What am I missing?

So... this is my game after 31 moves.

I think 33. Qg2 would have been crushing, but the computer analysis says I should have pushed a pawn (33. d5) instead. 

What am I missing?

You are threatening mate on e8. And e8 is defended by Black's Rook and Queen, so attack on Black Queen - pawn d5 and then Black Qd7, and White Re7 should be enough.

but why isn't 33. Qg2 the best move? If 33. d5 Qd7, I still can't attack e8 yet.  With Qg2, I'm skewering the queen and threatening mate.

Thanks, Sirfatticus, I'm using Chess.com's diamond level (2500 rating) post game computer analysis. 

chess.com's diamond level has its flaws told me i blundered when my move gave me a forced mate in 6     what about   33 .... qxq?

I _LOVE_  33... QxQ - 34. Re8+, RxR  35. RxR#

That's what makes it such a crushing move, IMO.  Take the queen, you lose.  Move the queen, you lose.  Give up your queen and rook, mate in 8.

So... it's chess.com that's wrong?  Or what I missing about 33. d5 that makes it the best move?

you must have the pices in  different places than my diagram the white rooks are not connecrted 

@rickdaniels52

The diagram starts with 32.  Red1 (connecting the rooks) and 32... b4.  To provide a hint of context going into the question.  Then came my mistake of 33. c4.  Chess.com insists 33. d5 is the best move, but 33. Qg2 seems much better to me.  What am I missing? 

that's why i lose games your move looks good

Thanks, (and feel free to challenge me to a game ) but I'm bothered by the computer analysis.  This advanced analysis is one of the 'perks' of being a paid member.  So... either there is a problem with 33. Qg2 that I don't see and 33. d5 really is better... or I'm paying for crappy information. 

Annyone know?

You set up the board wrong.  With white rook on d1, you simply drop your queen.

You should correct the numbering, or adjust your explanation.  Whatever.  Your presentation has already mislead some folks earlier in this thread.

what if black plays 33..Qd7 in response to 33Qg2 ?

Then 34. Qxa8+...

@ zborg - the diagram starts at move 32.  The move in question is 33.  The numbering is correct.

@ Jitesh - if black moves the queen away, such as 33... Qd7, then 34. Qxa8+ and black can only delay mate by throwing the knight and queen in front of it.  Even if ...Qxa8, Rxe8+ Qxe8, Rxe8# takes it home. 

So... am I missing some aspect of awesomeness about 33. d5 that makes it even better than 33. Qg2, or am I paying for a bad analysis?

The computer on chess.com and the computer workout program both analyze too quickly to be of too much use at advanced levels. I am often told that winning moves are blunders.

The latter.

Here's the 'correct line' offered by the analysis, with Sifatticus' Houdini line as an alternate.