the ai dosnt attack in force unl;ess it

a) has no choice

b)it has you so massivly outgunned that you dont stand a chance, then when you bring in enough ships to even the odds it runs.

Seriously, one time the ai was just sending ships and retreating, for half an hour until i (i was their ally) took the planet.

it means that you can always be sure that they usually send their entire fleet, you can ravage their other planets at will and they dont stand a chance.

The AI in a LAN game in Dawn of War: Dark Crusades, rushes you within five to minutes as will, while also taking into consideration stategic points, critical points, and technology upgrades.

If they were to address the issue with the AI in Sins, then I would have to say, they'll need to fix, everything.

1) the AI don't attack unless they outnumber you.

1a) they only attack with maximum numbers, and ships such as heavy cruisers and caps, when both are destroyed, then they fully retreat.

2) they don't take into consideration, that holding a star, and not the planets could easily boost their defense of their worlds while they upgrade all areas within technology.

3) mostly they take worlds, defend with minimal numbers, then move their fleet around to bogus positions on the game, which are easy to pick up on, then counter follow their movements.

4) Stardock is too specific on what types of AI are to be in game, overall they should have it to where the Ai are both aggressive, technological, defensive, and economic.

5) It'd be more challenging if they AI didn't focus on allying with you, and focused more on getting planets fast, moving their fleet after taking a world or defending it until it's at max upgrades, and building support is finished.

5a) to make it more interesting, I'd like to see the AI focus on upgrades, then once they've built up to their max in technology focus all res, on you and only you.

6) it'd be interesting in game to see all AI ally with each other, and focus all ships, and such on your only world(s).

Stuff like that I guess, there's more to it, but as always it's a insufferable task to just think on them all.

Another thing that might help is if the AI examined the structure of the graph/network of planets/stars and identified chock points to mass fleets in order to guard against attack.

What are you talking about, unfair is still too simple, I can beat nine unfair in maybe six hours depending on the map size, which usually the case is when I choose one of my mega maps I've made, which it would take weeks to even play the game.

I think part of the problem is that the developers want to make the AI play more like a human player, but sometimes that's not necessary.  I was one of those who pushed for the "unfair" option for that reason.

I think the devs should let the AI use strategies that a human player cannot use.  For example, raid/attack from multiple fronts at once.  A human player will have serious problems dealing with that.

Also, random attacks all over the map keeps players off balance.  It doesn't have to make strict strategic sense.  The AI can still do more strategic major attacks, but being unpredictable by using randomness is something the AI can do better than a player (and is easier to code).

It's kind of like throwing a wild pitch in baseball.  It won't get you a strike, but it sure keeps the batter honest.

Human players can't raid/attack from multiple fronts at once?  News to me - as it stands, attacking on multiple fronts   is the best way to defeat an AI faction, as they seem to have serious trouble building secondary fleets.

Too much cheating on the AI's part means that the player can no longer formulate offensive/defensive strategies since they can no longer accurately predict what effect their efforts are having on the enemy.  We may still be able to defeat them because...they're stupid, but some of the fun is lost if your deep strike on their core production worlds has little effect on their war machine, or a system you know they can't reach, and thus have left undefended, is suddenly attacked from nowhere...