Sid Meier's Starships

[Wiglaf]

YOU WOULDN'T KNOW ANYTHING ABOUT EITHER, AND NO "CRISPY APPLES" AREN'T REAL APPLES NOR DOES DRINKING DIET COKE MAKE THEM OKAY TO EAT

[Urban Ranger]

Also, for an interesting tactical element of a game they need a good AI.



They promise a big variety of ship designs, and to make tactical combat interesting they need to teach AI how to use all these different kinds of ships with different features like speed or stealth. I think they should have made combat semi-arcade, like in Star Control 2 or Space Rangers (in hyperspace). It's not as fast-paced as in fighting games, and different abilities of different ships make it tactical rather than just a pure fighting game mode.

Tactical arcade doesn't need a good AI to be fun - in Star Control 2 it was fine on 80286 20 Mhz in real time, with a very taxing (for it's time) sample-based music that was played on PC speaker, so it had a really primitive AI. Faster Than Light doesn't really have an AI (other than the simplest random target selection and patchfinding for invaders), but it's tactical and real-time so its fun as well.

However, most turn-based tactical games with complex mechanics suffer from a bad AI, ranging from old Archon Ultra to a recent Blood Bowl (which has rather good AI for it's genre, but it's still far beyond human capabilities). I think it's also the biggest weakness of tactical JRPGs - they have an interesting game mechanics but AI in tactical JRPGs is just abysmal. I really hope that Firaxis doesn't make the same mistake...

Unfortunately, AI is notoriously hard to do.

It takes years using sophisticated machinery to finally beat humans in chess, and that's a game with very rigid rules.

That's why they always resort to cheating.

[onodera]

Unfortunately, AI is notoriously hard to do.

It takes years using sophisticated machinery to finally beat humans in chess, and that's a game with very rigid rules.

That's why they always resort to cheating.

I can't beat my mp3 player in chess, so you don't really need a very sophisticated AI for your game, most players aren't grandmasters.

[Sava]

Unfortunately, AI is notoriously hard to do.

It takes years using sophisticated machinery to finally beat humans in chess, and that's a game with very rigid rules.

That's why they always resort to cheating.

WHOA


UR sighting

[Proteus_MST]

I can't beat my mp3 player in chess, so you don't really need a very sophisticated AI for your game, most players aren't grandmasters.

But most games don´t have such a limited set of rigid rules as chess (with just 16 pieces for you to move each belonging to 1 of 6 types)

To give a comparison of another game: Go

It has a few easy rules, but you can place your stones anywhere on the board, thereby giving each player an enormous spectrum of placements during their move ... also there is no intrinsic value for each piece, but rather, their value is determind by their placement on the board (and their vicinity to other pieces of your or he enemy colour).
Therefore any Go-AI has to analyze a space with a huge number of possible placements each turn (-> huge branching factor)
The result is that, while we nowadays have programs that can reliably can beat chess masters, most Go-Programs only can defeat advanced pupil levels ... and those few programs that managed to beat someone of the Dan (i.e. master) level, aren´t capable of doing so consistently.

Games like Civ (due to the huge number of possible actions per turn) can surely rather be compared to Go, than to games like Chess


As for OP:
GMG offered me a 25% discount on Starships, which I used ... so now I´m also among the crowd who may are able to play it when it comes out

[FlameFlash]

I am guilty of having pre-ordered Starships. I hadn't been aware, sadly, of the GMG discount.

[Sava]

I have no interest in Starships. They lost me with CivBE

[FlameFlash]

But space... it's like, the final frontier and all that...

[Koyaanisqatsi]

Go get Offworld Trading Company. Seriously. It's a great game and it's not even done yet.

[Aeson]

I think I have 2 more gift edition codes ($20?) for Offworld Trading Company if anyone is interested.

[Guynemer]

How close to completion is OTC?

[Koyaanisqatsi]

They say it'll come out in 2016. They worked on MP first, there are a bunch of people playing that, but I haven't tried it yet. I've been playing the SP campaign mode quite a bit. They keep saying that they haven't put much work into it yet, but for me it's a fun game as is. It could use more companies (civs) and the AI is a bit dull until you turn up the difficulty. I haven't noticed a lot of AI cheating though.

The balance right now is fun. Individual games are relatively short, maybe a half hour max. The campaign mode is several individual games strung together with progressive customization of your company; you earn certain bonuses by winning each game, and they stack. You're given a choice of maps with different playing conditions and different rewards in each round so there's some strategy to choosing your battleground. The way the game is balanced, the market starts slow and heats up quickly so that by the end of each game you're trying to keep things from flying apart at the seams as there are massive market swings, other players sabotaging you, etc. I hope they don't change that by the time it's finished, the simplicity and pacing of how it's set up now is a virtue.

[Guynemer]

I've heard lots of good things, I think I'll give it a try.

[Hauldren Collider]

Star Citizen looks better. It will be the next chapter in the Wing Commander series.

If star citizen ever comes out mark me down as surprised.

[Hauldren Collider]

But most games don´t have such a limited set of rigid rules as chess (with just 16 pieces for you to move each belonging to 1 of 6 types)

To give a comparison of another game: Go

It has a few easy rules, but you can place your stones anywhere on the board, thereby giving each player an enormous spectrum of placements during their move ... also there is no intrinsic value for each piece, but rather, their value is determind by their placement on the board (and their vicinity to other pieces of your or he enemy colour).
Therefore any Go-AI has to analyze a space with a huge number of possible placements each turn (-> huge branching factor)
The result is that, while we nowadays have programs that can reliably can beat chess masters, most Go-Programs only can defeat advanced pupil levels ... and those few programs that managed to beat someone of the Dan (i.e. master) level, aren´t capable of doing so consistently.

Games like Civ (due to the huge number of possible actions per turn) can surely rather be compared to Go, than to games like Chess


As for OP:
GMG offered me a 25% discount on Starships, which I used ... so now I´m also among the crowd who may are able to play it when it comes out

Not really. Go and Civ can't really be compared, for a few reasons. The space of Civ is really huge, but with a few basic heuristics you can eliminate really huge sectors of that space. That's not easy to do with Go. The other thing is that they can make the rules of Civ in a way that inform AI design and make AI easier to do.

AI is still tough, but when you're the one making both the game and the AI, it's definitely possible to make one that can beat your average player. Not easy, but possible. And sure it'll be vulnerable to someone who comes up with some game-breaking strategy that your AI doesn't account for, but you can patch that.