It's approaching Game 3 of the AlphaGo vs. Lee Sedol match. There's a distinct possibility that AlphaGo will win the match, and close out its victory. Lee Sedol, who just last week displayed extreme confidence, has changed his tone:



Is Go still a "big deal"?
Needless to say, even three days after Lee Sedol's initial loss, I am still depressed. Illuck's post was somewhat encouraging, but nonetheless, I'm still saddened. I've been reflecting for awhile, and I wondered to myself, "Why do I feel this way?"

I've come to the conclusion that it might be the case that seeing AlphaGo's apparently easy victory somehow belittles the complexity of Go. The optimistic perspective is that computer AI has advanced to a complex level, capable of achieving super-human level strength. The pessimist in me, though, feels that AlphaGo's success somehow indicates that - maybe playing Go at a high level isn't really a "big deal".

Thinking about it more thoroughly, I'm (a little bit) comforted.

The Old Days
I'm reminded of studying Calculus back in high school. When I first learned about integration, solving problems seemed like a "big deal". Eventually, I got the hang of it, but it felt like I was doing something cool. Nowadays, you punch the problem into an online calculator, and you have an answer. An elementary school kid can do that. Sure, I can go through the tedious work of doing things the "old fashioned" way; but the end result is that a 3rd grader can probably achieve the same result with some online app.

Transportation technologies - horse and buggy, trains, cars - may be the same way. Maybe hundreds of years ago, it was a big deal if you could run faster than anyone else. It could be a means to get someplace fast - or a way to survive. But these days, if you want to go downtown, just hop in a cab, and you'll be there in 5 minutes.

What about today?
These days, technology is everywhere. Is there still a place for doing things the old fashioned way? Why do integration by hand when you can punch the problem into a calculator? Why run when you can take a car? Why learn karate when it might not practically help you against someone with a powerful enough weapon? In terms of actual results for the given problem, I'm at a loss to find an answer. I can solve math problems more quickly with calculators. I can get from point A to point B faster in a car. Does the old fashioned way have its place?

I suppose that it does in this: self improvement. Learning how to solve math problems without a calculator improves your mind. It can help you to solve more difficult problems. Running has a number of health benefits. It can help you to lose weight; it can increase self-discipline; it can improve your attitude. Riding in a car doesn't do this for you.

Hard work, regardless of the efficiency of the result, gives a feeling of accomplishment. It gives something to strive toward. It gives a reason for living.

Punching a complicated math problem into a calculator may give you a quick and easy result. But the experience of obtaining that result on your own; the experience of working toward a solution; the experience of failing, improving upon that failure, and trying again... These things are obtained the "old fashioned" way.

Looking Ahead
Go is a beautiful game. Regardless of how far computers have come... It is a complex game. Looking back at the hours I've spent studying the game, playing the game, and the years I've tried to improve... That's proof.

Humans may never beat computers at Go again. On a more personal level, I may never beat a professional human player - even if I study hard for the rest of my life. In terms of real results, if I can never beat a professional player; if the top professionals can never beat AlphaGo... Does it belittle our effort in trying?

Finally, I can realize to myself with confidence: No, it absolutely does not.

Lee Sedol is much further than me on the path toward excellence in Go. AlphaGo may be further along that path than Lee Sedol. But neither Lee Sedol nor my own efforts are in vain. Regardless of who is further ahead, we are all on this path. And this path improves each one of us through our effort. Self improvement and hard work are not "old fashioned". They are the essence of being human. Efficient or not, through our efforts we become better.


Good luck, Lee Sedol
With this, I will say: Good luck on game number 3, Lee Sedol. It might have been humbling to lose. But hold your head up high. The years of effort you have given to this game are not in vain. The hours you spent solving problems that your father gave to you were not a waste.

That effort and that work is intrinsically commendable. And nobody can ever erase it.

I believe in you. Hold your head up high, and play with pride.
---
[1] https://www.cyberoro.com/news/news_view ... num=521390

_________________
be immersed

Well said. I think we can take comfort from the fact that chess is alive and well. The 100m sprint is still exciting at the top levels, even if a machine could be equipped with "legs" that could run faster than humans. Also, I think a lot of the satisfaction of games is being part of communities they are played within, seeing improvement in one's own game, and the rivalries with other people or communities.

Good point. Incidentally, I heard today about "Centaur Chess" (https://en.wikipedia.org/wiki/Advanced_Chess), where a human/computer team can compete.

In some tournaments, human+computer teams were not only more skilled than human-only players, but also more skilled than computer-only players.

_________________
be immersed

I would be more likely to want to play in person games if pro level computer software was widely available...oh, who I am kidding, I would probably be sitting in front of the computer taking a handicap from the pro level program and trying to learn something from my losses. On a serious note though, I wonder if chess had an increase or decrease (or neither) in face to face games after pro level software was available to all players.

Nice post Kirby. For me, I always knew it was a matter of time (certainly within my lifespan) before AI would take on Go, but to me it doesn't matter and doesn't change how I feel about the game. All that matters to me is that Go has blessed me with the opportunity to meet so many interesting people of all cultures and backgrounds, many of whom I consider to be friends. When I look at it that way, it doesn't matter whether or not the game or hobby has been beaten by AI. From my Go trip in China to US Go Congresses, from weekly Go meetups and even the AlphaGo viewing parties going on right now on at the SGC, my life has only been enriched with experiences of which Go was the mean to such ends. And for that, I'm thankful.

Plus I still have StarCraft, which probably buys me a year or so, maybe less .

AlphaGo is exciting but also makes me sad, for weird reasons. First, it is now too late for me to be involved in the first computer program that defeated a top professional. Now, it's not as if that was really a serious goal I had pursued and was just beaten to the punch---that would have been okay---it's just that that door is closed. Plenty of other doors are still open, though. It is also too late for me to invent the light bulb, but I'm not losing sleep over that. It is, however, a gentle reminder that says to me: what are you waiting for?

Is the perceived complexity of go something that attracted me? Somewhat. I don't do Sudoku because I know I can write a program that can solve those better than I can manually, so it seems silly. It's too tractable.

The beauty of go is unchanged, in my mind, even if there are programs that can beat top professionals. Its inherent complexity it not reduced, either. It did not suddenly become Tic Tac Toe overnight.

In coming years, I expect subtle changes. Maybe phones and tablets will be banned at tournaments, so we'll go back to recording on paper (or in human brain). Maybe some new josekis will show up, but that would have happened anyway. Maybe we'll just have to get used to the idea that many of our online opponents are cyborgs (which may not be a wholly bad thing if it brings people back to the clubs).

Access to technology can also make people arrogant. People walk into doctor's offices thinking they know more about their condition than the doctor because they've spent hours on the internet researching it. Sometimes they are right and sometimes they are misinformed, but either way it's a kind of arrogance.

A player a few stones stronger than me could point out a mistake in my game or provide a useful opinion. Will I then be tempted to verify it on a computer engine? Probably. I'm already tempted to verify those things with yet stronger human players, especially if I am skeptical, so in a way there's nothing new there. But do we increasingly become slaves to computer analysis?

All, in all, back my first paragraph in this post, if I'm honest with myself.

Well, to me, playing go has always been about playing people in many senses of that phrase. I already lose to programs all the time, so a new superbot doesn't really change much for me on a personal level. So what if there is one more thing that machines/computers/technology can do better than I can? It's a complex world and being the absolute best and something is pretty darn tricky.

So back to just me and go and computers, this actually won't change much for me. My goals of personal improvement will still stand. I'll keep playing people online and at clubs and (hopefully) enjoying some aspect of each game and encounter. I still have no illusions about becoming the best player around (even in my house, much less my city, state, country, or the planet). If anything, I think it is still impressive that any human can stand up against all the processing power represented by AlphaGo and play a close game. That's a true testimony to Lee Sedol's skills...

Bruce "UnBionic Man" Young

_________________
Currently reading: Plutarch, Cerebus, and D&Q 25th Anniversary

Watch out! I heard somewhere that DeepMind might be tackling StarCraft next

_________________
be immersed

It may come from seeing Go as part of your identity, rather than simply a game that you enjoy. I am guessing this is more of a tendency in the West, where Go is a niche game still, and there can be a certain sense of pride in being a Go player. Speaking personally, I liked the idea of Go having a bit of a mystical quality to it (something along the lines of Sai seeking to play the divine move). I think AlphaGo defeating Lee perhaps demystifies Go to a certain extent in our imaginations and may threaten our identity and pride.

Thanks, Kirby. Your post echoed many of my thoughts.

AlphaGo has not just won some games against Lee Sedol, but it made it seem like beating Lee Sedol is not something extraordinary, but rather simply a matter of processing things correctly. Playing go well has been reduced from a superhuman ability to an algorithm, and I too am saddened. My conclusion too however is not that Lee Sedol's greatness is diminished, but rather that well implemented algorithms have tremendous power.

Like you, I have seen many domains of human expertise fall to technological advancement, and I am left wondering and worrying whether our acquiescence to the new norm will cause more harm than good. If it has become no longer necessary that we perform skills with expertise, is our worth diminished when we abandon these skills? What I observe however in myself and others however is that the drive to become skillful remains strong, and that the technological advancements enable us to as humans to push the boundaries of our abilities further. This is not a bad thing.

The developers of AlphaGo have shown us that our perceptions of what is doable must be re-evaluated. The complexity of go has taken some people such as Lee Sedol to the limits of human ability, and we have been shown that this limit can be overcome by the crafting of new tools. In the face of other challenges that humanity seems challenged to master, this can only be good news.

P.S. Nobody is wondering why nobody boxes against machine guns.

_________________
Patience, grasshopper.

I wouldn't be too upset. It is only a game. What you should be worried about is what this presages for the future of the human race. If a computer can master that imagine what it will be able to do in another 40 years, well within the lifetime of most people here. It is my opinion that life will not be worth living with computers in total control of everything.

_________________
Still officially AGA 5d but I play so irregularly these days that I am probably only 3d or 4d over the board (but hopefully still 5d in terms of knowledge, theory and the ability to contribute).

What about the space station? Computers are in total control there, and people live there for extended periods of time.

I think computers are groovy, yeah!

There have always been go players stronger than me. This is just one more.

_________________
Help make L19 more organized. Make an index: https://lifein19x19.com/viewtopic.php?f=14&t=5207

Thanks for that post, Kirby!

My short answer to the question with which you titled it, interpreting the “our” as referring to our species, would be:

“Not the least, quite the contrary: it shows just how strong our efforts regarding Go are!”

_________________
“The only difference between me and a madman is that I’m not mad.” — Salvador Dali ★ Play a slooooow correspondence game with me on OGS?

This doesn't sadden or depress me at all (though I can see how it would for some). It doesn't change anything for me. It will likely change the rules for tournament go somewhat. But I don't play competitively any more. I just simply enjoy the beauty of the game, and will continue to do so.

As a mathematician, I don't think calculators have degraded our capacity to think. They may have degraded our capacity to calculate, just like cars have degraded our capacity to run but enhanced our capacity to get somewhere.

In 1998 I made a private thesis in number theory, next to my official work. I was investigating a certain pattern. My initial hypothesis, based on insights in the lower number behavior was partly confirmed but also denied then refined, thanks to a program I wrote in Maple. I could have done the calculations by hand but I would have spent weeks to get where Maple got me in one day. I could spend those weeks instead on thinking about the theory that encompassed the larger number outcomes.

Galileo had a very rough insight in certain laws of physics, because he did not have algebra at his disposal. Today this is unthinkable but he really didn't have it and needed to formulate his thoughts in long sentences, which were overlapping. His energy was not well spent. Surely when algebra arrived there must have been a few romantics who lamented the lost art of formulating mathematical truths in long sentences. Today, millions of people can think on a higher level of abstraction than Galileo did, thanks to algebra. Of course, there are billions of people still dumber than Galileo was. Collective progress is not uniform.

I don't know what Alphago is going to bring us in terms of new insights in Go. I'm sure though that scientists will be able to crack formerly unsolvable problems with new data aggregations or even languages or other interfaces to the problem at hand, driven by new discoveries through AI. And still there will be lots of people who don't recognize a group in atari.

I just came across an article in Wired entitled The Sadness and Beauty of Watching Google’s AI Play Go - here is the link http://www.wired.com/2016/03/sadness-be ... i-play-go/

I found the part at the end about Fan Hui very interesting. He was able to improve his play from what he learned in the games against AlphaGo. I think that is encouraging.

I may have to retract what I just said... when I saw people changing their Facebook profile pictures to Lee Sedol, I must admit, I got a bit emotional