Kirby's Study Journal

[Kirby]

I think I understand the paradox problem better, now.

To me, this comes down to risk aversion. The expected value can give you optimal strategy if you can play the game over and over, if the probability of predictor being right is well known.

But given just a single chance to play the game, your risk aversion comes into play, including how important $1million is.

[Kirby]

Anyway, my point about AI being a black box isn't because I don't think the AI is correct. It's just that I cannot verify or understand why it is correct.

This is problematic since it's more difficult for me to extend the concepts to new situations.

Like if somehow I felt that "God wants me to invest in McDonald's stock", with no other basis- it's more like a religion than science.

AI has a good track record, so I should give it the benefit of the doubt- but for no concrete reason.

[EdLee]

Hi Kirby,

I think I understand the paradox problem better, now.

My suspicion lies with the predictor: it's magic.

[mitsun]

Often Wikipedia is a good source, but seems here it is not.

As Newcomb's problem was explained to me long ago, the predictor has a very high historical accuracy rate (think > 95% but not perfect) at guessing whether someone will take both boxes or only the opaque box. If the predictor thinks you will take both boxes, he leaves nothing in the opaque box. If he thinks you will take only the opaque box, he leaves a million dollars in it for you.

* From an expected value perspective, one can make an argument that taking only the opaque box is the right move. More than 95% of the time, you will get the million dollars.

* From a dominant strategy perspective, one can argue for taking both boxes. The predictor has already decided, after all, to put the million in the opaque box or not, so whatever the predictor has done, I should take both boxes.

As for why it is sometimes referred to as a paradox, I suppose it depends on who you ask. Some argue that backward causality is an issue (my choosing both boxes is in effect causing the predictor to act in a certain way earlier in time), which is strange and could be labeled as paradoxical I suppose.

Anyway, put me down as a one-boxer. And once I get an AI Go program on my machine, I'll probably blindly take its advice.

The paradox arises only if the predictor is 100% correct. But in that case, you have no free will, so you are fooling yourself if you think you have a choice Or maybe the predictor can violate causality (going back in time to fill the boxes after I make my choice).

Ignoring the paradox, the practical choice depends on the relative value to you of the two boxes. Change the rewards to $1 and $1M and I will take 1 box, not minding the loss of $1; change the rewards to $1M and $2M and I will take two boxes and be happy to settle for $1M.

[EdLee]

The paradox arises only if the predictor is 100% correct.

So... in various verions something like this came up: "almost 100%", "95 out of the previous 100 times," etc. Once again, the predictor is the vital point -- magic.

[Kirby]

Getting back on this silly horse.

I was "Kirby", and I played as black:



I misread the middle to be seki for the longest time. I guess my opponent thought it was too, so we just sat there while he could have killed me for the longest time.

In the end, he did. If I would have realized earlier maybe I would have had a chance, but it wouldn't have been an easy win.

[Kirby]

I have no idea how the bottom left joseki is supposed to go and I spent a good amount of time there, too.

[Kirby]

I also found out it's very hard for me to do two things in a day.

I want to study Korean and I want to study go. Hard to do both in one day, along with exercise.

Will keep trying.

[Kirby]

Today's game:



I was black, and lost the game. I played on Tygem via my phone. Since I've been having issues with doing what I want to do in the evenings, I played this game during my lunch break.

I didn't have a ton of time, but at least I got a game in.

[Kirby]

I took a quick spin on the last game and checked Elf's thoughts. I didn't spend a lot of time on each move, so it's mainly Elf's intuition.

But here's the winrate graph, which is hilarious:

As I interpret it, I generally get a lead in the beginning, and Elf thinks black will win, but then that confidents plummets when I invade the bottom left 3-3. But then when I invade the middle, Elf starts to think I might be able to live, and then black's projected winrate goes up a lot. But then I screw that up and die, and then the winrate plummets again.

Amusing.

[Uberdude]

Kirby, you are in good company, check Elf winrate for deciding game of Meijin title match Cho vs Iyama: viewtopic.php?p=238534#p238534

[Kirby]

Kirby, you are in good company, check Elf winrate for deciding game of Meijin title match Cho vs Iyama: viewtopic.php?p=238534#p238534

I guess that's how it goes sometimes!

[Kirby]

I went to the Ann Arbor Go Club again last night.

I started off by staying later at work. I got a few things done, chat with a coworker, then went to have dinner.

Here is a picture of my dinner:

After that, I went to the Espresso Royale cafe, where the club is held. Only two folks were there, both around 4K, so I watched their game. Later, Don showed up. I beat him in a handicap game, then one of the 4K challenged me. It was our first time playing, so I just took white. I got a lead early on, and kept it without much eventful stuff happening.

Later, I met a girl who a 1d said was probably about my level. I had beaten the 1d when we played, and he said she usually beats him.

I think she's a university student.

Anyway, she took a very loooong time for each move. I have played that way at tournaments, but for club games, I don't typically play that slowly.

Anyway, due to the slow pace, the cafe closed before we finished the game.

She played an interesting move at the end, but I think I have a decent chance at winning - she has few points, and didn't attack me much in the center.

Here is the game - I was black:



All in all, I had a fun time. It was a good social event.

[Kirby]

Reviewed how the week went, and I am going to take a break from studying go for a bit - more interested in Korean right now. Will probably still go to the go club on Thursday, and if so, I might post the game then.

[EdLee]

Hi Kirby,

Here is a picture of my dinner:

Korean restaurant ?