I was analyzing The World vs Gert Jan Timmerman game with Shredder 11 and Rybka 1.0 beta when I encountered a position that, at 18-ply, Shredder rated as +1.38 and Rybka as -0.64:
Playing a few moves reveals that Rybka is probably more correct here. Are you aware of any other positions where any two engines would have so wildly differing evaluation ratings?
Well, try playing a Gothic Chess game between Joker80 against TSCP-Gothic. Very often TSCP-G thinks it is at +6 ahead while Joker80 has +0.50 for himself. Usually Joker80 then wins.