FA-100571 / Mixture of experts dispatch / Open access

Second expert is omitted from a top-two token dispatch · case 01

Second expert is omitted from a top-two token dispatch.

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

ROOT CAUSE

Only the winning route is retained even when the contract requests two experts.

THE FAILURE

Only the winning route is retained even when the contract requests two experts.

Unsuccessful approach: Taking the first two input routes can omit the runner-up after score reordering.

Case contract

Return up to two (expert,weight) routes sorted by descending weight then expert ID.

Why this case matters

A deterministic toy router makes this dispatch decision independently inspectable.

1 / The failure

Exit 1

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

N = 1
observations = []
def solve(routes):
    return sorted(routes,key=lambda x:(-x[1],x[0]))[:1]
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
rows=[('z',N),('a',N+1),('b',N-1)]
check('ranked pair',solve(rows),[('a',N+1),('z',N)])
check('two only',solve(rows[:2]),[('a',N+1),('z',N)])
check('singleton',solve([('q',N)]),[('q',N)])
check('empty',solve([]),[])
check('tie order',solve([('z',N),('a',N)]),[('a',N),('z',N)])
check('negative',solve([('x',-N),('y',-N-1)]),[('x',-N),('y',-N-1)])
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
ranked pair	`[['a', 2]]`	`[['a', 2], ['z', 1]]`	Failed
two only	`[['a', 2]]`	`[['a', 2], ['z', 1]]`	Failed
singleton	`[['q', 1]]`	`[['q', 1]]`	Passed
empty	`[]`	`[]`	Passed
tie order	`[['a', 1]]`	`[['a', 1], ['z', 1]]`	Failed
negative	`[['x', -1]]`	`[['x', -1], ['y', -2]]`	Failed

SHA-256 / 8dccbf84fdbe08d60287addfe3ca9e92b135710311cd195192b079312aaf4a5e

2 / The unsuccessful fix

Exit 1

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

N = 1
observations = []
def solve(routes):
    return routes[:2]
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
rows=[('z',N),('a',N+1),('b',N-1)]
check('ranked pair',solve(rows),[('a',N+1),('z',N)])
check('two only',solve(rows[:2]),[('a',N+1),('z',N)])
check('singleton',solve([('q',N)]),[('q',N)])
check('empty',solve([]),[])
check('tie order',solve([('z',N),('a',N)]),[('a',N),('z',N)])
check('negative',solve([('x',-N),('y',-N-1)]),[('x',-N),('y',-N-1)])
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
ranked pair	`[['z', 1], ['a', 2]]`	`[['a', 2], ['z', 1]]`	Failed
two only	`[['z', 1], ['a', 2]]`	`[['a', 2], ['z', 1]]`	Failed
singleton	`[['q', 1]]`	`[['q', 1]]`	Passed
empty	`[]`	`[]`	Passed
tie order	`[['z', 1], ['a', 1]]`	`[['a', 1], ['z', 1]]`	Failed
negative	`[['x', -1], ['y', -2]]`	`[['x', -1], ['y', -2]]`	Passed

SHA-256 / 52d11140bdf93ca6183a0e028ba7998c820d22f7c0dc41ef8985a1660d6ce013

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 finite toy scores and routes; the contract does not model a production model, accelerator, or learned routing quality. 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 2026-09-30T16:48:37.194377+00:00.

Case digest / 4a06c52398507822036e2131079cf7a793667ff1143ca73f08e6c0c1f901d740