It seems to me 4.Nc3 is playable, hoping to lead on the Black d-pawn: 4.Nc3 d5 5.Nce2. The Caro-Kann (you win your bet) has transformed itself into something like a King's Indian Attack, in which the Queen's Knight is not badly placed on e2. However, Black might instead exchange, something like 4.Nc3 dxe4 5.dxe4 Qxd1+ 6.Kxd1 Ng4 (say) 7.Ke1 with about an even game. Of course right here, Black could have played 3...dxe4 4.dxe4 Qxd1+ 5.Kxd1 Nf6 etc.

Their mobility is limited, it is true, but both retain considerable potential. Black's Queen's bishop isn't exactly a thing of grace and agility either, and its potential might be more limited than White's LSB (light square bishop. The latter can develop to e2, with an eye on h5. Sooner or later, White will "get in" f4-f5, which will release the prelate on e3 as well.

Maybe Black ought to have considered ....Be7.

True enough, but 11.bxc3 had its points, transposing into something like a French Winawer. Mind you, Black might then play ...f5 after which White will have to figure out how to break out of his position. I still like White's game, though.

If 14...a4, the Q-side would have been completely closed down, after which all the action would happen on the other side of the board. That would probably suit White better than Black. But instead of exchanging at once, Black could have waited for a more opportune time. 14...Nc6 comes to mind. White is unlikely to want to exchange on a5 (for a move or two at any rate), and a3-a4 is met by ...a5xb4.

Oops.

I rather like 18...e5-e4 instead of the capture on f4.

What a difference one move can make. Black returns the compliment, and the game completely opens up. Maybe 19...Rf5 or 19...g5 were indicated.