FA-100036 / Evaluation metric aggregation / Open access

Slice aggregation gives each nonempty slice equal weight · case 01

Slice aggregation gives each nonempty slice equal weight.

Verified by executionVariant 1 · 7 checks per implementationDownload source bundle ↓JSON ↗

ROOT CAUSE

Per-slice scores are averaged equally even though the contract weights examples.

THE FAILURE

Per-slice scores are averaged equally even though the contract weights examples.

Unsuccessful approach: Skipping empty slices still leaves nonempty slice means equally weighted rather than pooling their supports.

Case contract

Slices map names to [correct,total]; return total correct divided by total examples, or 0 when empty.

Why this case matters

A small explicit evaluation contract makes aggregation behavior reproducible across labeled examples.

1 / The failure

Exit 1

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

N = 1
observations = []
def solve(slices):
    return sum(c/t for c,t in slices.values() if t)/len(slices) if slices else 0
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('unequal slice sizes',solve({'a':[N,N],'b':[0,1]}),N/(N+1))
check('fixed unequal slices',solve({'a':[3,4],'b':[0,1]}),.6)
check('perfect slices',solve({'a':[N,N],'b':[2,2]}),1)
check('empty',solve({}),0)
check('zero support',solve({'a':[0,0]}),0)
check('one nonempty',solve({'a':[1,2],'b':[0,0]}),.5)
check('all incorrect',solve({'x':[0,N]}),0)
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
unequal slice sizes	`0.5`	`0.5`	Passed
fixed unequal slices	`0.375`	`0.6`	Failed
perfect slices	`1.0`	`1`	Passed
empty	`0`	`0`	Passed
zero support	`0.0`	`0`	Passed
one nonempty	`0.25`	`0.5`	Failed
all incorrect	`0.0`	`0`	Passed

SHA-256 / 41405c6444df223cba731e7a0ee42e6611abb9ee9a5f006d7d6c1ab1871d2ba8

2 / The unsuccessful fix

Exit 1

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

N = 1
observations = []
def solve(slices):
    vals=[c/t for c,t in slices.values() if t]; return sum(vals)/len(vals) if vals else 0
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('unequal slice sizes',solve({'a':[N,N],'b':[0,1]}),N/(N+1))
check('fixed unequal slices',solve({'a':[3,4],'b':[0,1]}),.6)
check('perfect slices',solve({'a':[N,N],'b':[2,2]}),1)
check('empty',solve({}),0)
check('zero support',solve({'a':[0,0]}),0)
check('one nonempty',solve({'a':[1,2],'b':[0,0]}),.5)
check('all incorrect',solve({'x':[0,N]}),0)
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
unequal slice sizes	`0.5`	`0.5`	Passed
fixed unequal slices	`0.375`	`0.6`	Failed
perfect slices	`1.0`	`1`	Passed
empty	`0`	`0`	Passed
zero support	`0`	`0`	Passed
one nonempty	`0.5`	`0.5`	Passed
all incorrect	`0.0`	`0`	Passed

SHA-256 / 9dc1ea9139512dd7574271f7ed01d3deb24136e4d25c078f9cd18f9099ffde8e

HELD IN THE MEMBER ARCHIVE

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

This mechanism has 7 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 labels and scores; this defines no benchmark protocol, production metric library, or model quality claim. 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 2026-09-30T16:48:30.840869+00:00.

Case digest / 358ad4535ea5acc78213991548acc6eeec46d73f7a024467196299d9e23e7637