This came up in a match today...

Unfortunately, I didn't do this as white...

No matter what black moves,

There are several mistakes (blunders) here...

Wait, Marvin, in your variation after 4. c3 Nf6, what if white goes now 5. Ng5? I will be in a better position to grab on e7 here than in your variation, as I'm holding the king-side in a death grip, a-la Carlsen vs. Anand 2012. 

The presence of Qs on the board favors white as the attacker. 

Moreover, exchange sacrifices on f6 are always a peril to fear, and so is an 2-rook invasion on f7. You can never chase my N from g5 as your Nf6 is also blocking the f-pawn. 

- - - - - - - - - 

On the other hand, it is true that in your variation after 4. Ng5 Qg7 5. QxQ+ (what I wanted to avoid with the move 4. c3) Kxg7 6. c4 Nb6. 7. Rdxe7 white still has a lot of pressure as black is really out of moves (7...Kf6 8. Nxh7+). Still it looks more cruel with the Qs on. 

I missed the 5. Ng5 move... But I don't think it makes white any better...

Dear Marvin, 

a pleasure to discuss this position with you!

I'm sure you had something in mind in case I go 5. Ng5 Rfe8 6. h3 Rad8 7. Rexe7 but just didn't write it in. 

If I were white, I would probably make my seventh move 7. Rdxe7, which is probably totally crushing! Let me know what you have against that. 

Now that I look a bit more into it I do see 7. Rdxe7 Qxa2 which will lead to 8. Rxf7 Qxf7 9. Rxf7 Kxf7 10. Nxh7 which I think favors white as the king is exposed and harrassable. 

About your 6. g3 variation - 

But wait - I now see your ideas with the Q sacrifice, and the checking resource on c1 which I missed and suddenly your evaluation looks just right... pretty much refuting my 5. Ng5 as a winning try and making 4. Ng5 clearly superior to 4. c3. 

Did you make this entire analysis on your own?

It looks like the capturing the e7 pawn on the seventh move is a blunder... It doesn't matter which rook does it - white seems to lose. It is a lot better to give up that pawn and actually make the position more solid.

The 1. c3? move is the problem. White keeps the advantage with an immediate 1. Ng5!

That's a really interesting game :D
Sacrifices and combinations all over the place. Difficult to keep track of all of them :P

BTW, could you post the whole game?

The game isn't mine. But wait - I now see that you're a robot, so I guess that answers my question from last post... anyway thanks for posting :-) I'm more interested in dialogue with human analysts as I can always fire up my own Houdini and find 'the truth' about any position if I so like...

Don't dismiss engines that analyze a position that interests you as something bad. I like to watch through grandmaster's games with the use of a computer. You can guess what's the best variation, then check the engine's opinion. It helps you to find patterns that you can apply when you need to think for yourself in your games.
By the way, that's how I learned to organise attacks against my opponents. Engines are really good at attacking play. I've played many games against engines and you can see when & how they prepare and initiate an attack against you (a pawn storm or a thematic sacrifice)... I've won a couple of OTB games in this style :D
Well, I guess it's a matter of preference how you want to learn to play chess. Some prefer living people, some prefer just engines, I prefer both :)

I work quite enough with engines - analyzing my games, going through opening lines that interest me... however when I discuss positions here in this website, what interests me is to see how much I actually saw, and to pit it against someone else's brains. If I want the engine's truth of course I can use my own, which is a pretty decent version of Houdini...

It's true that a computer is very helpful in improving one's game, and I'm also getting many ideas from my own computers for my own games...