Odd computer analysis


I was flicking through the computer analysis of one of my recent games and noticed that in a position evaluated at more than +15 for white the computer gives a variation which results in black mating white! Indeed it even claims that at the end of it's variation that white has a decisive advantage (though the comments at the end of variations are not given in the pgn and are hence not in the diagram). Anyone else have any examples of strange computer analysis?

(Note that I have promoted the computer line 24...Be6 to the main variation instead of the game continuation 24...c6.)


(presumably this is the horizon effect + lack of sanity check?)

Wait.. what move do you think the computer faultily analyzed?