Gould Puzzle #516
I started working through my analysis of Gould Puzzle #516, which my previous scan using the Cake 8 piece database, and revealed to be a loss for white, instead of the draw that Gould listed. Here’s the position:
The database reports this position as lost, but Gould thinks it is a draw. Let’s see where the published play goes wrong. Gould thinks that the play should proceed 6-2, 26-23, 19-16, 12-19, 2-7, 11-15, and 1-6, leading to the following:

Gould suggests that red should make 23-27, but that move tosses away win. 14-18 enables you to retain the extra man.
Gould listed play as proceeding with 23-27 (which is a drawing move), 6-10, 14-18, 20-16, 18-22, and 10-14 leading to a position where each side has three kings. But Red can maintain his one piece lead by playing 14-18 instead. You can walk a narrow line which is a pretty nice little ballet to advance all the red men, leading to a 4 on 3 endgame, which is tricky, lengthy, but very winnable.
Addendum: There is a similar flaw in the other variation presented.
I recall burning three or four weeks of a sabbatical getting Saccade.com on the air with Wordpress. So much tweaking…