> Eval Interpretations
> Now we enter a controversial part of my article: me trying to explain the numeric value engines push out. These are general observations based on my experience, and there are likely tons of exceptions to what I am writing. Nonetheless, I will try anyway:
> What is an Engine Evaluation?
SF16 new visible scoring does not transmit the internal leaf evaluation like when it was called classical.
So the tie to material count as the modulo part, and the crumbs around being decimals for the soup of named positional contributions, may not hold anymore (see evaluation web guide for SF, for the matrix of those names and intended components); if that is what is in the blog or that the blog depends on (I did not read yet, but flew over to see structure of it and where I might be more curious than others later). I will read later, no ETA. But I welcome such topic. I think needed a lot. SF16 now is using an internal score amplitude independent user end displayed score, liberating the internal score evolution and SF evaluation of the leaf evaluation quality do be what it needs to be, and with rescaling that its fittest parameter exploration training ground would have it explore (my current safe interpretation of that novelty, in absence of insider knowledge, needs salt).
Edit: I think my comment is a minor point about the numerical values on the left, more important take-home that the blog teaches, is about the right hand side of that interpretation progression.. Whatever the left-hand side numerical ranges landmarks should become for SF16+.
> Now we enter a controversial part of my article: me trying to explain the numeric value engines push out. These are general observations based on my experience, and there are likely tons of exceptions to what I am writing. Nonetheless, I will try anyway:
> What is an Engine Evaluation?
SF16 new visible scoring does not transmit the internal leaf evaluation like when it was called classical.
So the tie to material count as the modulo part, and the crumbs around being decimals for the soup of named positional contributions, may not hold anymore (see evaluation web guide for SF, for the matrix of those names and intended components); if that is what is in the blog or that the blog depends on (I did not read yet, but flew over to see structure of it and where I might be more curious than others later). I will read later, no ETA. But I welcome such topic. I think needed a lot. SF16 now is using an internal score amplitude independent user end displayed score, liberating the internal score evolution and SF evaluation of the leaf evaluation quality do be what it needs to be, and with rescaling that its fittest parameter exploration training ground would have it explore (my current safe interpretation of that novelty, in absence of insider knowledge, needs salt).
Edit: I think my comment is a minor point about the numerical values on the left, more important take-home that the blog teaches, is about the right hand side of that interpretation progression.. Whatever the left-hand side numerical ranges landmarks should become for SF16+.
I agree with this forum the most important part is the energy, time, ego, and curiosity.
Can you give me a like️ please?
Can you give me a like️ please?
@dboing said in #2:
> SF16 new visible scoring does not transmit the internal leaf evaluation like when it was called classical.
> So the tie to material count as the modulo part, and the crumbs around being decimals for the soup of named positional contributions, may not hold anymore (see evaluation web guide for SF, for the matrix of those names and intended components); if that is what is in the blog or that the blog depends on (I did not read yet, but flew over to see structure of it and where I might be more curious than others later). I will read later, no ETA. But I welcome such topic. I think needed a lot. SF16 now is using an internal score amplitude independent user end displayed score, liberating the internal score evolution and SF evaluation of the leaf evaluation quality do be what it needs to be, and with rescaling that its fittest parameter exploration training ground would have it explore (my current safe interpretation of that novelty, in absence of insider knowledge, needs salt).
I, in no way, get technical in this write up about how engines work. I am not capable of that. I did provide a couple of references for people who want that info. I take a more practical "how to improve with engines" approach, basically. I just think most peopl use engines too much.
> SF16 new visible scoring does not transmit the internal leaf evaluation like when it was called classical.
> So the tie to material count as the modulo part, and the crumbs around being decimals for the soup of named positional contributions, may not hold anymore (see evaluation web guide for SF, for the matrix of those names and intended components); if that is what is in the blog or that the blog depends on (I did not read yet, but flew over to see structure of it and where I might be more curious than others later). I will read later, no ETA. But I welcome such topic. I think needed a lot. SF16 now is using an internal score amplitude independent user end displayed score, liberating the internal score evolution and SF evaluation of the leaf evaluation quality do be what it needs to be, and with rescaling that its fittest parameter exploration training ground would have it explore (my current safe interpretation of that novelty, in absence of insider knowledge, needs salt).
I, in no way, get technical in this write up about how engines work. I am not capable of that. I did provide a couple of references for people who want that info. I take a more practical "how to improve with engines" approach, basically. I just think most peopl use engines too much.
@RyanVelez said in #4:
> I, in no way, get technical in this write up about how engines work. I am not capable of that. I did provide a couple of references for people who want that info. I take a more practical "how to improve with engines" approach, basically.
Well, I was reading about the score amplitude interpretation. As it might not be how our master oracle in chief will give it to us, from its programmed and engine pool optimized parameters, how it will show us its scoring anymore.
Or did I read too fast. I could not help but spill some of my salad, sorry. To help understand what has happened or will happen about SF score interpretability. I do not expect a human chess endeavor blog to explain under that carpet. I was just raising the possibility it could be pulled from under that part of the blog. And also, what will we do now, or then?
For a long while now, I have not used lichess or engine analysis except for full game graphic, for major swings, and not quantitatively.. just to find good position in past games to play again from...
> I just think most people use engines too much.
yes they do!
> I, in no way, get technical in this write up about how engines work. I am not capable of that. I did provide a couple of references for people who want that info. I take a more practical "how to improve with engines" approach, basically.
Well, I was reading about the score amplitude interpretation. As it might not be how our master oracle in chief will give it to us, from its programmed and engine pool optimized parameters, how it will show us its scoring anymore.
Or did I read too fast. I could not help but spill some of my salad, sorry. To help understand what has happened or will happen about SF score interpretability. I do not expect a human chess endeavor blog to explain under that carpet. I was just raising the possibility it could be pulled from under that part of the blog. And also, what will we do now, or then?
For a long while now, I have not used lichess or engine analysis except for full game graphic, for major swings, and not quantitatively.. just to find good position in past games to play again from...
> I just think most people use engines too much.
yes they do!
<Comment deleted by user>
<Comment deleted by user>
Great post I am really glad I learned this and there is a lot of helpful info here! Thank you.
@chessychurro said in #8:
> Great post I am really glad I learned this and there is a lot of helpful info here! Thank you.
You are welcome - glad you benefited. Grateful to you for reading and commenting.
> Great post I am really glad I learned this and there is a lot of helpful info here! Thank you.
You are welcome - glad you benefited. Grateful to you for reading and commenting.
This is the best blog article I have read on here in months. Thanks. I'll be sharing it with friends.