Spaceward Ho! (4x Strategy by Ariton Games/Delta Tao) [Now Universal!]

[christophercotton]

Originally Posted by DaviddesJ

On the other hand, I haven't played the game

Well what are you waiting for! Definitely go play it and let me know. Be sure to watch the Basic Gameplay video. It shows you the basics of the interface.

[athros]

Originally Posted by DaviddesJ

Hmm. I don't think credentials are very useful here, but I have a PhD in math, I studied computer science at MIT, I was one of the first software engineers at Google, my boss was the author of the world's leading AI textbook. I think I have a fair idea about what's possible.

I was more curious about your background than anything. You made a pretty strong statement, and I wondered what was backing that up. I hope you didn't take offense, because I didn't mean to give any.

Originally Posted by DaviddesJ

On the other hand, I haven't played the game, I shouldn't really be expressing an opinion about this game without playing it. I know that computer opponents can't play games like Civilization or MOO anywhere near as well as humans. But maybe this game is simpler/different and the AI can do better, I shouldn't venture an opinion without trying it first.

Anyway, thanks for the responses. Giving the AI information advantages seems like potentially a reasonable compromise (predicting what the opponent is going to do is one of the most difficult problems in this sort of game).

I have no doubt that the AI has some level of cheats built in at high levels, even in Spaceward Ho!. There was an article on AI in 4x games a while back..I'll have to run it down again.

[Mythbuster]

@David: You have a huge advantage, the language. English is just my 5. language so I get fast in trouble to write complex things.

But: Monte Carlo is a search logic in Go (as far as I know, SmartGo for iOS uses it ... I am a Betatester for several Go Programs, also for SmartGo), not in Backgammon. As I wrote, JellyFish uses a complete neral Network ... but nevertheless. I know, that games like Civ cannot beat a human player. I love intelligent board games due the fact that my real life IQ is 146 ... that is why I like chess ... oh, in my 'younger years' I was in the world champion chip team of chess programming and it was a great time ... I was also at the human world chess champion TM in the team of G. K. ... that is a long, long time ago ... 1990 ....
Now I am just a writer and author ...

Most games do bore me because they are too easy to beat or you see that they are cheating. I prefer to play vs. a 'real AI' 1:1 ...

As I wrote before, I found that the AI in SpaceHo uses different strategies to win. It is interesting that they can even change their strategy during a game when they realize that you are ahead.

Due the fact that the game parameter very limited I really think it is possible to play strong for an AI without cheating.

I think, you should give it a try ... it is not that expensive ... and give a report .... and if you are that good I am waiting for the MP update to meet you online ...

[Mythbuster]

Originally Posted by athros

I have no doubt that the AI has some level of cheats built in at high levels, even in Spaceward Ho!. There was an article on AI in 4x games a while back..I'll have to run it down again.

Well, most 4x games are more complex and have more variable parameter. Spaceward Ho itself is very simple ... but do not understand me wrong: Simple does in this case not mean boring or easy to master ...

[DaviddesJ]

Originally Posted by Mythbuster

But: Monte Carlo is a search logic in Go, not in Backgammon. As I wrote, JellyFish uses a complete neral Network ... but nevertheless.

Well, that's sort of true, but backgammon computer programs are only strong because they can use Monte Carlo simulations (playing lots and lots of games) to train the neural nets. Without extensive training, the neural network can't play well at all. This is one of the things that's hard to do in simulation type games---if you just "play out" the position, with an algorithm that doesn't understand the strategy, then the result is meaningless. It's also hard to distill the right inputs to the neural net---in a game like backgammon the state of the game is so simple that it's fairly obvious what they are.

That said, I still should actually try the game. Soon. I did buy it already.

[athros]

Originally Posted by Mythbuster

Well, most 4x games are more complex and have more variable parameter. Spaceward Ho itself is very simple ... but do not understand me wrong: Simple does in this case not mean boring or easy to master ...

Oh, believe me, I know. I'm still struggling with it in some ways, but that's more due to UI issues than anything, and my hang ups with other 4X games (like MoO 2, Starbase Orion and the newer ones.)

[Mythbuster]

That is indeed correct ... JF uses MC to train it's brain.

I am sure, you will not regret the purchase if you like games like that ...

[DaviddesJ]

For $4.99, I won't regret the purchase even if I never get around to playing it.

[christophercotton]

Originally Posted by athros

Oh, believe me, I know. I'm still struggling with it in some ways, but that's more due to UI issues than anything, and my hang ups with other 4X games (like MoO 2, Starbase Orion and the newer ones.)

If you have UI issues, please let us know about them. The 1.0.2 version we just released will update some of the issues people have had, and we are working to redoing some of the design to be better. The big issue was that we have had people who have played the Mac version quite well and now are getting more feedback on how it is for a new player.

[Mythbuster]

A very good review from 148Apps: http://www.148apps.com/reviews/space...o-ipad-review/