Help with pawn tension and positional ideas

Hello,

So I just finished a correspondence game (I played as black) and I was hoping to get insights from the community about moves 17 and 27.

Move 17:
I notice that whenever my opponent pushes the pawn forward to create some tension with my pawns, I'm generally not sure whether to push that pawn or capture it. In this case, clearly my decision to push was wrong (as pointed out by computer), but my thinking was that if I capture the f4 pawn and land my knight on e4, white can just take it out with his light square bishop. And given that I don't have a light square bishop, I thought I should grab some control of the light squares and have a pawn in the enemy territory.

Move 27:
I admit I calculated for a long time here. I DID consider Rd8 with the idea of deflecting the bishop away from the diagonal to create some kingside attack. However, I didn't play it because I noticed that after my opponent plays Bxc6 and I go Qe5, I didn't see what to do next and my opponent restored his/her material. Of course, the computer points out I should go with the g5 - g4 idea, which completely went over my head. So what should I do in the future to "see" these ideas??

Thanks for your time.

EDIT: Feel free to comment about my other moves as well! I'm trying to learn and improve as much as possible.

For Move 17, your push e4 isn't bad per se, Stockfish says it is about even (+0.3). However, note that your Bishop is now permanently blocked by the f4 pawn, and your knight has no good advancing moves.

The reason why capturing exf4 is better may be partially due to tactics. After 17... exf4 18. exf4 Ne4, you (black) are threatening several rather serious threats like:

1) ... Qa7+ winning at least the exchange
2) ... Nc3 winning the bishop

The only good way to meet this two threats seem to be 19. Qd3. After which, you can play 19 ... a5, which cannot be captured otherwise ... Bc5+ is a serious check that wins the exchange again. After ...a5, overall, you are at least a pawn up in terms of material hence Stockfish gives you a (-1.3 for black).

For Move 27, firstly your pawn on c6 is a weak backward pawn that will most likely eventually be won, hence you should not worry too much about it being taken, and your opponent "restoring material".
...Rd8 is a hard move to find though.

For the "g5 - g4 idea", it is good and standard attacking idea, especially since your opponent does not have a dark squared bishop. Playing g5-g4 primarily weakens the diagonal a1-h8, which is not that serious here since your opponent does not have a dark squared bishop (and you do).
Whenever your opponent plays h3, it creates a "hook" that allows you to play the "g5-g4 idea" to open up a file towards his king.

The main point why 17...e4 is bad is that it takes away an ideal square for your knight. a knight in the centre, that cannot be driven away by pawns is no weaker than a rook.

27...Rf6 is not bad and does not spoil anything, you are still winning. 27...Rd8 was stronger indeed. Your Rh6 is already active and your Ra8 still stands on its starting square. So activating Ra8 is more logical than moving Rh6. Also after Bxc6 Ra8 is under attack, so it makes sense to move it away right away, so as to attack the black squares weakened by his h3 with ...Qe5 and ...Qg3, where the active rook at h6 threatens Rxh3. So the rook is better at h6 than at f6, while the other rook stands better at d8 than at a8.

I didn't like move 19 trading bishop for knight.. the knight really doesnt do much and keeping your bishop around longer seems like it would be useful.. (unless im missing something)

oh god wrong color... never mind i quit

Nice game, the corrections to mistakes are bit more challenging to find than expected. Very much about backward, isolated pawns and ...

after 17. f4 e4, minority attack with the a-pawn (stockfish analysis after the game is ofc 32/20, though it's a little more obvious when kingside pawn structure looks gummed up).

Though maybe worth noting that pushing the backward c6-pawn instead of a6 can run into Qxd5+ in some variations, potentially forking king and rook in addition to reinforcing the c5 square. Albeit you probably spotted that!