How Checkers Was Solved(theatlantic.com)
theatlantic.com
How Checkers Was Solved
https://www.theatlantic.com/technology/archive/2017/07/marion-tinsley-checkers/534111/?single_page=true
35 comments
It's good, really, that this article focussed on Tinsley. I saw a talk by Schaeffer about ten years ago, just as he was finishing up solving checkers, and it was an astonishingly depressing talk. He invested his whole life—decades—as I recall his marriage failed partially as a result of his obsession with this—on what, in the end, was a lookup table. There was a lot of neat engineering he had to do along the way (storage, parallel work) but that's not what he cared about: he cared that he had solved checkers. Were there any insights about patterns that could be induced, maybe to teach humans how to do analysis or to learn some interesting mathematical fact about the graph structure of the board? No—you just see what the board state is, and look it up in the database to tell you the next move. I can't even tell you how demotivational this speech was.
Back in my college days I had a summer job as an intern at a maximum security psychiatric hospital. My main ward was the admittance ward which was essentially where men accused of a crime were evaluated if their fitness to stand trial was in question. However, there was this one 20 something who had been there for years. He was accused of a horrific crime and had been diagnosed with an IQ of 65. This kid loved to play checkers. I started playing with him and over the course of the summer probably played 100 or so games with him. He beat me the first 50 or so times I played him and my assessment was that he had every possible board setup memorized and knew the right move for each of them. Whenever I made a mistake, he capitalized instantaneously. After the 50th or so game, I had memorized enough setups that I could split games with him.
Sometimes I fear that this will be the end result of brain research. A vast table that says: when these neurons are activated in this pattern, it means happiness. When these neurons are activated in this pattern, it means the brain is perceiving a tree.
We will understand the brain and consciousness and it will be completely unenlightening... :(
We will understand the brain and consciousness and it will be completely unenlightening... :(
There's no way to map "what is the meaning of life?" to a table, so you'll never have that problem. As long as our universe and existence isn't 100% solved, the human brain will never be solved!
It's been a few years, but I recall there being research into 100% mapping out crab or ant behavior and it was getting to a point of being quite accurate.
Mammal brains have a few extra layers of complexity, but none of those layers involve magic. If ever we get around to "solving" one animal's brain, it shows that others can be as well.
Mammal brains have a few extra layers of complexity, but none of those layers involve magic. If ever we get around to "solving" one animal's brain, it shows that others can be as well.
Maybe we'll be able to finally prove that free will doesn't exist.
It's not that it does, or doesn't exist. It's that the whole common sense notion is literally incoherent.
Thank you! I agree. "Free will" doesn't even mean anything.
Indeed. I've always found this a thoughtful discussion:
https://www.rep.routledge.com/articles/thematic/free-will/v-...
Religions and denominations of the same religion won't even agree on the nature or existence of free will. Why everyone else is so concerned about the same, I don't understand. It's more philosophy than science in most discussions. If acting as if your actions had internal agency offers better chances of survival, that's a good enough fudge for nature.
There's no way we'll be able to do that until we have an agreed-upon definition of free will. Good luck with that. We can't even come up with rigorous definitions of many common, related words: intelligence, thinking/thought, life. Heck, we can't even define pornography!
All of these ultimately boil down to "I know it when I see it". This sort of definition is not good enough for scientific inquiry. For that reason, scientists tend to avoid such questions.
All of these ultimately boil down to "I know it when I see it". This sort of definition is not good enough for scientific inquiry. For that reason, scientists tend to avoid such questions.
> It's been a few years, but I recall there being research into 100% mapping out crab or ant behavior and it was getting to a point of being quite accurate.
We are nowhere near this point. There are specific, intentionally constrained behaviors in model animals such that we can monitor the activity of specific sets of neurons and make predictions (meaning anywhere from better-than-chance to perfect accuracy) of behavioral output. This is far from "mapping out behavior" for an entire organism.
As for the GPP about a comprehensive look-up mechanism from all possible states of dynamics in the human brain to semantic descriptions of mental states: the problem is that trying to implement such a mechanism within the physical and thus computational constraints of our world will result, with high likelihood, in something that functions rather like a brain.
We are nowhere near this point. There are specific, intentionally constrained behaviors in model animals such that we can monitor the activity of specific sets of neurons and make predictions (meaning anywhere from better-than-chance to perfect accuracy) of behavioral output. This is far from "mapping out behavior" for an entire organism.
As for the GPP about a comprehensive look-up mechanism from all possible states of dynamics in the human brain to semantic descriptions of mental states: the problem is that trying to implement such a mechanism within the physical and thus computational constraints of our world will result, with high likelihood, in something that functions rather like a brain.
>none of those layers involve magic
This is a bold claim. How do you know? The brain could be magical.
I think everyone on this forum who is not a Christian can agreee that the Christian God is in some sense "magical". Further, Genesis 1:27 says "So God created man in His own image; in the image of God He created him; male and female He created them."
While this falls short of saying that, like God, humans are magical, to date we have no proof they are not.
In that case the brain might be, for example, a magic antenna to another plane.
See my comment here:
https://news.ycombinator.com/item?id=14770230
The choices are that the brain is magical or that strong AI that can do anything a human brain can do (make judgments, parse text and images, underdtand culture, communicate intelligently with humans), is in some sense inevitable.
You cannot simply discount the possibility of magic. You must let people who are committed to denying the future have their out.
If I were speaking with Lord Kelvin I would allow him the possibility that birds are magical and not subject to the laws of physics, whereas any machine humanity might ever build is - so that it is certainly possible that in 100,000 years, no one will build a heavier-than-air flying machine.
The essence of flight might be magical. So, too, the essence of thinking.
This is a bold claim. How do you know? The brain could be magical.
I think everyone on this forum who is not a Christian can agreee that the Christian God is in some sense "magical". Further, Genesis 1:27 says "So God created man in His own image; in the image of God He created him; male and female He created them."
While this falls short of saying that, like God, humans are magical, to date we have no proof they are not.
In that case the brain might be, for example, a magic antenna to another plane.
See my comment here:
https://news.ycombinator.com/item?id=14770230
The choices are that the brain is magical or that strong AI that can do anything a human brain can do (make judgments, parse text and images, underdtand culture, communicate intelligently with humans), is in some sense inevitable.
You cannot simply discount the possibility of magic. You must let people who are committed to denying the future have their out.
If I were speaking with Lord Kelvin I would allow him the possibility that birds are magical and not subject to the laws of physics, whereas any machine humanity might ever build is - so that it is certainly possible that in 100,000 years, no one will build a heavier-than-air flying machine.
The essence of flight might be magical. So, too, the essence of thinking.
> the Christian God is in some sense "magical" ... In that case the brain might be, for example, a magic antenna to another plane.
If those things were shown to really exist, then they would be part of what we understand to be our reality. So they wouldn't be magical, it's just that our understanding of what makes up reality could be incomplete.
If those things were shown to really exist, then they would be part of what we understand to be our reality. So they wouldn't be magical, it's just that our understanding of what makes up reality could be incomplete.
OK, but they would be non-physical. And that would mean that a materialist understanding of humans would be incomplete. It might also mean that we cannot ever build a human-equivalent AI.
well I meant it to be far-fetched. If we had no access to the other plane, there's not much we could do to understand the brain calculation happening there.
whereas if it's not an antenna then it's just 3 lbs of meat.
whereas if it's not an antenna then it's just 3 lbs of meat.
Ultimately that boils down to a semantic distinction, and not one advancing the state of knowledge. Calling some phenomenon magical or mysterious just means you don't want to study it, not that it can't be studied.
I don't believe in the existence of magic so at a practical level I have to agree with you.
For posterity I would like to mention that my reference to a magical God is made ironically: the brain is driven by simple laws of physics and not attached to any ethereal plane! There is no magic involved.
Clarke's Third Law: Any sufficiently advanced technology is indistinguishable from magic.
That kind of understanding has a philosophical limit to what it can describe though - consider that an alien being, with an alien physiology would still experience what we would call pain, without the c-fibers firing.
https://en.wikipedia.org/wiki/Philosophical_zombie
https://en.wikipedia.org/wiki/Philosophical_zombie
What a coincidence: I'm a philosophical zombie.
(Well... prove I'm not!)
(Well... prove I'm not!)
Ignoramus et ignorabimus?
Your story suggests value in another article focussing on the personalities of Tinsley and Schaeffer. Both obsessives, both pre-eminent in their domain, but in quite different ways - and with different consequences on the rest of their lives. Tinsley's variety of obsession seems much healthier than Schaeffer's.
A truly fascinating story. I recall reading another version with an additional detail: after Checkers was solved Schaeffer evaluated every recorded game Tinsley had played in his life - and determined that there were only a handful of moves that he ever made in competition that were not perfect (in the sense of perfect play). The vast majority of the time, if you were playing Tinsley, you literally could not possibly win.
Having read Jonathan Schaeffer’s great book One Jump Ahead, I couldn’t help feeling that this article was a (well-written) book report which just barely mentions its primary source. Nice summary, but read the original. I haven’t read the second edition which apparently includes the follow-up story of solving the game, but I found the first edition very compelling. I wonder if the reporter was following the recent Deepmind/U of Alberta news and then came across the checkers story. Disclaimer: I was in Jonathan’s compiler course 30 years ago, prior to Chinook.
A great recent documentary on Checkers is King Me (trailer here; https://www.youtube.com/watch?v=Eh8wyKabQiY). I say that as someone who is terrible at the game and doesn't enjoy playing it.
One of the players it features is Ron King, who is currently one of the strongest Checkers players. He holds the record for playing (and beating) 385 people simultaneously, though I believe he was beaten by Tinsley. Also his trash talking is fantastic.
One of the players it features is Ron King, who is currently one of the strongest Checkers players. He holds the record for playing (and beating) 385 people simultaneously, though I believe he was beaten by Tinsley. Also his trash talking is fantastic.
There is a fantastic story on the first checkers AI here: http://relprime.com/chinook/
That was one of the greatest stories I've ever heard
If you like this story I suggest you take a look at the book "One Jump Ahead" by Schaeffer for a lot more in depth view!
There's also Blondie24[0], which's got a pretty humorous backstory to its namesake.
[0] https://www.amazon.com/dp/1558607838/
[0] https://www.amazon.com/dp/1558607838/
can second this recommendation. good, quick reading.
You might like The Master of Go - humans, but a similar arc of the master in fading health facing an opponent for whom the game means something quite different.
I had no idea that there was once a real-life incident reminiscent of the film War Games!
[Solving chess] took Schaeffer harnessing computers all over the world, drawing on and expanding his expertise in parallel computing. He conscripted machines everywhere from Switzerland to Lawrence Livermore National Laboratory, a major Department of Energy facility that often deals with nuclear weapons.
“There was somebody else there [running a program] called BOMB and I was running checkers programs,” Schaeffer told me. “It was a very strange situation. Security should have been concerned.” And they were. They paid him a visit after discovering gigabytes of data flowing out of a national lab to Edmonton, Alberta.
(edit: add the remainder of the second paragraph.)
[Solving chess] took Schaeffer harnessing computers all over the world, drawing on and expanding his expertise in parallel computing. He conscripted machines everywhere from Switzerland to Lawrence Livermore National Laboratory, a major Department of Energy facility that often deals with nuclear weapons.
“There was somebody else there [running a program] called BOMB and I was running checkers programs,” Schaeffer told me. “It was a very strange situation. Security should have been concerned.” And they were. They paid him a visit after discovering gigabytes of data flowing out of a national lab to Edmonton, Alberta.
(edit: add the remainder of the second paragraph.)
Awesome story and well written.