Anyone know? I saw a thread talking about Champion Go (e.g. Crazy Stone). I searched the App Store and found one called Ginseilgo which has the advantage of being a Universal App, and is only $4.99. It also says that it is the strongest AI engine. Obviously when ported to iOS it depends a lot on how optimized the engine is to maximize computational efficiency, but anyway the point is that it's not clear to me which fo these two apps is stronger, or if perhaps there are stll other apps that are even stronger.

I tried playing Smart Go on 19x19 and even as just a 9k it was making moves that were embarrassingly bad. So I'm hoping to find something else maybe.

Champion Go + Server Engine Mode is the strongest I think but that's the app + a subscription I think. Actually that might be iPad only, I'm not sure.

Unless the app calls a server that does the actual computing, which I think is the solution used by Champion Go's "Server Engine Mode".

Having said that, the interest on having a portable go AI can be precisely for playing when no connection is available. If you're going to depend on a connection, you could connect to kgs/dgs.

Not answering your question directly: I don't think a smartphone has enough computational power/capacity to play a reasonable game on 19x19. I am only 4k, but I have Smartgo Pro 9x9 at level 18 (I have always white with 0.5 komi). I play quite fast though.

edited

I've seen Champion Go HD on the iPad give an EGF 3d a hard hard game (i.e. winning sometimes) on 9x9. Now on 19x19 it struggles against an EGF 2k. SmartGo is a lot weaker than Champion Go/Igowin.

Sorry I meant 19x19, as this was what was requested. And it was especially requested for iPhone, not iPad.

Fair enough, on 19x19 these engines aren't very strong. He mentioned Universal Apps though so I think he's also interested in iPad strength.

If we include "simply as communication devices" completely changes the question. The "+ Server Engine Mode" means the program has zero playing strength. It's not being used as a go playing engine. It's simply acting as a communication device. In theory you don't even know what program might be running on the bot you are playing against on some go server somewhere.

"Lamp", what are you saying? Although the makers of these programs (the MCTS programs) are attempting to widen their market by providing a wide range of playing strength this isn't really possible without introducing some of the weirdness you describe. At best they can calibrate for average effective strength overall, not strength of isolated moves. I believe 9k to be below where the MCTS algorithm can function properly and in this level of strength and below some sort of go AI would be a more satisfactory opponent.

Sorry, iPad is actually fine too. Universal would be preferable though, because I'm more likely to be carrying my iphone than my ipad.

Champion Go looks interesting, but I wish they had ticket options for a fixed number of games rather than a fixed time period.

Greetings. Although I usually play on IGS with Panda-Tetsuki, Clever Go Pro has the best AI that I have played against. The strength is adjustable at weak, medium, and strong levels. The moves of the AI are very textbook responses so you can also learn practical applications of joseki patterns and common tetsuji as you play.

Someone else suggested playing on KGS if you have an Internet connection available. Is it possible to do this from the iPhone now?

Take Care,
Desmond

You can play if you have a computer with java and access to remote pc access software.
See http://www.lifein19x19.com/forum/viewto ... =18&t=3648 for an idea. This specific way doesn't currently work, but something similar might.
I think it would work better on an iPad due to screen size, but it might be workable.

Please be aware that this is a completely unofficial solution. wms, the KGS developer doesn't support it (but has no problems with people using it).

_________________
a1h1 [1d]: You just need to curse the gods and defend.
Good Go = Shape.
Associação Portuguesa de Go

Greetings,

Thank you for the reply. It seems like a lot of extra work and money to get that up and running, especially when IGS works so well through the Panda-Tetsuki App. I wonder why the KGS developer doesn't want iPhone support?

Take Care,
Desmond

As William Shubert (the developer, his nick here is “wms”) has stated in another thread here, he is working on an HTML5 version of KGS which will then run in a browser window, no Java needed anymore, and this should therefore also run on any iOS device.

_________________
“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?

Greetings,

This is fantastic news! Thank you for passing on that important bit of information. I look forward to trying out the new interface.

Take Care,
Desmond

It may take quite a while, though, until that’s done … you may want to read viewtopic.php?f=24&t=6350and viewtopic.php?f=24&t=2719 on this topic.

Greetings, Tom in Germany

_________________
“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?

you guys have the question backwards -- why doesn't Apple support Java?

wms has no problems supporting iPhone, if Apple used Java.

Probably b/c it would mean a HUGE support necessity for all Java stuff that would make problems on iOS, not to mention that because it’s an interpreter language, people would then be able to load any Java code into their iDevice, including malicious software.

But this might be interesting: http://apple.slashdot.org/story/12/07/2 ... ava-to-ios


Greetings, Tom

_________________
“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?

Not supporting Java is one of Apple's wisest decisions. I often find software written in Java to be slow, and it needs a lot of updates with every new Java version, and has bad looking graphics and often has silly bugs.
Java was designed to keep programmers safe, to prevent them from making errors, but the result seems even worse. I guess for great software you just need great programmers.

There may be a problem here.

Shockingly bad moves? We need to remember that on a standard home computer the AI (go knowledge based) programs were only 6-7 kyu before the MCTS based algorithms came along. That represented a huge jump in performance BUT (very big but) there is little evidence that this algorithm can give good results below some relatively high playing strength. In my opinion, if below 1-2 kyu there will be gaping holes in what it does so it will sometimes make "shockingly bad moves".

The question being asked is essentially "what is the best program when using hardware inadequate for the MCTS to perform at the minimum level at which it doesn't make some "shockingly bad moves". I'm afraid the answer may be none. The hardware may simply be below what is required for the MCTS based programs to perform satisfactory.

Discussing whether these progams can also operate as interfaces to playing against a remote bot is another question entirely. One unrelated to how well they can play go on the specified (weak) hardware. It doesn't take much crunch to act simply as a communications interface, just the right software.

This is the only thing you said that is true.

There is great software written in Java too.