With grateful acknowledgment to Rochade Europa 9/96 (which
itself acknowledges borrowing from Schachreport 8/96) EG offers
its readers this fascinating insight into the o-t-b skill level of
top-level masters confronted by (or confronting?) a Q+P vs. Q
ending. The game conclusion comes from the top board in a
hard-fought pre-Olympiad training match between Germany, and
Armenia.

Kindermann vs. A. Petrosian position after 63 g8Q a1Q 
White: Kh7, Qg8; pawn - g2
Black: Kb6, Qa1

The commentator reports with some astonishment that the player
of Black, trainer to the Armenian team (and, incidentally, chosen to
write the obituary piece on Kasparyan for New in Chess),
confessed to having no idea of how to play this ending.

64.Qg6+? Ka5 65.Qf5+ Ka6? 66.g4 Qhl+ 67.Kg7 Qb7+ 68.Qf7
Qg2 69.Qf6+? Ka5 70.g5 Qb7+ 71.Kg6 Qg2? 72.Qe6? Qhl 73.Kg7
Qb7+? 74.Kf6 Qhl 75.g6 Qh4+ 76.Kg7(?) Qd4+(?) 77.Kf7(?)
Qf2+ 78.Ke7? Qa7+(?) 79.Kf6 Qb8?? (Well! Time trouble,
apparently. Yawn.) 80.Qe7?? Qb6+? 81.Kf7 Qb3+ 82.Kf8? Qf3+
83.Kg8 Qd5+ 84.Kg7 Qd4+ 85.Kf7(?) Qd3 86.Qe5+. At this point
the scoresheet dried up. White, we read, had 8 seconds left. It was
enough because Black resigned a few moves later. An ('optimal')
win from here on: Ka4 87.g7 Qf3+ 88.Ke7 Qb7+ 89.Kf6 Qf3+
90.Kg5 Qg2+ 91.Kh4 Qhl+ 92.Kg4 Qdl+ 93.Kg5 Qd8+ 94.Qf6
Qd5+ 95.Qf5 Qd8+ 96.Kg6 Qa8 97.Qc2+ Kb4 98.Qb2+ Kc5
(Kc4;Qa2+) 99.Qc3+ Kd6 100.Qf6+ Kc7 101.Qf7+.
The Rochade commentary concludes: "Anyone scared stiff by this
endgame can take comfort. IGM Roberto Cifuentes scanned
457,000 games electronically to find just 321 QPQ endings. This
means that on average one game in 1424 reaches this material. And
even 24-carat masters commit blunder after blunder, unpunished!
Cifuentes found 45% won by the stronger side - 164 were wins and
177 were draws, which, he says, was about 'right' according to the
theoretically 'correct' results. c-pawn/f-pawn scored best (41 wins
in 70 - 58.6%); d-pawn/e-pawn 16 from 31 51.6%; a-pawn/h-pawn
33 from 110 - 30%. Reprinted from EG with permission.