Strange blundercheck results in Rybka Aquarium 2011

Hi, guys!

Nevertheless I decided to buy Rybka 4 Aquarium 2011 software and installed it successfully. Now I play chess with computer using this environment. Mostly I play Weak Delphi engine at increasing ELO levels. My progress is good, I guess, but Aquarium game analysis modes work strange to me. Here you are latest example of odd blundercheck results: 

After move no. 17 blundercheck inserted a variation line which is completely idiotic in my view because I would lost a rook for a bishop! Yet playing on my own I was able to choose better solution. Am I right or maybe I set bad blundercheck parameters? For your knowledge it is:

- 1st pass: 1 second AND 1 ply per move

- 2nd pass: 5 seconds AND 1 ply per move

- Thorough blundercheck - ON

Maybe this is too little time to make a proper blundercheck?

I never used rybka aquarium but recomended 17. move is good cos you get knight and a bishop for a rook and that is a good trade. If u look at relative value of pieces you will see that rook is 5 points , knight 3 and bishop 3 so in this case 6>5 Also there are many peons on the board , no open files so rooks are not so good 

It's not stupid, it's two pieces for a rook, two pieces are usually better than a rook and a pawn (same point count)

I think you made the right choice +3 is better than +1

OK, but why blundercheck's variation doesn't include getting knight by queen in move no. 22 for instance?

Anyway I guess playing with WeakDelphi at 900 ELO level means my adversary constantly chooses bad moves and that is why I was able to get bishop while not losing a rock in exchange. In other words blundercheck is much more useful analyzing games with stronger opponents.