Tremor, freeze, and tornado are a big part of why this average skill player can survive against Insane AIs.  They allow a few strong armies to slow/reduce the many strong AI armies that can swarm us (providing the AI armies don't just sit there).  I want the AI to do to me what I can do to them -- good for the goose, good for the gooser!

My apologies Illauna, I should have understood what/why you were saying.

Lesson learned! 

I have been playing a few games and some things pop up in regards to the AI. The AI is indeed very sophisticated but rarely puts up a challenge, because it does some very dumb things. Just played a Large map and ended up facing Roseln, who did terrible. They built stupid units like widows and all their champions had tons of injuries. They also failed to kill the Darkling lairs all around them. After examining the situation I think I know the root of their problem. The AI is just horrible inefficient when it only has a few cities. Later if they get a ton of cities they can compete but they are horrible at managing a small empire of 2-4 cities. 

First of all the AI still doesn't get that it should be prioritizing workshops and bell towers. Instead it always wants to build pointless monuments in new cities. Unless they are in dire need of some resource the AI should always be building unrest and production buildings before everything else, and they never ever do even though there have been many threads about it. This cannot be overstated.

Secondly, the AI still doesn't use specialized cities at all. I have played a lot of games and let me tell unless an early empire without both a fortress and a conclave is seriously handicapped. If you even have 1 of every city type and properly specialize them you can keep up with huge empires but without 1 of each you will fall behind hard. The AI not only is usually missing one type unless it gets a lot of cities but it doesn't use them optimally at all. Every city type has a single purpose that takes priority over everything else. However the AI seems to just build randomly from them all. Even claiming world resources is almost always a waste of time early game unless the city is already specialized and has basic improvements already.

Towns are for producing gold, food, and usually pioneers. After production and unrest you should build gold buildings and food buildings in that order. Only then should other things be built. This actually means that they are the least necessary. Don't build them first unless you think you will have lots of room for the other types. They can be built on suboptimal locations like 2/3/0 with no real consequences. Unless you have no fortresses you should not be building troops from this city type.

Fortresses are built in high production spots and having some essence is a bonus. This is where all your troops should be built. However building the improvements that help build troops like workshops and barracks is first necessary.

Conclaves are very important for research. Without one early game your research will be very very slow, and you will fall behind. Build them in places with essence but even without it you can still get a fair bit of research from them early game.

Simply put all cities should prioritize in the below orders.

First Village

Unrest-Production-Pioneers-Growth-Other

Village

Unrest-Production-Resources-Other

Towns

Unrest-Production-Resources-Gold-Food-Growth-Other

Fortress

Unrest-Production-Resources-Training-Units-Other

Conclave

Unrest-Production-Resources-Research-Mana-Other

*Resources means World Resources around the city and includes monuments if they would reach them. Obviously these should only be built if the AI needs them.

If the AI could just grasp how to manage a early game empire of only a few cities then it would be much more challenging. Sometimes through luck it will do the right thing and be very dangerous but most the time it doesn't come close and stalls hard. This early game lack of tech and units is what gets it's champions killed and allows lairs around them to evolve and trap them.

Oh I understand that but even even increasing the weight on these items would really help. The guidelines I give above are very general and basically always the best thing to do. I don't expect the AI to be perfect and never make mistakes but right now it is not so much making mistakes as playing completely wrong. If it even loosely held to the above strategy I think it would play much much better.

I build them all the time and they are very useful. The problem I have is the AI likes to build them in cities that don't even have workshops yet. In fact they almost never build bell towers or workshops. They spend so much time building stuff like monuments, studies in Fortresses, or units in Towns, that they never develop those cities. Build the monuments or whatever after the city has basic production and unrest buildings, then it will be built faster and not delay the development of the city. The monument is just not a priority.

A Fortress with production buildings and training improvements can pump out a lot of buffed units really fast. But instead the AI likes filling their queues with studies, shards, monuments or whatever. They don't even get production improvements so it takes forever to build them. Then they train their units in their Towns and Conclaves which in turn prevents those cities from developing because they don't build fast enough to have time for units and improvements. It's just a horrible way to play. The AI needs to realize that it's better to build some improvements before others and to specialize their cities. Even if they lack research they should not be building studies in Fortresses if they have Conclaves. 

I don't know about any of you, but I definitely build studies in my towns and fortresses, and often the upgrades of them.  It is too simplistic to say build a study in Conclaves and skip it in towns and cities.  Yes the conclave will ideally have many more research than what the fortresses and towns will be able to put out, but they still produce a decent chunk of research themselves.

I am not entirely sure how much of a difference it makes that the AIs don't build the unrest buildings.  They might be programmed such that unrest doesn't really apply to them.  Even if it does, players can't hope to be challenged by Challenging and below difficulties anyway.  When you amp up the difficulties above that the AIs bonuses start to eclipse the penalties from unrest even if it does apply to them.

Plus, there needs to be some element of randomness to the whole thing, because formulas are going to be exploited by the players mercilessly.  The last thing we want is players developing opponent specific build orders.  Making the AIs plays random to some extent at least helps reduce the chance of that happening.

I agree that it would be nice if the AIs would build more units in fortresses and less in towns, but it just wouldn't do to restrict them to only building units in fortresses.  They could never even hope to defend their empires this way.  Part of the reason they will build units in towns is on the off chance they don't have a fortress nearby they want to be able to get that city up to 6 or 9 or whatever additional defenders so they can try to keep control of it.

They need to do that, because they are horrible fighting multi-front wars which are actually quite common for them to get themselves into.  If they didn't have city defenders in every city they would have zero hope of being able to keep control of any of their cities any time they were fighting 2 or more opponents at once.  See above, their best hope of getting defenders into a city is to just make them in the city in question.

I do build Studies in my Fortresses, but most of the time I build troop enhancing buildings first, and if I need units, I'll be training them rather than building Studies. (The big exception is if my capital is a Fortress, +1 Research at the start of the game is actually quite a lot.) I get the impression that the AI doesn't really specialise its cities like a good human player does.

Certainly I've captured large AI cities late in the game which had built almost everything conceivable except a Workshop, which seems a bit wrong. It's a fair point that the AI might not be using the same Unrest rules as the player, but I assume more production is always a good thing. Perhaps the AI should build production increasing buildings (regardless of personality) if any other buildings would take more than twice as long, or some other such rule of thumb.

It's not unrealistic to expect an AI to play to WIN? That seems to be what's missing a lot in games. They are just delaying action opponents.....one is going to win (human) if he has any brains at all. I sometimes wish I were some of these guys that's having hard times against the ai, but sadly it's the same ole outcome every game. I have 2 say though the ai in Gal CIV II on higher than normal difficulties was pretty good yet I could still get them into fights with one another and I just think that should be taken out of games. Ai's are already too easy as they are let alone the ability to get them to fight one another. The other thing was being able to SEE what they were teching up in. If I can see what they are doing without actually spying...why should I invest in spies. I could always see what tech they were teching up in and sell my tech to them right at 1 turn to go. I told you about this time and time again and you never seemed to listen or take that ability away. It's on that list where it tells you about the AI's and the specific things about them and what they are teching up in is on that list at the bottom.

So, I don't think it's unrealistic to want you to take exploits out of the game. There are so many holes it's unrealistic in that aspect sometimes. Whereby you can make a good fighting ai but you forget all that surrounds it. Look for the exploits, play for the exploits and then do something about them. That's all I'm asking. Hire some experienced testers, Players that will look for exploits for you. Fix them. )

Also, I don't know if the game has this ability but I love it in SPARTAN 1.013 and that was the ability to choose how the ai built and in what order. The main developers have the ai going so much research and civ stuff in the main program that the ai always was building craap units. I went in that program cause they let us and changed the order of build to more and better units and put research and civ stuff in its place behind units. I made one of the best ai mods I'd ever played against.. The ai started producing lots of better units and armies. Hordes of them. Drove me out of Athens onto a lil island. Of course I was still playing on hardest and impossible levels but the amount of armies and units was what I was looking for and surrounding me and attacking me from 4 sides was awesome fun. No exploiting that ai cause I never had time to look for exploits I was defending myself so much. That's what ai's need to be like, hate humans is 1st and ally and attack them at all times is 2nd. Then you'll have a good ai and crybabies all over the forum claiming the ai is too hard.....even on easy. That's what they did in the SPARTAN game and made a 1.017 cause players cried the ai was too HARD.. lol I could not believe it. Even on easy they were crying it was too hard. lol cyrbabies. LOL

I think Willie Sanderson is in a group of players that constitutes roughly 0.0001% of all players.

I think Willie Sanderson believes he is in a group that constitutes roughly 99.999% of all players.

I think that is the major disconnect here.

I think if Stardock implemented his ideas, it would piss off the 99.999% group that he thinks he is in while making the 0.0001% group more happy.

I can't think of any better reason not to use any of his ideas.

Good lord, can we ban this willie sanderson character already? I don't care what it says under "Join Date".

I want to say otherwise, because he's basically a ridiculous troll - but the general idea of finding a way of encapsulating expert players' decision trees into AI scripts is a good one. That's basically what put Civ IV over the hump.

Restrict the "help" the AI gets to start at Challenging and ramp up from there.

Where can one mod the ai DsRaider? To select the way the ai builds things. Another game I modded years ago was Kohan II. Look for a Ravinhood101 SAI mod on the now defunct Gamespy. They ran a test on numerous player made AI's and mine was in the top 10. I didn't get #1 because they also used my unfinished and unworked really Huaron mod I was piddling around with and it was attached to my other mod.

But that's no matter, if I can adjust how the ai builds armies and production in this game that would be most beneficial. But, Brad has to make he AI aggressive in the first place. That was the thing about the ai in SPARTAN it was already aggressive but it just built everything stupidly. It's armies before I got ahold of it consisted of weak line units. I made it produce more elite units like Hoplites. An ai that just builds to delay is no fun, an ai that builds to WIN is a lot of fun even if it does make silly mistakes from time to time.