Just Askin', I don't know (Winning/Losing)

[jpwrunyan]

So far the game difficulty seems confined to the early game with the exception of the shakturi (whom I rarely encounter <sadface>).
Part of this is due to problems with game balance talked about exhaustively before, part of it is the snowball effect.
Snowballing can be dull, but on the other extreme are capricious events and ai cheats that effectively make pointless any skilled play on your part. Even if you played against another human, if you played better than them you could still snowball. My point is snowballing sucks, but is a sign of a weak opponent more than anything. Luck can also cause snowballing of course but I believe its pretty mitigated in this game.
So to address snowballing is to address the ai.

I like the idea of playing a game with no races of your family so all other ai empires ally with each other and against you.

[BigWolfChris]

Like all games, once you learn how to play to a decent level, AI opponents are simply speedbumps to your ensured victory
You just alter the length of time to victory and how steep those speedbumps are through handicaps to yourself (or advantages to AI)

I've yet to play a game that is otherwise - and I don't include games where the AI blatantly cheat, as that is not competitive AI, just lazy programming

[jpwrunyan]

In defense of programmers everywhere, giving the ai unfair advantages at higher difficulty levels can still yield better results than more advanced ai logic. As a player also I feel some satisfaction beating a cheating ai. Remember that moo2 gave its ai more racial picks at higher difficulty. This had a pronounced effect on game play and made things more challenging at all stages of the game. It also felt good winning against an opponent as an underdog having only your human intellect as an advantage.

I think ai cheats, when done right, can be very effective and satisfying challenges. And they have the benefit of being easier to implement than deep blue.

[JonathanStrange]

"Snowball effect" or as I usually call it "Steamroller" effect:

I think that's a good point to bring up. Once your (hopefully it's your) empire, space navy, military, economy or whatever gets ramped up in many games, victory becomes a bloody drill of knockin' out the enemy one by one.

Or being knocked out.

I've no problem with the AI occasionally ganging up on a winning empire. Does the AI do that?

It feels real that the alien empires would notice that the herd was being thinned out and they'd do more than just mention at the local space cantina, "You don't see too many Heechee anymore, do you?" They'd, one would hope, would ally against someone's steamroller. Wouldn't they?

[MartialDoctor]

ORIGINAL: Haree78
I have been wiped out but almost always from near the start of the game. When I stack the odds in my opponents favour or the Galaxy heavily stacks it against me, like 3 insects on my borders. I have also lost from the Shakturi a number of times. Generally though this game suffers a little from steam roll after you start to win syndrome that all games Vs the AI suffer.

Basically, I'm in a similar boat with Haree... I have only been challenged very early in the game; only losing if the AI either is far too powerful for me to handle, since I always start them much larger than myself, or the AI sneaks in and takes my homeworld. If I manage to get to mid-game, it's basically all finished and lacks any challenge. Although the Shakturi can add some challenge mid-game.

Hopefully, we can see some AI difficult level additions / modifications and a change in AI behavior in the next expansion. If they'd actually start spending money on ships, among other things, the AI would be far superior to what it is now. If those changes happen, this game will change from a really cool game that is far too easy to a game that is truly excellent in all respects. We then won't be having these types of posts anymore [:D]

[BigWolfChris]

ORIGINAL: jpwrunyan

In defense of programmers everywhere, giving the ai unfair advantages at higher difficulty levels can still yield better results than more advanced ai logic. As a player also I feel some satisfaction beating a cheating ai. Remember that moo2 gave its ai more racial picks at higher difficulty. This had a pronounced effect on game play and made things more challenging at all stages of the game. It also felt good winning against an opponent as an underdog having only your human intellect as an advantage.

I think ai cheats, when done right, can be very effective and satisfying challenges. And they have the benefit of being easier to implement than deep blue.

I think you may misunderstand what I mean by AI cheating
Giving the AI something like more racial points or bonus to production I would call handicapping the player, I don't object to this to much, since game AI will probably never match human intelligence (I say never because games are more and more ignoring AI forgetting not everyone has time or desire to play against humans), and normally the player chooses the AI to have these advantages, either through the difficulty level
What I mean is, AI being able to ignore rules/physics of the game, like having no fog of war, not even needing to explore, some versions of both Civilizations and Gal Civ have been proven guilty of this at one time
Another being knowing how to counter every single one of another players design without ever setting eyes on them, either in combat or through espionage, or which planets are weakest with no recon - this one I think DW may actually do itself
And there are other methods used as well (Some FPS AI well known for sniper accuracy using standard firearms at ranges no human player could ever recreate for example)
Alot of it is lazy programming, quite often given with the excuse to save CPU cycles... and these examples and some I've not said are blatant AI cheating


TL:DR - AI having an advantage is fine for a challenge (but at a player choice for it) is fine
AI being able to ignore the rules/physics of the game that the player cannot is cheating, and is to be avoided at all costs

[BigWolfChris]

Also, programmers need to stop doing this, AI is there to occupy the players time, type of design
AI should being having their own goals and be doing what it can to reach them, regardless of what the player is doing
If the player isn't seen as a threat to it's goals by that particular AI, then the player is not factored

These goals would be picked from the same options as the players (domination, culture, etc)
Some games have pulled off this type of AI to a small extent (Civ V claimed to... but last I checked the AI was poor throughout in that game), but really this is the ultimate type of AI for a game, because that's how most humans play, we ignore other players/AI unless we think they are a competitor/threat to our own goals

[jpwrunyan]

@bigwulf
I dont think we really disagree. I wanted to draw a distinction between ai cheating and player handicap. I think you concluded my sentiment well. And I agree that there are satisfactory ways to compensate ai that feel like handicaps and unsatisfactory ways to compensate that feel like ai cheating. Wasnt it unreal tournament that had excellent ai bots while its competitor quake 3 (4?) had ai that one person described as: "easy mode ai acts like a spastic monkey, hard ai acts like a spastic monkey that never misses".

Ah unreal tournament... The memories.

[Grotius]

I also draw a distinction between cheating and handicapping. I don't mind handicaps in a game vs the AI; in fact, for most computer strategy games, such handicaps are essential to provide a challenge.

Cheating by the AI is different: that's when the AI gets completely different rules to play by. That's less satisfying.