I like how the eventual checkmate comes from the fianchettoed bishop - playing to your existing positions is a good concept to keep in mind.

wait how is that 1-0... oh. That's my thought process when I see the end of that.

To answer to the Q, 14. exd5 is worth being played and is playable. You sacrifice the B to get more F, which is a lead in attack. The more force, the better. Splitting force is bad because when your opponent has more force contacting your force that got split, then your force will be killed by your opponents force, and you will be in for trouble. As happened in tis game, trouble came as black had less force where there was contact between both sides. The place where both sides have contact can also be said as where the attack is. So having more force then your opponent will help you in a game. I learned this from Napolean's plan and connected it to chess.

What about 17...f6 or 17...Be7?    17...Bc5 seems to make it easy : /

Nice games. The plans were super. I cant even believe masters play this way. The game ZL vs. PS, the PS guy played really well. He tried and gave his most effort to try to win. I was amazed from how he played and thought on move 36, you would have to give up the R to keep opponent from playing mate. I didnt see through all the way for Q check to keep the R and B. 

I'd like to find out how I didn't get 31. Qc7. Basically I committed the cardinal analytical sin: bouncing around analysing random variations. I looked at some forcing moves like Rh3, got a hallucination of a knight fork with the queen on h3 and king on h8(thought it'd be on h7), then I asked myself if there was a quiet move.

Since the solution was just that, the moment of rejecting such a move is also important. I noticed Black's "threat" of ...Rxe6 and just rejected a "quiet move"; notice how I didn't specify Qc7 or even Qxa7, both of which stop ...Rxe6. If I had seen the line 31. Qc7 Rxe6 32. Rxe6 Qxe6 33. Rh7+, which is extremely easy to find, I would know that 31. Qc7 wins because the attack plays itself from there.

In conclusion, I need to find a better method of analysing. Perhaps I should test Kotov's(and Silman's and many more) method of picking candidate moves before doing any analysis and only then working them out one at a time. No doubt I would've found Qc7 had I done this.

they r too good