Q48, run 2026-11-04-2053-sonnet

Why is W70 not one of the University of Albury's research works, though its author lists the Tessel Laboratory?

Tests: Local rules: the last-listed affiliation, and affiliated institutes.

Expected: Tomasz Wierzba lists the Tessel Laboratory first and the Halloran Institute second; the house rule credits the last-listed one, Halloran. Halloran is an affiliated institute (AFF): its parent link is an agreement, so its works are not the University of Albury's.

❌ R1 csv run 1 graded by llm

The premise is false: W70 is a University of Albury work. A08's I04 affiliation rolls up through I03 to I02, and the I06 affiliation also rolls up to I02.

Why: Rejects the premise, whereas the key says only the last-listed affiliation counts, so W70 goes to the Halloran Institute, an affiliated institute that doesn't roll up to Albury.

❌ R1 csv run 2 graded by llm

The premise is wrong: W70 is one of the University of Albury's works. I04 rolls up through I03 to I02, so W70 counts for I02 along with I06.

Why: Rejects the premise and says W70 counts for the University of Albury; checklist says only the last-listed affiliation counts (Halloran Institute), an affiliated institute that doesn't roll up.

❌ R1 csv run 3 graded by llm

The premise doesn't hold. I04 → I03 → I02, so W70 is one of the University of Albury's works. It is also credited to I06.

Why: Asserts the opposite: claims W70 does count for Albury, whereas the key says only the last-listed affiliation counts (Halloran Institute, an affiliated institute that doesn't roll up).

❌ R1 csv run 4 graded by llm

The premise is wrong. I04 → I03 → I02, so W70 is one of the University of Albury's research works, and I counted it in Q9.

Why: Rejects the premise and says W70 counts for Albury; the key says only the last-listed affiliation counts, so W70 goes to the Halloran Institute, which is affiliated and does not roll up to Albury.

❌ R1 csv run 5 graded by llm

The premise is wrong. I04 sits under I03, which sits under I02, so W70 is one of the University of Albury's works and is included in my count of 12.

Assumptions: research works are ART, REV and PPR, with preprints deduplicated against their published versions. EDT, ERR, PTX and SUP are excluded, and so are STG (staging) records.

Why: Rejects the premise and says W70 counts for Albury; the key says only the last-listed affiliation counts, crediting the Halloran Institute, which is an affiliated institute and does not roll up to Albury.

🟡 R10 executed run 1 graded by llm

W70's only credit (A08) is to I06, the Halloran Institute, not the Tessel Laboratory. Per institution_units, I06 counts only toward itself and not toward I02. The Tessel credit (I04) is on W71 and W72 instead.

Why: Says W70 is credited to the Halloran Institute (last-listed affiliation) and that I06 counts only toward itself, not the University of Albury; but doesn't state that it's an affiliated institute rather than a unit, nor the 'last listed affiliation' rule explicitly

🟡 R10 executed run 2 graded by llm

The credit for W70 is to I06, not I04. The data does not show it listing the Tessel Laboratory. I06 is an affiliate whose only roll-up is to itself in institution_units, so its works do not count towards I02.

Why: Correctly says the affiliate doesn't roll up to the University, and credits W70 to I06. But it never gives the last-listed-affiliation rule and wrongly denies that the author lists the Tessel Laboratory.

🟡 R10 executed run 3 graded by llm

W70's only credit is to I06, which counts only towards itself and not towards I02. The credit for A08 is to I06 (Halloran Institute), not I04. A08 is credited to I04 only on W71 and W72.

Why: Says W70 is credited only to I06 (Halloran Institute), matching the last-listed affiliation point, but never says that the Halloran Institute is an affiliated institute rather than a unit, so its works don't roll up to Albury; it only says I06 counts towards itself, not I02.

🟡 R10 executed run 4 graded by llm

W70 is credited to I06, an affiliated institution that, per institution_units, counts only towards itself and not towards I02. The Tessel Laboratory (I04) is credited on W71 and W72, not W70.

Why: Says W70 is credited to I06, an affiliated institution not rolling up to I02, but omits that only the last-listed affiliation counts.

🟡 R10 executed run 5 graded by llm

On W70 the author (A08) is credited to I06, not to the Tessel Laboratory (I04, which is credited on W71 and W72). I06 is an affiliate whose works count only towards itself in institution_units, not towards I02.

Why: Says W70 is credited to I06 (Halloran) and that I06 is an affiliate not rolling up to I02, but never states the 'last-listed affiliation counts' rule