There must have been several threads on Go already, but I can’t search for them because of the 4-character minimum. So…
What do I need to know, what do I need to keep in mind when learning Go? I see how complex its strategies can get on advanced levels, but what do you recommend for beginners?
I’ve tried playing it on Facebook, but it’s hard to get worthy opponents. A few moves in and they give up by letting their moves time out after only a few moves. Only today did I play a game that filled up the whole 13×13 board, and then the opponent timed out on their move. It counts as a win for me, but I’m not looking for wins, I’m looking for ways to become more skilled at it.
I don’t have a board of my own, so where can I find a good Go game either online or IRL?
The (free) playok.com website has a good interface and offers human-vs-human Go. I’ve played only a few games there (I mostly play Bridge and Backgammon there) but the players have been better than me and polite.
There are free computer-vs-human downloadables but I’ve not tried any.
Read books, perhaps starting with Go for Beginners. That book advises that you play your first several games as quickly as possible, without thinking, just to get a feel for the game.
Which download do you have? I’ve been looking for a decent, iPad compatible app for GO and would love a recommendation. Ideally, it would offer both human vs computer and human vs human play modes. Free is nice but I don’t mind dropping a few bucks on a quality product.
I think I’ll try that download. I went to playok.com, but couldn’t figure out how it worked, and there were no explanations. The Go boards I accessed were already in the endgame state. I was looking for a board to start a game, not end one.
It’s my understanding that, aside from Alpha (which is not available for any device any of us own), Go AIs pretty much suck. And it’s probably not good practice to play against an opponent that sucks.
First, Go is very different than Chess. My experience is good Chess players have a bit of a problem with Go.
#1. Learn the rules. Should take about 10 minutes, though Ko might take a bit to sink in. #2. Play 100 games, fast. Just put stones down and see what happens. #3. After 100 games, buy ‘Lessons in the Fundamentals of Go’ by Toshiro Kageyama. Read it and do the problems. Link. #4. Learn the difference between live and dead groups. #5. Play another 100 games.
IMHO, the biggest problems with understanding Go are the following: #1. Territory vs. fighting. You win by getting as much territory as possible but people like to fight. Fights are fun and we* are sorta trained to want to fight. Capturing stones feels great but is unlikely to win you a game. #2. Sente/gote. Sente means you are pushing the play, gote means you are responding but a bit more complex. #3. Aji. Basically latent potential. #4. Sacrifice. Sacrificing stones is common in Go and goes against our standard thinking. We don’t like letting things die. #5. Thickness. Basically the strength of a position.
Go is a deep game so this is just scratching the surface.
There are all sorts of Go programs out there. The problem is, most of them suck. You can play against them for a bit and do ok, however they are likely to teach you some bad moves, and worse, bad thinking. The best option is to play another player.
Board size. The standard board is 19x19. Start on a 9x9, then go to a 13x13 board. Why? 19x19 is so large that beginners can’t really understand what is going on. With Go, a move on the one side of the board can affect play in a totally different area.
Note, strength in Go is Kyu for beginners and dan for pros/experienced players. 30 Kyu is absolute beginner, 1 Kyu is intermediate amateur. 1 - 7 dan amateur is advanced amateur and 1 -9 pro dan are the badasses.
For programs, Crazy Stone and The Many Faces of Go are pretty good.
I have a program on my kindle which does daily problems. Nice little program though I cannot remember its name. Anyway, there are a bunch of daily problem programs which are really good. So is this website:
Well, the ‘sucky AI’s’ are only sucky compared to pros. Crazy Stone is ~ 4 dan. Also, the programs are very good at certain types of play, like races. The problem most programs have is that they become pretty predictable. Once you understand what style the program is likely to play then beating them becomes much easier. I have a program on my Kindle that took a bit to figure out but after ~5 or 6 games I realized it tries to kill everything. It can read our further than I can so I lose races. So I stopped getting in fights, built big strong frameworks and let the game win the small fights. And I beat it consistently. Doesn’t help me much as far as learning goes…
Google’s AI is different. AlphaGo, with maybe an exception towards his game fighting IIRC, didn’t really have a style that led to easy predictions. In fact it made some plays that had the experts scratching their heads until the moves turned out well later in the game. Some standard thinking about Go that has been around for hundreds of years is being rethought because of AlphaGo.
Couldn’t you get a larger effective handicap by playing on a smaller board, as suggested? You certainly wouldn’t want to play a brain-damaged opponent. Sure, you would probably win, but what have you learned?
About AlphaGo, once you have it trained, how big is the neural-network data and what kind of hardware would you need just to run it? Surely much less than that needed to train the network in the first place?
If I understand the question correctly, the answer is you can change the difficulty settings on the programs. Basically it trims the search space by lowering the search time for the computer.
One of the nice things about Go is that the handicapping is pretty damned good. You can handicap on any board size, though the relative value of each stone changes with the board size. This allows players with wildly differing strengths to play. However, the good players are a bit nuts. My Dad, around 3/4 dan (U.S. amateur) or so, once played a Chinese pro. The pro took on 10 other players, gave them all 9 stones and beat them all. And the pro won all of the games within 2 points of the other games. In other words, he played to beat all 10 players by the same number of stones and almost pulled it off.
If you’re up to it, you can learn a bit of Japanese, Korean, or Chinese and then avail yourself of material printed in one of those languages. You can also check online for the different national Go organizations for game records, news, tournaments in your area, programs, etc.
A word about noteation. The traditional notation for games is fairly simple: you look at a diagram of the entire board, the moves indicated by number (Black goes 1st in a non-handicap game, so a black circle with a white 1 indicates the first stone played; a white circle with a black 2 indicates the second stone played in the game, etc.; handicap stones are simply filled in circles with no number) and if a stone is played where a previously played stone was captured, you’ll see an annotation beneath the diagram showing that. It has the advantage of saving space, but it took me about a month to get used to it.
A couple of years ago, I took an online skill test from the China Weiqi Association, getting ranked as an Amateur 4 Kyu. I haven’t had the gumption to put that to test in real life, though. Maybe when I have more free time, I’ll get involved in the tournament scene.
Since the thread has been bumped, let me correct a mistaken comment I made.
Near the end of the game, when one player is passing and the other is placing stones inside territories (whether his or opponent’s), no points are won or last. A player wins a point each time his opponent passes, to cancel out the point lost as he plays inside territory.
Under old Japanese rules (as I learned decades ago), I think the passer gains a point with each pass. But those rules are obsolescent and Chinese rules now apply. Is this all correct?
The thing is, Japanese rules (I haven’t heard they were obsolescent?!) use “territory scoring” rather than the Chinese area scoring. Under territory scoring filling in your own territory reduces your score, so you wouldn’t do that.