influence concensus map

[jaca]

Even if they can't say why.

[Elom]

Everyone makes decisions with some combination of intuition and analysis. Some are more intuitive; some more analytical.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4519675/

It's probably the case that the best Go players are more analytical than intuitive, as reading isn't something you can afford to not do.

But what to read??

i only ever speak for myself, and everything i say is only my opinion, and in this case my opinion is that i would like some help in figuring out what to read...

Perhaps you could try reviewing a long game backwards after reviewing it forwards. It will likely take quite a while, but the fewer options towards the end of the game should make it easier to be thorough at the start.
It might also show the moves you intuitively select as candidate moves without being influenced by previous moves (neural nets essentially read the same way as humans).

All About Thickness by Ishida Yoshio

Kiseido sell an e-book download of All About Thickness
http://www.kiseidodigital.com/kdb1-toc.html

Ishida was called "the computer" by his peers, so who better to guide us to a better understanding of influence!

...

1974 #1— the Computer
2017 #1— the Computer

[Elom]

On the more general side, a reservation I've always had is how is an influence map, most algorithms for which I've seen are basically some smooth proximity decay sort of function, going to cope with the non-smooth nature of go (stones have to be on the lines) and the way shifting one line can make a big difference to how much that player controls that area.

IMO, you hit the nail on the head.

For example what would an influence map say is going on in these corners?

I thought it might be fun to show the maps for my influence function. I dipped my toe into influence functions in the early 2000s. I found a function that produces correct territory estimates in certain well defined situations. To my surprise, when applied to the whole board, it produced a komi estimate of around 8. Surprisingly close. So I thought it might make a reasonable first approximation, despite its oversimplification and other flaws.

. . .

Did you test it on 9x9? It would be interesting to see the results .

[Vio]

A wall has different value of influence according to his own life status. A wall with eyes, or an extension or a ponnuki to take some examples are much more influent as a simple line of stones.
That's so fundamental that I don't see how to modulate a punctual influence value without implying much more concepts like life and death...

[jaca]

I don't see how to modulate a punctual influence value without implying much more concepts like life and death...

agree 100%.

in the beginning, we all marvelled at the new wunderkids

and then some people started to worry about Go engines going haywire or being manipulated by megalomaniacal oligarchs. I don't mean Michael, who is a shining star that even Neil could see and an example to us all - how he manages to keep a straight face with the antics of that middle-aged actress pretending to be a 13 year-old schoolgirl cavorting around and grinning like a Cheshire Cat at the camera i will never know!

Ishida set what many believe to be an impossibly difficult challenge for Go programmers, because people don't think like computers like him

but have no fear, Ishida+Redmond - inspired https://github.com/gogre/irgobot is here to save the day and tell you the plain facts of life, death and influence:

the next episode will compare Michael Redmond's analysis of a position in game 9 of Zero vs Master with what irgobot makes of it, focussing on estimating territory in the early part of the midgame.

irgobotgit.png (455.73 KiB) Viewed 6679 times