Some Tichu Answers

I've seen quite a few questions about Tichu (for the iPhone) lately that I thought would be best answered in a public forum. I posted most of this message elsewhere but I thought I'd also share it here where hopefully more people will see it.

First, performance:

The target platform was the 3G S. It also will run reasonably well on the later generation iPod touch. Older models will, of course, work just fine but be a little slower. There's a lot more going on here than there was in Mü and that takes a LOT more cycles. (You should expect that on a first generation iPod touch this will be VERY slow.)

Network play:

As with Mü, local network over Bluetooth is supported (and it's much more stable than Mü's network play is). When Apple releases their matchmaking service here in a few months I plan to look at supporting it but I can't promise that it will happen.

Finally, about the AI:

With Tichu having so many more plays possible on a given turn, the decision tree is HUGE and as a direct result, the AI is not as strong as Mü's. Also problematic is the fact that I've discovered that people feel VERY strongly about their style of Tichu game and styles vary WIDELY between players. Some folks think that certain play decisions are set in stone. Others feel very differently. The fact that the AI is usually playing as a human's partner doesn't help. :) You'll enjoy my game best if you treat it casually and just try to enjoy the game. Don't think this is going to replace playing a game with four expert players who are familiar with your style of play. I wish it could but it probably won't.

Here's a little bit about the three difficulty levels that matter (the easier ones are not as interesting).

The "Normal" setting plays a reasonably competent game. The AI will ocassionally make a poor decision but it's good enough to enjoy (at least for me). This is the level that will probably feel the most "human-like" to most people. The AI tends to make slightly safer plays at this setting.

The "Better" setting plays slightly stronger. It thinks farther ahead which will cause perf problems for folks with older model devices, but may also cause it to make moves that you might think are a little risky sometimes. The thing to remember is that the AI has a perfect memory of all cards played so if it does something you think is odd, it's probably because it thinks that the odds are high that it will pay off. If you trust your partner a bit then usually it will work out.

The "Best" setting is the one that I fear might throw people off the most. At this level the game is allowed to peek into other players' hands when deciding what to play. (This only applies when the AI is deciding what to play on a trick; never when deciding to call Tichu or deciding what cards to pass.) Because of this, you really need to trust what your partner is doing if it's playing at this setting. For instance, if you gave it the Dog and it leads something else, it's probably because it thinks it's going to be able to set itself up for the double victory by taking a few tricks and then dogging to you later. You may think that's a horrible play but if you could see all the cards then perhaps you might not.

I am continuing to improve the AI so you can expect to see a few updates in the future that (hopefully) improve it some. If you would like to provide me with feedback or (gentle please) criticism then please feel free to send me email. I'd love to hear from you. (I would prefer that bugs or criticism be reported via email rather than in a public forum.)

Hopefully that helped answer some of your questions!

I've had a blast with Tichu so far. It's been wildly successful in the few days since its release and I hope to keep on improving it. This has easily been one of the most challenging pieces of code I've ever worked on. Hopefully you'll enjoy it too.

Steve

email: steve-at-housefullofgames-dot-com