We left each other on Monday - and thanks for all the contributions to the comments box by the way - with me about to play the more than somewhat rank ... Qxd5.
As Will pointed out, exchanging queens just gives White a big pawn on d5 and forces the knight to a horrible square. It's not too hard to see that is it? So how did I come up with something so self-evidently bad?
Well I'd wanted to get my e-pawn going and since the immediate ...e5-e4 drops the bishop on f5 I thought
1. ... Qxd5, 2. cxd5 Na5, 3. Nd2 - to stop the knight bouncing of the rim to c4 - ... e4
would be just the ticket.
Naturally, when the game arrived here I saw at once that I still couldn't push my pawn because White would respond 4. Bb4 defending my target on c3 and hitting the wayward knight.
Even if White didn't have this resource my intended
4. Nxe4 (?) Bxe4, 5. Bxe4 Bxc3
would be very bad anyway.
When pondering the queen exchange I stopped analysing at this point thinking I'd be gaining a tempo on the rook. I was quite oblivious to the fact that White can hit the bishop straight back with 21. Re3 which leaves me in trouble since, aside from the fact that I've given up the bishop pair on an open board, my knight is still hanging around on the edge doing nothing but be vulnerable to attack.
Of course if White didn't have 4. Bb4 or this he could always play 4. Bxe4 which is much better than taking with the knight since after 4. ... Bxe4, 5. Nxe4 c3 would be defended and I'm just a pawn down.
So ... Qxd5 was based on a calculated sequence that could be improved by not one but two alternatives for my opponent on the third move and wouldn't even have worked if I'd got what I thought I wanted.
Exchanging queens was a horrendous tactical error then? Well perhaps but in a broader sense I think there's something much more fundamental to consider - the question of what I was doing analysing ... Qxd5 in the first place. I'll be coming back to that next week.