lichess.org
Donate

Analysing Armageddon Games: Time Usage, Scores and Sharpness

White starts with a 3-minute time advantage, but Black closes the time gap from move 1

While the absolute time difference might be reducing, I think the relative time advantage might actually increase. Simply put, having 5 minutes versus 3 minutes ( eyeballing ~ move 25 in the graph) is a more favorable then the initial 10 minutes versus 7. I would speculate that there is some critical " low time alarm* " equivalent in OTB games, where the top players will have an increased blunder chance.

>White starts with a 3-minute time advantage, but Black closes the time gap from move 1 While the absolute time difference might be reducing, I think the relative time advantage might actually increase. Simply put, having 5 minutes versus 3 minutes ( eyeballing ~ move 25 in the graph) is a more favorable then the initial 10 minutes versus 7. I would speculate that there is some critical " low time alarm* " equivalent in OTB games, where the top players will have an increased blunder chance. * https://github.com/Antiochian/chess-blunders/blob/main/README.md

hmm now I wonder what a chart of draw-probability vs time-control looks like for super gms ...

hmm now I wonder what a chart of draw-probability vs time-control looks like for super gms ...

I was unaware of asymmetric time controls.
I do have many popping up questions.
Is there subjective recounting of what that does? Or third-person report or review with hypotheses.
From their subjective accounts, does it make a difference to thinking?
How long has this type of thing existed?
Should I google Armageddon (minus “flat earth” if that might correlate with some bubbles of the internet, kidding, I can just add “chess”)

I would appreciate your input as well. Maybe filtered pointers if that is on your radar.

You mentioned that there is some variability within the time ranges. Controls with a longer duration? Where speed of execution is less important for planning? Possibly, that would dilute the interesting effects on differential thinking strategies. (debating).

What were the reasons that led to the creation of such categories?

Some of my questions might not be answerable. That is ok. Take your pick.

I was unlikely to be interested. But given the previous blogs' creative approaches, I thought I would still allow the distraction from my goal of no reading in chess diet. Just a pressure to avoid tilt of the chess reading kind. I did puzzle sprees (going back there).

And now, this:
I wonder if engine tournaments have considered such categories. They may consider exploring the possibility of allowing slower versions to surpass faster versions in different dimensions of the optimization objective (like covering of chess position wilderness, from reduced conservative compromising choices in their exploration, either in RL or in partial exhaustive tree preemptive pruning).

On the sharpness thing, I think I missed the boat before. I do not follow well. For I have no real intuition, or my intuition interferes about what sharpness might be. Please rehash an explanation of what it means, in your intuition, now that you have explored some tangible algebraic candidates based on engine scores tools available.

Would position difficulty or "criticality" in some sense be good enough to look catch the graphics story? (if you can explain which senses, even better)

I was unaware of asymmetric time controls. I do have many popping up questions. Is there subjective recounting of what that does? Or third-person report or review with hypotheses. From their subjective accounts, does it make a difference to thinking? How long has this type of thing existed? Should I google Armageddon (minus “flat earth” if that might correlate with some bubbles of the internet, kidding, I can just add “chess”) I would appreciate your input as well. Maybe filtered pointers if that is on your radar. You mentioned that there is some variability within the time ranges. Controls with a longer duration? Where speed of execution is less important for planning? Possibly, that would dilute the interesting effects on differential thinking strategies. (debating). What were the reasons that led to the creation of such categories? Some of my questions might not be answerable. That is ok. Take your pick. I was unlikely to be interested. But given the previous blogs' creative approaches, I thought I would still allow the distraction from my goal of no reading in chess diet. Just a pressure to avoid tilt of the chess reading kind. I did puzzle sprees (going back there). And now, this: I wonder if engine tournaments have considered such categories. They may consider exploring the possibility of allowing slower versions to surpass faster versions in different dimensions of the optimization objective (like covering of chess position wilderness, from reduced conservative compromising choices in their exploration, either in RL or in partial exhaustive tree preemptive pruning). On the sharpness thing, I think I missed the boat before. I do not follow well. For I have no real intuition, or my intuition interferes about what sharpness might be. Please rehash an explanation of what it means, in your intuition, now that you have explored some tangible algebraic candidates based on engine scores tools available. Would position difficulty or "criticality" in some sense be good enough to look catch the graphics story? (if you can explain which senses, even better)

@TBest said in #2:

I would speculate that there is some critical " low time alarm* " equivalent in OTB games, where the top players will have an increased blunder chance.

  • github.com/Antiochian/chess-blunders/blob/main/README.md

Interesting. The stress psychology, I think people might have their own variations of flight or flight subjective threshold about time left and odds perception. How far can this noise in the cogs trend of stress chess keep going (I only find the asymetry here theoretically interesting, would not touch with 10 foot pole... even the symetric ones.. my brain is too small, this would be like freezing in front of the class room, and no time to look around, and not trusting my impulses etc..). but glad to know it is a thing. now engines have no lack of imagination excuses for not diversifying their optimiztion or improvement goals if they want us to keep adoring their purpose shift as accuracy standards.

@TBest said in #2: > I would speculate that there is some critical " low time alarm* " equivalent in OTB games, where the top players will have an increased blunder chance. > > * github.com/Antiochian/chess-blunders/blob/main/README.md Interesting. The stress psychology, I think people might have their own variations of flight or flight subjective threshold about time left and odds perception. How far can this noise in the cogs trend of stress chess keep going (I only find the asymetry here theoretically interesting, would not touch with 10 foot pole... even the symetric ones.. my brain is too small, this would be like freezing in front of the class room, and no time to look around, and not trusting my impulses etc..). but glad to know it is a thing. now engines have no lack of imagination excuses for not diversifying their optimiztion or improvement goals if they want us to keep adoring their purpose shift as accuracy standards.

@dboing said in #4:
The idea behind armageddon games is that White gets more time, but Black has draw odds. White gets the win if they win and Black gets the win if they win or draw. So every game is decisive and they are sometimes used to break ties.
As I said, there is no standardised time format and in the online tournaments White starts with 10 minutes and the players bid with the time for Black. The lower bid wins and gets to play Black with the time they have bid. So basically every game has a different time control.

The sharpness is calculated by using Leela Chess Zero's win, draw and loss scores and the idea is that a position with a higher draw score is less sharp than a position where both sides have winning chances. The idea is that a position where both sides have a 50% winning probability would have an eval of 0.00, the same as a position with a 100% draw rate. But in practice, they are very different positions and the sharpness score tries to capture that.

@dboing said in #4: The idea behind armageddon games is that White gets more time, but Black has draw odds. White gets the win if they win and Black gets the win if they win or draw. So every game is decisive and they are sometimes used to break ties. As I said, there is no standardised time format and in the online tournaments White starts with 10 minutes and the players bid with the time for Black. The lower bid wins and gets to play Black with the time they have bid. So basically every game has a different time control. The sharpness is calculated by using Leela Chess Zero's win, draw and loss scores and the idea is that a position with a higher draw score is less sharp than a position where both sides have winning chances. The idea is that a position where both sides have a 50% winning probability would have an eval of 0.00, the same as a position with a 100% draw rate. But in practice, they are very different positions and the sharpness score tries to capture that.

@jk_182 Thanks for that. Yes, the ideas are what I understand better or can keep in memory, as that comes with some explicit logic to glue the parts, and I think long-term memory of mine is fond of that, slow learner that it is otherwise.

I see that Armageddon is tied to other categories; that part I might have to let simmer a bit. But you did give me food.

The sharpness idea is now a lot clearer. You may have already explained it that way, but such rephrasing seemed to be sufficient consolidation. I still would call it sharpness_WDL or sharpness_LC0, so we can have chess community parallel terminology not colliding in case other sharpness "entendre(s)" (plural) woud collide. But your explanation does resonante about a quantity I do find has some characteristics going for it.

And the natural language meaning of sharpness, might even apply. As draw-ish statistic on a collapsed information single linear combination of W D and L, having 2 humps versus a one hump in profile, means something got split within the same surface under the center symmetric profile. A knife or something split the thing. Does that compile? loosely (the point is rephrasing for myself, you be the coach).

@jk_182 Thanks for that. Yes, the ideas are what I understand better or can keep in memory, as that comes with some explicit logic to glue the parts, and I think long-term memory of mine is fond of that, slow learner that it is otherwise. I see that Armageddon is tied to other categories; that part I might have to let simmer a bit. But you did give me food. The sharpness idea is now a lot clearer. You may have already explained it that way, but such rephrasing seemed to be sufficient consolidation. I still would call it sharpness_WDL or sharpness_LC0, so we can have chess community parallel terminology not colliding in case other sharpness "entendre(s)" (plural) woud collide. But your explanation does resonante about a quantity I do find has some characteristics going for it. And the natural language meaning of sharpness, might even apply. As draw-ish statistic on a collapsed information single linear combination of W D and L, having 2 humps versus a one hump in profile, means something got split within the same surface under the center symmetric profile. A knife or something split the thing. Does that compile? loosely (the point is rephrasing for myself, you be the coach).

also note link to sharpness article from word 'sharpness' in blog. Armageddon is interesting but not really chess. Forcing good players to blunder through +1s increment is some steps down for chess. Rather 960

also note link to sharpness article from word 'sharpness' in blog. Armageddon is interesting but not really chess. Forcing good players to blunder through +1s increment is some steps down for chess. Rather 960