Why jumping move?

[baduk17]

I see this jump recommended a lot, but it never feels like a better move when I see it. Can someone explain?

[Station_11]

How did B get to this situation? Locally W has 6 stones, B 4.
Example: W R3, B attaches R4, W extends Q3, B follows Q4, W hanes P4, B hanes P5, W extends O4; B plays tenuki, W cuts Q5, B ataris R5, W extends Q6.

Why did B tenuki? Was it worth it? What does the board look like? What’s happening on the top right corner? If an engine suggests R7 jump, it has to do with the whole board, global evaluation. When you say “a better move,” compared to R6 push? Again, what’s happening globally? How can we evaluate if B should tenuki again, without seeing the whole board?

[baduk17]

here it is

kakuna17 vs. Spectral-10k

[taatelikakku]

There’s a proverb ‘dont push from behind’. So the jump obviously avoids breaking this proverb. Whether this explanation is just dumb and wrong idk…

[teapoweredrobot]

I’m far from being an expert but I think AI will often give up a corner for influence towards the centre. And in the game we can see that pushing (from behind) earned black 15 points while white built a magnificent wall that ended up being worth about 200 points!
So I think it depends what you mean by “better” it could be that the jump gives up the three stones (not “better” for them!) but earns many more points elsewhere (“better” for winning the game). Also it still leaves the potential not to give up the stones.

But all in all it might be that the actual answer is simple that jumping ahead is usually better than pushing from behind as explained by @taatelikakku

[Station_11]

There’s a proverb ‘dont push from behind’ – this is a very unfortunate mis-transmission, but it’s so pervasive now, it’s kind of hopeless. If you ask pros trained in China, Japan, Korea, they’d tell you they’d never heard such a thing when they were coming up.

A bad move happens to push from behind – don’t play it.
A good move happens to push from behind – of course, play it!

[mark5000]

Just one question for @baduk17: Which of the next positions do you prefer for Black?

[timuzhti]

One other thing to keep in mind that Kata’s winrates are for when it’s playing itself, so the actual score difference of the push vs the jump is a whopping 2 points. Which white promptly loses back to black because white didn’t actually take advantage of that inaccuracy (with the hane, R7 after R6). So besides being a very small difference score-wise, your opponents might not even notice these small inaccuracies or take advantage of them, making it feel even smaller. Unless both players are much stronger, a 2 point advantage, or even a 10 point one, isn’t that much.

I think switching the analysis view from win % to score (which is something I just realised you could do. Maybe it’s a newish feature) would make it much more helpful: It make it clear that you only really “lost” the game with the sequence from move 64 to 86 (where you lost ~6 to 25 points every move, though to be fair white also lost a lot by not perfectly following through). You don’t see that in the winrate really, because at that point it was already >90% for white.

[baduk17]

The second one looks better. I’m not that smart though.

[baduk17]

I gotta admit I laughed out loud when I rewatched that just now.

[bugcat]

Yeah, seeing that White can hane at (1) (because the Q7 atari followed by P6 R8 S7 S8 S6 works for White) really changes your perception of the position.