FA-99466 / Vector retrieval ranking / Open access

Vector retrieval: A later retrieval channel overwrites earlier candidates · case 01

A later retrieval channel overwrites earlier candidates.

Verified by executionVariant 1 · 6 checks per implementationDownload source bundle ↓JSON ↗

ROOT CAUSE

The right candidate source replaces the left source, or concatenation preserves duplicate hits.

THE FAILURE

The right candidate source replaces the left source, or concatenation preserves duplicate hits.

Unsuccessful approach: Concatenating and sorting leaves duplicate identities in the result.

Case contract

Return unique IDs from both channel lists, ordered by descending score then ID; repeated IDs keep their greatest score.

Why this case matters

An offline vector retrieval toy model isolates one ranking contract from embedding generation, storage, and serving.

1 / The failure

Exit 1

"""Failure Map reference implementation. Python standard library only."""
import json

N = 1
observations = []
def solve(left, right):
    return [x[0] for x in sorted(right,key=lambda x:(-x[1],x[0]))]
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('both channels',solve([['a',N]], [['b',N+1]]),['b','a'])
check('duplicate collapse',solve([['x',1]], [['x',N+2]]),['x'])
check('best duplicate score',solve([['x',N]], [['x',N+2],['y',N+1]]),['x','y'])
check('left empty',solve([], [['b',N],['a',N+1]]),['a','b'])
check('right empty',solve([['b',N],['a',N+1]], []),['a','b'])
check('both empty',solve([], []),[])
print(json.dumps({"observations": observations, "passed": all(x["passed"] for x in observations)}, ensure_ascii=False))
raise SystemExit(0 if all(x["passed"] for x in observations) else 1)

Boundary fixture	Actual	Expected	Outcome
both channels	`['b']`	`['b', 'a']`	Failed
duplicate collapse	`['x']`	`['x']`	Passed
best duplicate score	`['x', 'y']`	`['x', 'y']`	Passed
left empty	`['a', 'b']`	`['a', 'b']`	Passed
right empty	`[]`	`['a', 'b']`	Failed
both empty	`[]`	`[]`	Passed

SHA-256 / 76ac37b43f36f072adb7a1362c655df6aba19a6e98306873baf2b42edb11b8c4

2 / The unsuccessful fix

Exit 1

"""Failure Map reference implementation. Python standard library only."""
import json

N = 1
observations = []
def solve(left, right):
    return [x[0] for x in sorted(left+right,key=lambda x:(-x[1],x[0]))]
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('both channels',solve([['a',N]], [['b',N+1]]),['b','a'])
check('duplicate collapse',solve([['x',1]], [['x',N+2]]),['x'])
check('best duplicate score',solve([['x',N]], [['x',N+2],['y',N+1]]),['x','y'])
check('left empty',solve([], [['b',N],['a',N+1]]),['a','b'])
check('right empty',solve([['b',N],['a',N+1]], []),['a','b'])
check('both empty',solve([], []),[])
print(json.dumps({"observations": observations, "passed": all(x["passed"] for x in observations)}, ensure_ascii=False))
raise SystemExit(0 if all(x["passed"] for x in observations) else 1)

Boundary fixture	Actual	Expected	Outcome
both channels	`['b', 'a']`	`['b', 'a']`	Passed
duplicate collapse	`['x', 'x']`	`['x']`	Failed
best duplicate score	`['x', 'y', 'x']`	`['x', 'y']`	Failed
left empty	`['a', 'b']`	`['a', 'b']`	Passed
right empty	`['a', 'b']`	`['a', 'b']`	Passed
both empty	`[]`	`[]`	Passed

SHA-256 / b3f6cf4591dcaceff3a841f313e361bef463acb7fbd5ad5c4e84f7d9542ffa1a

HELD IN THE MEMBER ARCHIVE

The verified repair and its recorded checks are member-only.

This mechanism has 6 recorded checks per implementation. The open-access tier publishes the failure and the unsuccessful fix; the repaired source that passes every check, and the observations that prove it, are available to members.

Every case sharing this mechanism uses the same contract and the same repair, so this one record is held back for all of them.

Member access is invitation-based. Sign in with your invited account to inspect the repair.

Verification & scope

Inputs are small exact toy vectors and scores. No production index, ANN recall, embedding quality, or latency claim is implied. This reproducer isolates one failure mechanism. Results cover the supplied fixtures. Variants within a family share a test contract and should remain grouped when constructing evaluation splits. Related mechanisms with a shared evaluation_group must also remain together; these controlled models are not independent production incidents.

Observations recorded using Python 3.12.14 at 2026-09-30T16:48:24.474259+00:00.

Case digest / eb96ef34703fe7b60d4100e4049ca87964bc5a9bef135253559b525a29749836