How To Lose In 13 Moves

I PLAYED at the Highlands Open, a nine-round swiss held in Trebic from Sep 23-30 as part of the Czech Tour.
My score of +4=1-4 was good enough to gain 25 elo points, but in the sixth round I had the embarrassment of losing in 13 moves.
The game began reasonably enough ...
Jan Bartos (FM2209) - Spanton (1858)
Catalan Opening
1.d4 d5 2.c4 e6 3.Nf3 Nf6 4.g3 c5 5.Bg2!?

Black to make his fifth move

This is quite a common idea in the Catalan: White ignores Black's threat to demolish his centre and instead gets on with developing his kingside. I was already out of book.
5...cxd4
Most popular is 5...Nc6, but the text has been played by some strong players, including Aronian.
6.0-0 Nc6
6...dxc4 is possible, but White replies 7.Qxd4, leaving Black even further behind in development.
7.Nxd4

How should Black play in the centre?

First, it is important to realise that the ambitious 7...e5? loses a pawn to the simple 8.Nxc6 bxc6 9.Nc3, eg 9...Be6 10.Bg5 Be7 11.Bxf6 etc.
7...Bc5 drives the d4-knight from the centre, but there is no gain of tempo as White hits the c5-bishop with 8.Nb3.
Instead, I came up with ...
7...Qb6?!
This drives away the knight with gain of tempo, and I later discovered it had been the move of some strong players. But I think it is more significant that the queen becomes a target for White's minor pieces.
8.Nc2
By no means the only reply. The most popular line runs 8.Nxc6 bxc6 9.Nc3 Be7 10.cxd5 cxd5 11.e4 dxe4 12.Nxe4, with an excellent position for White thanks to the powerful Catalan bishop on g2.
8...d4?
The text is liked by the analysis engine Stockfish8, but Bartos's strong reply shows it to be an error.
Komodo9 prefers 8...dxc4, but that hardly convinces after the straightforward 9.Nba3.

Bartos's ninth move proved very strong

9.e3!
I had not been worried about White giving up the bishop-pair to grab a pawn with 9.Bxc6 Qxc6 10.Qxd4 - Black's compensation is obvious. But the text gives White a strong game, whatever Black replies.
9...Bc5??
I rejected 9...dxe3 10.Bxe3 Qxb2 as too risky. The analysis engines at first disagree, until they find the move that caused me to reject this line, namely 11.Nba3.
Best may be 9...e5 10.dxe4 dxe4, although Black's passed d-pawn is probably weak rather than strong.
10.b4 
Even better, according to the analysis engines, is 10.exd4! eg 10...Nxd4 11.b4, which wins a piece.
10...Nxb4?
Black is still just about in the game after 10...Bxb4, but 11.exd4 gives White the upper hand.
11.exd4 Be7
11...Qd8!? is a computer-suggested near-save, and does keep the game going, but White should win comfortably enough after the forcing 12.Nxb4 Bxd4 13.Nc2 Bxa1 14.Qxd8+ Kxd8 15.Nxa1.
12.c5 Qb5 13.a4 Qd3
Losing instantly, but the b4-knight could not be saved, eg 13...Qc4 14.Nba3 Qb3 15.Rb1.
1-0
So how do you lose in 13 moves? There are many ways, but this game shows three things that help:
1. Pick a variation where you know very little theory.
2. Develop your queen early.
3. Push a pawn across the halfway line when it will be difficult to support it.