I searched for information on this but the forum seems to have a bunch of posts about AlphaGo drowning out any useful posts.

I'm trying to increase a few stones to be able to play even games against some of the local players. I'm studying books, working problems, and playing games in person.

I want to get a few more games in and I have to travel to play in-person so that leaves playing online. I get really bad online go anxiety and I've not been able to kick it. I don't have that same anxiety playing bots on IGS or playing against GNU Go. A couple of people have told me that the bots will just cause me to start adopting bad habits and I'm not sure I understand why.

I'm sitting at 21kyu right now. Are bots and GNU Go useful opponents at such a low level or should I steer clear of them?

Thanks!

while its true that bots tend to have general weaknesses that can be exploited by stronger players, and those exploitations can be considered bad habits that you should not learn to imitate, for lower level players, as long as you try to play honest moves, you should be able to learn a lot and improve from them.

First let me say: DO NOT PLAY BOTS if you want to get stronger! The reason is that bots tend to play a lot of bad/useless moves. For a beginner it makes it very confusing which moves the bot makes are genuinely good and which are just flaws with its programming. This among other things will cause you to learn bad habits. The bottom line is: don't not play bots as a beginner, you will come to regret it soon enough if you are hoping to get from kyu to dan level at some point.



The real problem is that weak players can also exploit the weakness of a bot. A bot will make the same mistakes over and over, and a beginner player will think that what they are playing is 'correct' because it work, even though in reality it's just due to a programming flaw.

This is the line of reasoning that I was presented with by a couple of people and I guess I'm not at a rank where I can understand it. I see little difference in the moves the bots play and moves stronger opponents play unless they are substantially stronger than me.

At some points of the game the bots will play normal moves and at other they will make you stop and think: "what the hell". Bots make a lot of moves like that. The stronger you get the more you will be able to see these odd/bad moves. Trust the people you talked to and don't play bots for now. Soon you will realize just how utterly random and inappropriate bot's moves can be.

Play the strongest opponents you can find, as often as you can. If the strongest opponent you can find is a bot, play it.

Gnu Go is a generation behind, but both fuego and pachi are free, open-source programs that could easily beat you, even if you took a 9 stone handicap. Any "bad habits" you pick up from a 2d program will be much better than the habits you would learn playing other 20 kyu humans.

This may have been true in the past, but it isn't true anymore.

As your are 21 kyu, may I suggest playing against bots that can give you 9 stones? It's fun and you will learn more than playing even against them without picking up bad habits. Besides, a lot of human players who can give you 9 stones might not be willing to do so, as they are more used to taking stones than giving them.

Good luck!

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

But a lot has been said that needs paying attention to:

"Even weak players can learn to exploit a bot's mistakes. The bot always makes the same mistake."
Perhaps SOME behave that way. In case, use a bot that doesn't. Or with a bot that gives you a choice whether to "randomize" its moves, use that setting.

"learn bad habits from them"
You might if you try to play them in even games (make the bot's strength setting so low that you can do that). But if you have the bot's strength level such that you have to take 3-9 stones any mistakes it is making will be ones way above your current level. In other words, it will be able to punish your mistakes so that you learn not to make them, and as you improve, you keep upping the strength level of the bot.

Learn different things. When you are playing high handicap games, you are learning about how to keep your stones connected and to interfere with the connection of the opponent. When you play low handicap games, you learn about direction of play and how stones already on the board affect what is joseki and what is not (what is an equal division in isolation might not be equal when other stones are present -- a variation normally not as good might become excellent and one normally good could become bad.

Understand the limitations of the types of programs (with regard to YOU). For example, the algorithm behind the MCTS programs isn't going to work well if "dumbed down" to your level. But that's OK, because the pre MCTS programs had gotten up to around 6 kyu, you couldn't handle that even with 9 stones.

Don't start by paying for software, you don't need that yet. For example, the "free trial version" of MFOG 12 offers the lowest level (18 kyu) and the easy problem set. I suspect even the free trial will let you select "randomize". It should be enough to get you started and give you an idea if you like playing against a bot and what features you use. Time enough to make decisions about what to do next when you have some experience.

From personal experience, i can say that at 21k you will rapidly go to perhaps somewhere about 11k or 10k if you play bots. Even if you only use GnuGo. After that, it will become more difficult, and bots are less helpful for improvement (unless you use dan strength bots).
I don't see a point in things like -bad- habits at my level (somewhere around 9k or so), because I think I have no 'habit' or general playing style. See emerald's post.

I'd suggest not giving up in overcoming online go anxiety. Maybe play 9x9 or 13x13 games were less ego is attached, but eventually you have to tackle the core problem there. There is some advice in Dwyrin's video here.

Maybe some of the players at your club will play handicap games with you online when the club is not open.

I am not in favor of spending the majority of one's training against bots, especially any one bot, but I wouldn't say it's because of bad habits. It's more that improvement requires learning to punish a large variety of opponents' mistakes. For this purpose, it helps to have a variety of opponents.

"unless you use dan strength bots"

We ave already concluded that in order to be learning from bots the bots should be substantially stronger than you are. For example, nine stones stronger and you try to work down to below three. BY DEFINITION that would require a dan strength bot once you have gotten past 11k or 10k.

Luckily dan strength bots are readily available (as long as we are talking about running them on a computer and not a phone or weak tablet). And these would be MCTS based programs, so would not always play the same way.

aw..phooy...
some of you guys put too much emphasis on a bot's weaknesses, the hardware its running on, or that it must be so far above you to be of any use.

AlphaGo is not that much more above pros, yet they are all eager to see what they can learn from it to improve their own game already.

to the OP... find a bot... have fun. Learn what you can from whomever you can.

When I learned Go for the first time, I was stuck at around 18k. I was afraid to play people, because it was intimidating. I didn't like showing my weakness.

So I started playing GnuGo, which was about 6k at that time, I think. I started by playing black with 9 stones, and the bot beat me. I kept playing it, because I wasn't afraid of losing to a machine - didn't seem that the machine would judge me based on my mistakes.

Eventually, I could beat the bot evenly, and I felt more comfortable playing in person. When I went to the Go club at the University of Michigan for the first time, I was about 5k. At that point, I was more confident in playing against humans, and kind of stopped playing bots.

I don't think there's anything wrong with playing bots - though, they do play a bit differently than humans sometimes. More than that, the bad part about playing a bot over and over again like I did is that you get used to how it plays. It's kind of like playing the same human opponent over and over again.

It's nice to spice things up a bit, play against a variety of opponents, and get into a variety of situations.

The good thing about bots, though, is like I said - they don't judge you... At least until engineers incorporate the ability to criticize into their bots. :-p

_________________
be immersed

It isn't just bots. I recommend that beginners play with humans who give them at least 5 stones.

When I was a beginner, my main opponents were a 5 kyu and a 2 dan. I did not pick up many bad habits from them.

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

I expect that a few lines of code could remedy that.

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

The result of over generalizing? Not looking carefully at what is being said?

Look, it depends VERY MUCH on the algorithms the program is using and that is closely related to strength levels and strength levels are in fact what strength levels can be achieved related to hardware power.

a) For strength levels up to about 6 kyu the programs almost certainly "go knowledge" based AI's. These don't need a great deal of computer power to be able to play with reasonable time controls. But they tend to be systematic and not aggressive. Many of them will make the same move in the same situation all the time.
It is THESE you can learn bad habits from unless set at a level well above your own.

b) For strength levels above about 2-3 kyu up to perhaps 3 dan (with a strong home computer) and up perhaps 6-7 dan (with a machine well above that in power) the MCTS based programs can be used. But they cannot be easily* weakened below that and this is more the underlying algorithm than the implementation (it would get erratic). But these do not have SYSTEMATIC weakness, you won't learn bad habits, won't ALWAYS play the same, so you can use them set to your own level.
When people say "hardware matters" it is THESE that they are talking about. Needs to have enough crunch to do enough playouts within the time control.

c) AlphaGo --- here we are talking about an approach and a program that requires a machine at least an order of magnitude more powerful yet, say two orders of magnitude more than a powerful home machine. As an neural net trained on moves made by high ranking professionals, of course its analysis would be respected by high ranking professionals.
There is no point in discussing this one as a bot you could use.

* In theory could have a MCTS program that could play weaker by combining "a" with "b". Picture a sort of tag team play, randomly selecting whether the "a" player or the "b" player made the next move, strength level set by adjusting that probability. Sort of like a weaker player who sometimes makes brilliant moves. AFAIK, nobody has implemented something like that.

IMO (as a 9k player) this (above) advice by emeraldemon is good solid advice.
In addition to the above, IgoWin software (which is the older freeware version of "Many Faces of Go" ) is very good for beginners. It only plays 9x9 games but it is quite good at teaching beginners, say 25kyu - 15kyu, some basics. Some of the things I like about it are: 1) It is always ready to play a game, 2) It is very patient, if you make the same mistake over and over it will never complain or get bored, it just plays about a 5kyu player response (or better*) which is usually good enough to beat beginners players, 3) at 9x9 IgoWin plays quite well... IIRC the program you can load on your machine (for free) will actually learn (or more precisely will stop playing some bad moves) after you have beaten it a few times.

I agree that you should play at least half of your games with humans for best learning results, but as was already mentioned humans also make mistakes, so you can pick up bad habits (weak plays) from them too, so play the strongest players you can find, and if that is a bot, GO for it!!!