What's the best Android app to play as an opponent?

_________________
Assertion without proof takes up little time; misrepresentation is always beautifully brief. –John Stuart Mill

Champion Go
https://play.google.com/store/apps/deta ... mlnb2VuIl0.

But really, I suggest
DGS
https://play.google.com/store/apps/deta ... 5hbkRHUyJd

and OGS
https://play.google.com/store/apps/deta ... NsaWVudCJd

and play some people.

I have 6 games going on DGS now and can play those on my Fire tablet but I'm looking for a good app for when I lack Internet access. Right now I have AI Factory's Go which is adequate but I'm looking for something with more features.

_________________
Assertion without proof takes up little time; misrepresentation is always beautifully brief. –John Stuart Mill

WegoIgo is quite nice, and so is Gobandroid - you can play against gnugo on either of them.

_________________
Patience, grasshopper.

I like Champion Go, especially for quick 9x9 games when I'm killing time somewhere.

I suggest spending your time doing problems on gogrinder instead of using one of the android computer opponents. You will get much more out of it in terms of benefiting your play. The android AI options are all pretty bad.

I will look into that. I need to see if I can figure out how to sideload Gogrinder into my Kindle.

_________________
Assertion without proof takes up little time; misrepresentation is always beautifully brief. –John Stuart Mill

You'll probably need to root your fire or sideload google market to get it to work. These things are well worth doing anyway (Im running a rooted nook tablet myself)

I'll look into that as well, although I am somewhat leery of rooting my Kindle and hence voiding the warranty. I just wish it was easier to find and download apk files via PC.

Update: I have managed to install Gobandroid and gnugo on my Kindle Fire and it works perfectly (thus far).

_________________
Assertion without proof takes up little time; misrepresentation is always beautifully brief. –John Stuart Mill

However questions about the usefulness of computer opponents running on this or that hardware should be answered in the context of the strength of the person asking the question.

So yes, pretty weak (what's available on such minimal hardware) but we should give an indication of what is meant by that in more absolute terms. What's too weak to be an opponent for a 4 kyu can be plenty powerful for a 20 kyu to learn from. Some bad habits, yes, but those are again relative to strength. The bad habits of a 10 kyu bot might be one heck of a lot less of an issue than the bad habits of an 18 kyu human opponent (if we are talking about a 20 kyu playing against either to learn from).

Seems to me there is merit to both studying problems and playing opponents (human or otherwise) to improve at go. Besides, I want to play the game and sometimes getting online is simply not feasible for me since I lack 24/7 net access. At least Gobandroid HD lets me play against Gnugo and even has commented pro games to study.

_________________
Assertion without proof takes up little time; misrepresentation is always beautifully brief. –John Stuart Mill

but playing a computer opponent is like doing problems

Not for me.

I think we have perhaps reached the stage where it might be informative to perform a "Turing test" on this question. Somebody prepares a set of game records (say 20) half of which are human vs computer and half of which are human vs human. That means that of the 40 players (two in each game) 10 were a computer. Then post the game records and have people place 10 votes. See how much better than 25% right they can do.

If the results of this are not a whole lot better than 25% people who consider the computer a poor substitute might want to rethink their position. For example, a 50% result would mean the computer is as likely to be mistaken for a human as not.

Such a computer-go Turing test was done in Japan already:
http://www.jaist.ac.jp/jaistcup2011/index-e.htm
http://www.mail-archive.com/computer-go ... 02740.html

Rémi

Interesting. There are some obvious questions, though: how many games were actually used? And of course 19x19 would be a better target for such a turing test.

_________________
Occupy Babel!

Wow, complete failure to read. SCWilson is asking about playing an AI on his android. Tofu is talking about the incompetence of various computer programs on the android. The sudden tangent into the normal defense of the honor of computer players is uncalled for. A go Turing test would be a great topic for a completely different thread.

How well these programs play depends on the strength of the machine.

How useful a person might find a program playing on a weak machine is going to depend very much on the strength of that person.

I would expect SCWilson and Tofu to have very different experiences trying to play against a program running on an Android.