drops the genre workaround from the SimilarArtist test.
bvandeusenandClaude Opus 5.5