Q47, run 2026-11-05-1235-haiku

The ranking flips because the two methods weight the same works differently. Full counting gives Rostval a credit for each of its four 2024 works (4 vs Kesting's 3), but Rostval holds only one of four author slots on each. Fractional counting gives Kesting 1.75 (W22 is single-authored, and W20 is split between two authors) against Rostval's 1.0.

Why: Kesting's larger shares (W22 single-author, half of W20) are covered, but it doesn't state that each Rostval work has four authors, so a quarter each, and it misses W21 as a half share.

Rostval has more 2024 works (4 vs 3), so it leads on full counting. Its four works each have four authors, giving it only 0.25 per work. Kesting's W22 has a single author (1.0) and W20 has only two, so Kesting leads on fractional counting (2.0 vs 1.0).

Why: Gets Rostval's four-author works at 0.25 each and Kesting's single-author W22, but omits Kesting's half share of W21 (only mentions W20) and doesn't clearly state one Rostval author per work.

Rostval has more 2024 works (4 vs 3), but each of its works has 4 authors, so its fractional share is split thin. Kesting has fewer authors per work (2 on W20, 4 on W21 with 2 Kesting authors, 1 on W22), so its fractional total is higher.

Why: Gives the Kesting side (W22 single-author, W21 half) and says Rostval's works have 4 authors, but doesn't state that only one author per work is from Rostval, i.e. a quarter each; also W20 is described as 2 authors rather than a half share, which is imprecise.

Under full counting, Rostval has 4 works and Kesting 3. Under fractional counting, Kesting gets 2.0 and Rostval 1.0, because Rostval's 2024 works are multi-author with one Rostval author each (shares of 0.25), while Kesting's W22 is single-authored and counts fully.

Why: Gives the one-quarter shares for Rostval's multi-author works and Kesting's single-author W22, but omits that Kesting also holds half of W20 and W21 (its larger shares).

Rostval (I09) has 4 works in 2024 but only one I09 author on each (1 of 4), so its fractional total is 1.0. Kesting (I07) has 3 works with a larger share of their authorship (0.5, 0.5 and 1.0), so its fractional total is 2.0. Full counting rewards the number of works, which favours Rostval, while fractional counting rewards share of authorship, which favours Kesting.

Why: Gets Rostval at one of four authors per work and Kesting's larger shares, but the shares given (0.5, 0.5, 1.0) are consistent with the key's W20/W21 halves and W22 single-author; it omits naming W20-W22 and the 'a quarter each' framing is only implied via 1/4. Stated total of 1.0 for Rostval is consistent with 4 × 0.25.

Fractional counting splits each work across its authors. Rostval's 2024 works each have 4 authors with one Rostval author, which gives 0.25 per work and 1.0 in total. Kesting's works have fewer authors (W20 has 2 and W22 has 1), so its shares are larger and total 2.0.

Why: Gives the quarter-share for Rostval's four-author works, but for Kesting it cites W20 (2 authors) and W22 (1 author) without mentioning W21 or the half-share of W20 and W21 explicitly.

Rostval has four 2024 works with four authors each, so its fractional share is 4 × 1/4 = 1.0. Kesting's 2024 works have short author lists (W22 is solo, W20 has two authors), which gives it a larger fractional share (2.0), while full counting gives each institution one unit per work regardless of co-authors.

Why: Gets Rostval's four-author works and Kesting's solo W22 and shorter author lists right, but never says Rostval has only one author per work (a quarter each) or that Kesting holds half of W20 and W21; the claim that W20 has two authors is only loosely matched.

Under full counting, each work gives 1 to every institution on it, so Rostval's four 2024 works give 4 and Kesting's three give 3. Under fractional counting, Rostval's four works each have four authors with one Rostval author (1/4 each, total 1). Kesting's works include a single-author review (W22, worth 1) and shared papers (W20 at 1/2, W21 at 2/4), totalling 2.

Why: Gives Rostval's quarter-each shares and Kesting's W22 single-author review and half-shares, but never says the institutions swap places or explains the rank reversal; the W21 share (2/4) is also oddly stated, though it equals 1/2.

Rostval's 2024 works each have four authors with only one from Rostval, so its fractional total is 1.0. Kesting's 2024 works have fewer co-authors (W22 is solo-authored, W20 has two authors), so its fractional share is 2.0 even though Rostval has more full-count works.

Why: Rostval quarter-share reasoning is correct (1.0), but Kesting's explanation is inaccurate: it says W20 has two authors but omits W21 and misstates the shares (W22 solo plus half of W20 and W21).

Rostval (I09) has four authors on each of its 2024 works (W21, W27, W28, W29), so each author's share is 0.25 and the total is 1.0. Kesting (I07) has fewer authors per work (W22 has one and W20 has two), so its shares are larger and it totals 2.0.

Why: Rostval quarter shares are correct, but the answer says Kesting has W22 (one author) and W20 (two authors) and omits W21 being half Kesting's; it doesn't cover the half of W20 and W21.

Full counting credits each work to every institution on it, so Rostval's four 2024 works (4) beat Kesting's three (3). Fractional counting splits each work among its authors, and Kesting's 2024 works have fewer authors (2, 4 and 1), so its shares add up to 2.0 against Rostval's 1.0.

Why: Gets the swap and the totals (Kesting 2.0 vs Rostval 1.0), but never says each Rostval work has four authors with one from Rostval. It gives Kesting's author counts as 2, 4 and 1, which contradicts the key's half shares of W20 and W21 and does not add up to 2.0.

Full counting favours Rostval (4 vs 3) because it counts each work once, and most of its 2024 works have four authors. Fractional counting favours Kesting (2.0 vs 1.0) because W22 is single-authored, giving Kesting a full 1.0, while Rostval gets only 0.25 on each of its four works.

Why: Covers the Rostval quarter shares and Kesting's W22 full share, but omits Kesting's half shares of W20 and W21, and says 'most' rather than each Rostval work has four authors.

Full counting counts works: I09 has 4 (W21, W27, W28, W29) and I07 has 3 (W20, W21, W22). Fractional counting splits each work among its authors: I09 gets 4 × 0.25 = 1.0, while I07 gets 0.5 + 0.5 + 1.0 = 2.0 because its works have fewer authors.

Why: Explains fractional shares and Kesting's larger share (W22 at 1.0, halves), but doesn't say Rostval's works have four authors with one from Rostval; uses IDs, and doesn't name W20/W21 halves precisely or mention the swap

Full counting gives Rostval (I09) 4 works against Kesting's 3, but fractional counting gives Kesting 2.0 against Rostval's 1.0. Rostval's four 2024 works each have four authors, so each gives Rostval only 0.25, while Kesting's W22 has a single author and gives it a full 1.0.

Why: Gives Rostval's four-author works (0.25 each) and Kesting's single-author W22, but omits Kesting's half shares of W20 and W21 and doesn't say only one author is from Rostval

Kesting's 2024 works have fewer authors per work, so their fractional share is larger. Full counting counts whole works, and Rostval has four 2024 works (W21, W27, W28, W29) against Kesting's three. Fractionally, Kesting gets 1/2 from W20, 1/2 across I07 and I08 from W21, and 1 from W22, for 2.0. Rostval gets only 1/4 from each of its four works, for 1.0.

Why: Correctly says Kesting has larger shares (W22 single-author, half of W20/W21) and Rostval gets 1/4 per work, but says Kesting has fewer authors per work generally and doesn't state that each of Rostval's works has four authors with one from Rostval explicitly; mostly matches, though the 'half of W21' split across I07/I08 is muddled.

Rostval leads under full counting (4 vs 3), but Kesting leads under fractional counting (2.0 vs 1.0). Rostval's four 2024 works each have four authors with only one from Rostval (0.25 each), while Kesting's W22 is single-authored and gives it a full 1.0.

Why: Covers Rostval's four-author, quarter-share point and Kesting's W22 single-author review, but omits Kesting's half shares of W20 and W21.

Under full counting Rostval (I09) has 4 works in 2024 and Kesting (I07) has 3. Under fractional counting Kesting has 2.0 and Rostval has 1.0. Rostval's 2024 works (W27, W28, W29) each have four authors, so each gives Rostval only one quarter of a share, while Kesting's works have fewer authors (W22 has one).

Why: Gives the quarter-share point for Rostval (though it lists only three works, W27-W29, while claiming four), but never says Kesting's larger shares come from half of W20 and W21 as well as the single-author W22.

Full counting credits each work to every credited institution, so Rostval wins on the count of works, 4 to 3. Kesting's works have fewer authors, so under fractional counting its shares are larger: W22 is a single-author work that gives Kesting a full 1, and W20 gives 1/2, while Rostval's four-author works each give it only 1/4. Kesting totals 2.0 and Rostval 1.0.

Why: Gives Rostval's quarter shares and Kesting's W22 full credit and W20 half, but omits W21's half share for Kesting and doesn't state that W22 is a review; the totals are stated without a W21 breakdown.

Full counting gives Rostval 4 and Kesting 3, because it counts each work once per institution. Fractional counting divides each work by its author count: Rostval's four-author works give 0.25 per work (total 1.0), while Kesting's W20 (2 authors) and W22 (1 author) give larger shares (total 2.0).

Why: Gets Rostval's quarter-share per work, but says Kesting has W20 (2 authors) and W22 only, omitting W21 and misdescribing W20/W22 shares; doesn't state one Rostval author per work.

Full counting gives each institution the whole of every work it is credited on. Rostval has 4 works and Kesting has 3, so Rostval leads. Fractional counting divides each work by its number of authors. Rostval's authors are one of four on each paper (total 1.0), while Kesting's share is larger on its papers (W20 and W22 have few authors), giving 2.0.

Why: Rostval's quarter shares are covered, but Kesting's shares are not: the answer says 'W20 and W22 have few authors' and omits the single-author W22 and the half shares of W20 and W21.

In 2024, under full counting Rostval has 4 works (W21, W27, W28, W29) and Kesting has 3 (W20, W21, W22), so Rostval leads. Under fractional counting Rostval's four works each give a quarter share (1.0 total), while Kesting's W20, W21 and W22 have fewer co-authors and give 2.0.

Why: Correctly says Rostval's four works give 1.0 total (a quarter each) and Kesting holds larger shares (2.0), but doesn't state the four-authors/one-from-Rostval reason or the specific single-author W22 and half shares of W20/W21.

Full counting gives each work to an institution once, so Rostval's four 2024 works beat Kesting's three. Fractional counting splits each work across its authors, and Rostval's works have four authors with only one Rostval author each (about 1.0 total), while Kesting's are single-author or have a larger share of their authors. Kesting's total is 2.0.

Why: Gets Rostval's four-author, one-Rostval-author quarter shares, but does not identify Kesting's W22 single-author review and half shares of W20/W21 (only vague 'larger share'); also states a total without that basis

Full counting favours Rostval (4 vs 3), since it has four 2024 works. Fractional counting favours Kesting (2.0 vs 1.0), because Rostval's works have four authors each (0.25 per work) while Kesting's 2024 works have fewer authors, so its authors take larger shares. This has no JSON field, so it's answered in prose only.

Why: Correctly says Rostval's works have four authors each (0.25 per work) and Kesting has larger shares, but doesn't identify the single-author review W22 or the half shares of W20 and W21.

They do swap. Rostval (I09) has 4 full-count works against Kesting's 3, but Kesting's fractional total (2.0) beats Rostval's (1.0). Kesting gets a full credit for W22 and half credits on W20 and W21, while each of Rostval's four works gives only a 0.25 share.

Why: Gets Kesting's W22 full and half credits on W20/W21, and Rostval's 0.25 shares, but omits that each Rostval work has four authors with only one from Rostval.

Full counting gives each work to every institution with an author on it, so the count of works decides. Rostval has four 2024 works against Kesting's three, so it wins full counting. Fractional counting divides each work among all its authors, and Rostval has one of four authors on each of its four works (1.0 total). Kesting's shares are larger: W22 is single-authored and W21 has two of four authors (2.0 total).

Why: Gets Kesting's larger shares (W22 single-author, W21 two of four) and Rostval's one-of-four per work, but omits that Rostval's four works each have four authors with a quarter each as stated, and cites W21 instead of W20 and W21 as half-shares.

They don't swap. In full counting they tie at 3 each (I07: W20, W21, W22; I09: W21, W27, W29). In fractional counting I07 leads (2.0 against 0.75), because fractions shrink with the number of authors and I09's works have four authors each.

Why: Gets the fractional logic (four authors per I09 work, so quarter shares) but denies the swap, counts I09 as three works instead of four, and never names W22 or the half shares in W20 and W21.