Go Memes Pedantry

No. One expects to be attacked by enemies. What is more dangerous and triggering is to be attacked by “friends.”

2 Likes

As a first step why do we call chinese checkers, chinese checkers?

2 Likes

In the United States, the commercialization of Halma (invented c. 1880) occurred in the early 20th century. That coincided with a period of national fascination with the exotic side of Asia, particularly China (despite the so-called “Yellow Peril” propaganda in the first two decades of the 20th century). Consequently, associating the game with China was nothing but a sales gimmick, and “Chinese” checkers was born. I once saw an early Chinese checkers board in an antique shop. It was called “Hop Ching Checkers.” Especially interesting was that the board had 5 rows in the home triangles, rather than the modern 4 rows. I would have bought it, but it was very expensive.

1 Like

Looking back, it appears that no one ever answered @Lys’s question about why glass marbles are called “marbles.” The answer is that early marbles (aggies) were typically made of agate, a decorative stone of great variety and popularity in the lapidary field. People might well confuse some agate with marble, or perhaps marble was also used at that time (I don’t know). Glass marbles were a later development, in my day, and were considered an abomination by my parent’s generation. Aggies were still around and prized when I was a kid, and I had a few of them, including several “shooters.” Not sure where my collection is now. I think I passed it on to my “little brother” when I became a teen.

4 Likes

This meme confused me as we usually oppose big moves vs urgent moves to point that urgent comes before big.
But maybe that is that was intended?

The immediate perception is that the runner (in blue) stopped short of the base (white square on ground) in order to trick the defender (in grey) into thinking that he was already safely on base (preventing the attempted tag by the defender from having any effect) and thus the defender did not actually bother to complete the tag, allowing the runner to then touch the base and establish a safe position. Alternatively, one might view it as the defender knowing that the runner has not yet touched the base, but then having a mental lapse of not completing the tag, due to the runner inexplicably stopping short. Either way, it appears that the runner got away with something that he should not have.

The reality is not readily apparent without additional context about the state of the game. At the beginning of the clip, you briefly see the catcher (man in heavy padding standing behind the batter) throwing the ball towards the defender seen later in the clip. Since the batter (in blue) is not making an attempt to run to first base and the catcher had the ball, we can conclude that the runner was apparently trying to steal a base (attempting to advance, without the ball being put into play by being hit by the batter). Note that during a steal attempt, the defender needs to specifically tag the runner (with a glove holding the ball) before he reaches the next base. However, the crucial part of missing context is that the pitch just thrown towards the batter (and caught by the catcher, right before the clip began) was ruled as a “ball” (an invalid pitch) and happened to be the fourth thrown at the batter, which lets the batter freely take first base and advances any other runner(s) that must be pushed forward. Actually, at very the beginning of the clip, you can also just barely see a bat flying away from the batter, which is presumably due to the batter throwing the bat back towards a helper that will collect and organize equipment. The batter did this since he was already aware at that point that a fourth ball had been ruled and that he was granted a free pass to first base. However, it is very easy to overlook this subtle detail, since it is only a blur seen for a fraction of a second at the beginning of the clip.

Since the batter was awarded a walk (a free pass to take first base, due to four balls being pitched), any attempted steal is nullified (essentially treated as having never happened) and the runner is anyways awarded a free pass to advance to second base (since he started at first base and was forced forwards by the batter taking first base). Thus, there was no need for the runner to attempt to reach second base before being tagged, nor did the defender need to attempt to tag him, since all of that no longer matters. Thus, given the full context, it becomes apparent that most likely the runner and defender also became aware of situation as it was unfolding (possibly due to the umpire in black informing them), and hence there was an apparently abrupt cessation of effort. At the very end, the runner extends his foot to touch the base and smiles, perhaps just to create a funny moment, rather than making an actual sneaky play.

8 Likes

I’m not any wiser, but I appreciate the pedantry.

6 Likes

I wasn’t any wiser after reading the comment @Gia mentioned, but I am now somewhat wiser after having read @yebellz’ explanation. Thanks!

6 Likes

immagine

A similar graph (even if white was always winning) , and it’s setted to “score”.

5 Likes

That looks like a L&D situation where a large group was wrongly assumed by both players to just be alive.

KataGo would be going like:

Move 90: Blunder! -50.7
Move 91: Blunder! -53.1
Move 92: Blunder! -52.8

and so on, until someone finally realizes that “Oops, I guess that group wasn’t alive after all :sweat_smile:

Edit: Or maybe you were just being merciful to the 12k

5 Likes

Where does a Go Meme about Go Memes Pedantry go?

Should there be a Go Memes Pedanty Go Memes thread? I mean, it’s just pendantry all the way down.

4 Likes


What?

1 Like

Switch to “score”! :smile:

3 Likes

Very often if the winrate graph looks like sawteeth like this, it means:

“There is a very important, game-deciding move, somewhere on the board. Whoever plays it will win the game.”

Since you are both playing irrelevant moves elsewhere and not playing the game-winning move, the AI considers that whoever is on turn will win (by playing this important move right now).

Why?

If there is a fight going on, I often find the AI’s estimated score not very helpful. What I want to know is whether I’ll win the fight, not the exact amount of points that I’ll win or lose in some imaginary endgame sequence whose prerequisite is “after this fight I’m so far ahead that I can afford to lose points”

I find more meaningful seeing that the point difference is, say, 20pts at that point and that both players were missing a big move worth more than that, than simply seeing that both were missing a big move.

If the game is really even a “1 pt mistake” could bring the winrate from 0% to 99%.
I like thinking that a move is giving “n” points to the opponent. For me it makes more sense than discussing about “1%” winrate loss.

1 Like

Do you have any evidence that if you make a move work x points when the best move was worth y=x+n points, the score announced by katago actually drops by exactly n points?

I find that very hard to believe at any stage of the game except maybe in the very very late endgame.

I don’t think I understand your question.

Best moves by AI are usually worth zero points.
My moves are always negative numbers.
When my move is evaluated “-n” usually my estimated score drops actually by about “n” points.
There can be some decimals difference but I don’t mind.

It’s just a valuation but it works quite well for me.

My ranges are quite rough:
0 to -1 wonderful move, I play like katago,
-2 to -5 not bad actually,
-6 to -10 what the hell was I thinking of?
Less than -10 oh my! That’s why I’m stuck with my rank

2 Likes

It’s hard to define what you’re trying to convey, but here’s how I believe KataGo’s score should be interpreted.

Let’s say we have position (A), and two follow up moves to bring us to position (B1) and (B2), where KataGo estimates (B1) to be scored for X points and (B2) to be scored for X+N points, then on average if we let KataGo continue the game (as both players) from position (B1), it will end up with a final score of X points, and if it were to continue (B2) it will end up in a final score of X+N points.
Importantly, KataGo will estimate (A) to have the optimal score for the player whose turn it is out of all score estimates that follow up moves have (or at least out of those follow up moves that KataGo considers). So, if (B2) is the best move according to KataGo, then (A) will also have a score of X+N, and thus (B1) could be seen as an N point mistake, if KataGo were to continue the game.

The score is a statistical estimate by KataGo, though, based on its experience playing the game: its neural net is really good at evaluating what the score is at any “normal” position if KataGo is the one continuing the game. That it is good at estimating the score is exactly why it managed to become so much stronger than pre-AlphaGo AI.
However, statistical is the key point here: there exist positions where the “on average” correct estimation tends to break down and KataGo gets confused. These are quite rare in usual games, but quite common in technical tsumego problems (like the kind of endgame problems you might find in the book by Berlekamp & Wolfe).

The important things to note is that KataGo is one of the strongest players in existence at this moment, and certainly almost any human opponent playing instead of KataGo tends to do worse than KataGo in terms of average scoring. So although it might not be the case that the score is accurate against a perfect player, it is accurate to the best standard that we have “on average”.

2 Likes

Sofiam quoted a post from the future :hushed:

5 Likes