Breaking the BDG

In the most recent TCEC/CCC post, TCEC S21 Sufi Underway; CCC Still Romancing (August 2021), I noted,

CCC: The site continues to run multi-engine, multi-round-robin blitz/bullet tournaments based on thematic 'romantic openings'.

One of the romantic openings was the:-

Blackmar-Diemer Gambit (1.d4 d5 2.e4 dxe4 3.Nc3)

I've played the BDG in both OTB and correspondence events and know something about it. It's fun to play, although White struggles to get compensation for the Pawn. I once wrote a post about it, although you wouldn't guess the subject from the title: Naming the Opening Variations (October 2014).

A few months ago I worked out the procedure for downloading CCC PGN files for a particular event and documented it in a post: CCC PGN IV (May 2021). I downloaded the CCC BDG file, loaded it into SCID, and analyzed the games. The following chart shows that there were 528 games played using the forced opening and that all but one continued 3...Nf6. The stats at that point aren't good for White, who scores only 37.4% with 58% of the games ending in draws. The rest of the chart shows what the stats say is the best line for both sides: 4.f3 Bf5 5.g4 Bg6 6.g5 Nd5.

I asked Stockfish to analyze the position after 6...Nd5. It told me that Black has a small advantage after 7.Nxe4, a larger advantage after 7.Bg2, and an even larger advantage after 7.fxe4. In the 2014 post, 'Naming the Variations', this line was called the 'Kampars Gambit'; the only move given was 7.fxe4. If I ever wander into this line again, I'll try 7.Bg2.

It looks like the engines are breaking the romantic openings. They are also busting the old analysis. Should anyone be surprised?