About the game analyzed on Tuesday The main reason black lost wasn't the final tactical mistake, but rather a passive play in the opening. The only two real things black did were (1) bishop exchange, which was good (2) queen exchange, which was already necessary at that point. Lets consider the critical position:
Here black played 14... Qc7. Were there more active options? Note, that previously black forced the exchange of light-square bishops, which means white cannot play Bf3 to protect the center. On the other hand, black has more space. E4 pawn is therefore a target. And Nc3 is also a target. For example:
(1) 14...Nh5 (Houdini's favourite; threatening with Ng3) 15.Qf2 Bxc3 16.bxc3 Nf6 (the knight has nothing to do there) 17.Nd2. White has weak pawns, but the protection layer around black king has thinned. I would never play like that in a long game. The very fact that Houdini shows near zero evaluation in such a structure shows something.
(2) 14...Nb4 forces the queen to defend c2 pawn. Say white continues with his plan 15. a5 Nd7 -- the difference from the variation above is the threat to white's Q-side pawns is now immediate, while white has nothing towards developing an initiative on K-side, where he has to play if black takes on c3
(3) (my favourite) 14... Na5. White has no LSB to protect c4. From c4, the knight threatens to exchange the dark square bishop (making a2-a4-a5 plan pointless) and with Nxb2. For example 15. Rb1 Nc4 16. Bf2? Nxb2 and Rxc3. White would have probably respond with 15.e5 or 15.Bd4. For example 15. e5 dxe5 16.fxe5 Nd5 17.Nxd5 Qxd5 18.Nxa5 Qxa5 19. Bd4, and black rooks are active and positionally black is Ok.
The point is, you have to pose problems. Nowadays everybody can watch instruction videos and execute some plan - if you do something on your own, they just get lost.