A Surprising Defensive Resource I Almost Missed

This was just a casual 3+0 blitz game I played on Lichess. By the time we reached the critical moment, I was completely winning as Black. White was a bishop up, but their king had wandered all the way to h3, where it was completely exposed to my queen and rook. Objectively, there was no realistic way to save it.
Or so it seemed.
White's only remaining hope was to keep giving checks and pray that something unexpected would happen. Normally, this kind of desperate checking sequence simply delays the inevitable. But here, one very surprising defensive resource was hidden beneath the surface, and I came remarkably close to walking straight into it.
What makes the position especially instructive is that all of Black's candidate moves look winning at first glance. In reality, though, there is an enormous difference between them. One move loses immediately, another only draws, while a quiet defensive move simply wins on the spot.
Let's start with the move I almost played.

---

The Tempting Move That Actually Loses

After White's desperate 34.Be4+, my first instinct was to play ...f5.

It looked completely natural. The pawn blocks the bishop's check while simultaneously attacking the bishop, gaining an extra tempo. In practical terms, it almost feels like the perfect move: you defend and attack at the same time.
The problem is that this move accidentally creates a tactical resource that didn't previously exist.

The beautiful part about this combination is how unexpected it is.
If you look at the position before ...f5, White hardly appears to have any threats at all. There isn't some obvious tactical shot hanging in the air. The only thing White can really do is keep checking and hope something appears.
It is ...f5 itself that creates the winning tactic.
That is what makes this resource so instructive. Had Black simply chosen another move, none of this would have worked.
I also genuinely believe my opponent would have found 35.Rh8+!! over the board. They only had seven seconds remaining, but it was practically the only forcing move left in the position. When you're completely lost, those are exactly the kinds of moves you instinctively look for.
Fortunately, with just over twenty seconds on my own clock, I paused before playing ...f5 and decided to look for one more option.
That small pause made all the difference.

---

The Quiet Move That Wins

Fortunately, I didn't play ...f5.
With a little over twenty seconds remaining, I stopped myself and looked for another way to deal with the check. Eventually I found a move that looked much less ambitious, but was actually much stronger.

What I really like about this position is how subtle the difference is.
After ...f5, White suddenly has a brilliant tactical sequence that wins Black's queen.
After ...g6, that exact same idea completely falls apart.
White still sacrifices the rook, still gives the queen check, and still hopes for the same combination—but there simply isn't one. The bishop has nothing to capture on f5, there is no discovered tactic against the queen anymore, and Black's king is perfectly safe. The attack simply fizzles out.
That is why ...g6 is such an instructive move. It doesn't look nearly as active as ...f5, yet it completely removes every ounce of White's counterplay.
In fact, the move is so strong that after 36...Kh7, White's position is already beyond saving. The king on h3 remains hopelessly exposed, while Black still threatens to continue the attack with both the queen and rook. There simply aren't enough checks available to keep the attack alive.
One thing I also found fascinating is that my opponent chose 35.Rh8+ almost instantly in the game. Considering they only had a handful of seconds left, I think that's strong evidence they also would have found it after ...f5. When you're completely lost, forcing moves become incredibly attractive, and Rh8+ is by far the most natural one to investigate.
Looking back, it really feels like I dodged a bullet. I was one move away from turning a completely winning position into a completely lost one, all because I wanted to gain one extra tempo against the bishop.
But the position still had one more surprise in store.

---

Even Sacrificing the Exchange Was Better

There's one more continuation that I found fascinating after the game.
Suppose Black doesn't play ...f5, but instead decides to eliminate White's dangerous bishop immediately.
At first glance, this feels almost absurd. Black is completely winning and suddenly gives up one of their strongest attacking pieces for a bishop that isn't even threatening anything immediately.
Yet even this turns out to be a better decision than ...f5.

I think this continuation is particularly instructive because it shows that Rh8+ isn't just a one-off tactical trick.
It's a recurring geometric motif hidden inside the position.
Whether Black plays ...f5 immediately or only after exchanging the rook, the same sacrifice keeps appearing because the rook on h8 is able to drag the king onto exactly the square where White's queen wants it.
The only difference is that after ...Rxe4, Black has already removed White's bishop, so there is no longer a way to win the queen afterwards. White can only scrape together enough checks for a perpetual.
That means the three most natural candidate moves all produce completely different outcomes:

• 34...f5? loses.
• 34...Rxe4! draws.
• 34...g6! wins.

I find that remarkably elegant. If someone had shown me the original position without any context, I would have assumed all three moves were roughly equivalent. After all, Black is completely winning, White's king is hanging by a thread, and White's only apparent hope is to keep checking.
Instead, the position turns out to be incredibly precise.
One move creates loses, another just barely allows White enough activity to escape with a perpetual, while the quietest move of all simply shuts everything down.

---

Final Thoughts

One of the biggest mistakes we can make in winning positions is assuming that every reasonable move wins. When you're completely winning, it's tempting to play the move that looks the most active or the most forcing, especially in blitz where there isn't much time to calculate.
This position is a great reminder that winning positions still deserve respect.
The move ...f5 looked almost perfect. It blocked the check, attacked the bishop, and seemed to improve Black's position in every conceivable way. Yet it was actually the only move that threw away the win because it accidentally created a tactical idea that hadn't existed a move earlier.
Luckily, I hesitated for a few seconds and found the much quieter ...g6, which simply removed White's counterplay altogether.
I think this game perfectly illustrates how a completely lost position can still hide one miraculous defensive resource, waiting for just the right mistake to bring it to life.