stupid AI

I believe that Technological Victory was the solution to this problem that way when they reach the end of the tech tree its over.  I have to assume you turned it off so my advice is to turn it back on and try not stall the opponent out. Instant fun and challenge yay !  At the point in game where you have all techs and you're going for a military victory the AI is at a disadvantage anyway, because you can design ships specifically to counter their weapons and defenses(they stick to one of each even when they have all types in my experiance) to my knowledge the AI doesnt do that. Instead they stick to the prefabs. So yea tech victory on and beat them down before they get there.

Yes, the Research Victory seems essentially to serve as the game designers' feature to defeat the sustained stall strategy.  In which case, it may not be entirely fair to turn that Victory Condition off and then make that specific criticism.  I am not sure the programmed opponents alter their styles in accordance with victory conditions; I doubt it.

Note that the All-Factories strategy is another variation on your theme.

If you turn technology victory on with the technology speed at anything above normal, it's not even a game.

And there is a game option to play with all tech already researched.

If you play a very large game with many planets on a very large map, the only way to allow the game to play out in a military fashion is to turn tech victory off.

So if what you want is a large military game, you cant get it caues the AI is only good until the tech race is over.

Playing on the biggest maps, with tech victory on, and the tech speed normal or faster, there is little military involvement in the game.  YOu just hold everyone off while you race the other civs in research.

This is a military game.... but the AI cant deal with playing correctly once the full tech is unlocked? 

That's pretty lame in my opinion.

If the game had multiplayer internet support this would be a small oversite.  But as a single player only game, it takes this game from what I thought was a really cool turn based stragety game a few weeks ago into something that probably wont even get launched again.  Bummer.

Well while playing on the biggest maps with tech victory on instead of sitting there hitting the end turn button   take the ships youve been useing to hold off the opponent and go kill some off that way there's plenty of hot military action   in 1/3 the time of sitting in a long drawn out cold war waiting for end game weapons and the inevitable ai drop off you speak of.  Start wars early lots of fun in that have everyone fighting everyone else if you can, that'll slow them down.  Also I dont really see how you could hold off the ai while researching the whole tech tree with out at the very least matching them militarily could just be me though. 

I think the point is you should not have to carry the AI in that manner.  The AI needs to adapt to any strategy you adopt, especially a basic one like he was where you re-tool to fight a long, drawn out war.  The AI is broken in that regard because even given all the techs they still build a huge number of research facilities from the outset.

The AI does a pretty good job of expansion and occasionally springs a surprise on you by showing up somewhere you did not expect them, but it really never puts up a terribly great military fight.  It builds far to many defenders, never concentrates its forces in a manner to really threaten you and often times if they do send a few powerful fleets at you their transports are lagging turns behind and are easy pickings for the opponent.

In my experience the actual deployable military force for the AI is like 1/2 of what is listed.  Usually 1/3rd or more is composed of defender ships that are useless in offensive action.  I have never used defenders on any world.  I always just develop mobile fleets and a reserve of transports in case something happened to get through.  I may have half the military rating of the Korath but all of mine is in fleets rather than defending every world on the map.

It is such a basic strategy as to be laughable really.  I cruise into a sector with 4 or 5 planets.  One one is a major enemy fleet.  I casually move from planet to planet wiping out defenders, mines and starbases.  Their fleet never leaves orbit.  I leave unharmed and they begin rebuilding the same so we can do it again in a year or so.

I am reminded of what Napoleon said upon being presented a plan to defend France that spread troops evenly across the border "Is the idea to prevent smuggling?"  Concentration of force is one of the first military principals out there, particularly in naval engagements.

The AI is seriously broken on the military side.

I have a follow-up question.  Is there a way to tell what the AI slider settings are on any particular turn?

If all has been researched and all buildings built, the AI may simply reset the sliders to 100% military (ship) production, rather than divert some to fund the social production necessary to convert labs to factories.

As long as the AIs did that reset (100-0-0), the Suicidal bonus should make them pretty tough, given all the industrial sectors and stock markets they generally build

If the sliders are left at something like 33-33-34 even after research is complete, that would seem not hard to patch.

Jeez, you guys don't want much, do you?....just an AI that is as smart and adaptable as a human player. What are you running your games on, a HAL9000??

And multiplayer for a strategy game as layered and complex as GalCiv ABSOLUTELY WILL NOT WORK! its hard enough playing real-time strategy games on-line without one or more players spoiling or quitting or sabotaging, never mind a turn based game. I used to play MOOII on-line, and that is a much more basic game than Gal Giv, and I think out of the approximately 20 games there were maybe 2 that got played out to something resembling a finish. Players would quit, or stall, or not be available when the other players were, or spoil the game by surrending  their powerful empire to another player (usually pre-arranged) or cheat by communicating with other players outside of the in-game messaging...you think its frustrating putting up with stupid AI, trust me that putting up with stupid humans would be far, far more frustrating.

It doesnt matter what the sliders are at, it's about the AI's planetary improvements.  They build extensive research building which then drain cash and produce nothing.

Meanwhile I've converted all my research planets and research tile improvements into military production planets.  My military output dwarfs one of these AI's even with a fraction of the planets.

First of all I dont want a HAL9000 AI.  Simply an AI that can play with some sembelence of skill with the full tech tree.  Leaving hundreds of research building all over your planets when the tech tree is full is just horrible gameplay from the AI.

As far as multiplayer that's just flat out wrong.  There are plenty of turn based strategy games as complex and MORE complex than this one that play just fine multi-player.  The civilization series comes to mind...

Yuhjn1972 -

Please re-read my slider comment.

If the AI has the sliders on 100 mil and zero everything else, that is a simple programming compensation at Suicidal to keep the AI competitive.  I think the AI gets something like 300% as a base mil production at Suicidal.  So, each AI factory is worth 3 of yours, each 3x tile with a factory is really a 9x tile, etc.  Have you viewed an AI mil prod planet and compared its shields to your mil prod planets?  Count its factories, etc, and compare with one like it of yours with similar # of factories.

Yes, the AI would do even better to divert some production to converting research buildings to factories, but the AI is using it all to make ships IF THE SLIDERS ARE 100% MIL.

Thus, if the AI is not a challenge then, I suspect the sliders are NOT 100% mil.

Wait!  Can you tell from the Reports or any other intel screen what the AI is generating in Research points?

It would be pretty simple to write better military routines in for the AI, particularly in regards to its defensive and offensive planning.  The route taken was really fairly lazy as the defender ships are just a crutch to keep the AI from getting totally rolled.  It really is just insane that on high difficulty levels I can roll in and strip 3 or 4 worlds of their single defender ships and invade while a 6 or 7 ship fleet made up of battleships just sits there on another planet in the system and watches.  It is equally insane that the enemy will invade your space with the occasional surge of fleets but more often than not sends no transports with it, or they lag so far behind they are easy pickings for light forces.

What is needed are routines to have the AI engage in a proper fleet defense of certain areas rather than just scatter defenders across all of its planets.  When I come crashing across the border with 100 huge ships or more in the end-game I don't want to be met by piecemeal resistance where I take out the enemys fleet in ones and twos for the most part while running across the occasional planet with 4 or 5 ships.  I should be met by a fleet containing the majority of the enemy combat power, especially if I was overtly building up on the border for over a year.

Instead I overrun everything within a couple weeks travel of my borders and consolidate before a couple of powerful fleets come trickling out of the enemy's core worlds at the rate of 1 or 2 a week.  They cause problems but can't undo what is done.  Even if they have transports to retake many of the planets I am going to strip the worlds bare first so they are of no value.

My point is no one would look to defend like that.  If you can travel between systems inside of a week then it makes sense to cocentrate your forces and actually look to hold someone on your border, rather than engage in some suicidal policy of fighting it out over your own planets.

I think you vastly underestimate the coding tasks.

Vastly.

For example, what if you encroach with one fleet, spot a huge reaction fleet, and split into multiple TFs?  What if you cross the border with two fleets a turn worth of hexes apart?  What if you enter with one fleet, stop, and build a big mil SB deep within the AI borders and keep your fleet within its umbra?

What if one player fleet is armored and the other is shielded?  What if one fleet has transports and the other does not?  What if the transports are ~empty?  What if one fleet is very lightly weaponed but much faster and might slip around?  What if that fast sneak fleet gets into the back areas and refits into a heavy hitter?

This is tough coding stuff, and why AIs need bonuses to cope.

I would not complain if the AI were not responsding properly to feints and pincer moves and other more advanced tactics.  But it is not even necessary for the player to exert themselves that much.  I can make a blatant show of things for months prior to jumping the border, declare war prior to doing it and the AI is still unprepared for it and has its defensive forces scattered all over its empire.

As far as I can tell the AI makes no effort at fleet defense at all.  It has fleets that will attack you if the opportunity presents itself while they are on their way to do something offense, but defensively I hardly ever see them employed.  They just wait on planets to be eaten in small chunks.

I have said other places the whole fleet structure needs to be rethought to give the AI a fighting chance.  As it stands right now the AI basically needs to have out-built you by a 3 or 4 to 1 military factor to really have a fighting chance unless they just way outdistance you in weapons and defenses.  Once the tech tree gets maxed for everyone it really becomes a mess.  The AI can certainly wear you down, but it really only employs about 1/3rd of its military against you on offense.

Hell, I have had the AI wander a whole fleet right by an area where I was marshalling nothing but unarmed transports, like 30 of them with 2,000 troops each and it just crusied right on towards my pretty useless military starbase to conduct an attack there.  I was just floored.

I agree that the AI will ignore doings within your borders, including military buildups.  Also, the AI will react especially ineffectively during the initial invasion.  In fact, it reminds me Stalin's behavior during June-August 1941.  Papa Joe (per some biographers) closeted himself for three days after the start of Barbarossa and would not speak to anyone or give orders.  In the days ahead of June 22, 1941, Soviet commanders had been refused permission to patrol their own side of the border, were told not to react to provocations (some even were orderd into barracks and their weapons stored), and the fleet admiral was refused permission to up-anchor and stand out to open waters.  Incidentally, most of those same Soviet border commanders were later executed for failing to do the things they asked to do but were told not to do.  Stalin and his henchmen wanted to cover their tracks.  The only notable exception survivor was that Soviet admiral, who had insisted on getting those orders in writing.

I have played games where the AI shifted a force from destination to destination, or target to target, without ever getting to use those units on anything in any sort of timely fashion.  That, too, has happened historically, and even did so a few times on the German side of Barbarossa, mostly due to Hitler's interference.

The choice of Barbarossa as a source of examples is not w/o basis.  That conflict is the probably the closest thing we've had on this planet to a GC2 invasion.  Consider: huge borders, great distances, vast forces, lots of machinery needing ammo and fuel, fast communication yet limited intelligence, both sides under the command of a single decision-maker.

Also, I agree that the AI will often not react to visible enemy units, and stupidly continue on in conformance to its original orders.

One of the D-Day anecdotes involves a line soldiers tramping along one side of a wall and another line of soldiers tramping along the other side in the opposite direction.  At one point, the wall was lower, so that they could see each other.  The two units marched by on their separate ways and only too late did one or both realize that the other was an enemy unit.  When some turned around, the other was already out of sight.

At Jutland - arguably the greatest sea battle since the advent of steam, during the night after the vast but inconclusive battle (May 31 - June 1, 1916), the British admiral (Jellicoe) tried to put the Grand Fleet across the expected path of the German fleet to port.  The German admiral (Scheeer) chose a somewhat different course, but the German fleet still passed through the British line.  Many, many British sailors and officers and even ship captains and maybe even junior admirals saw the Germans but no one reported it.  They all figured that if they could see them, the admiral in charge must and they would do nothing without orders.  A few separated British ships actually fought the Germans during the night because the Germans essentially (and in one case literally) ran them down.  Yet the British admiral never was told and the Germans got through the line and back to port.

If humans suffer these things for real, imagine trying to code the poor AI to do much better!

I agree there are issues with it, like I said I think any fix would have to be a result of totally reworking the fleet system.  Fleets should be much larger (I would prefer them be unlimited in terms of size and have logistical ability basically serve as a multiplier of effectiveness) and should be able to be deployed in defense of a number of systems.  The player can do this, at a great deal of micro-managing currently.  I simply propose to automate that as much as possible.

Say I have 100 ships to defend 6 Systems with 3 planets in each.  Right now the AI pretty much just spreads ships across those systems, with the occasional concentration of ships it has not launched yet on production worlds.  What I would propose is a pretty simple system.

Those planets are assigned to a fleet, lets call it 1st Fleet for simplicity sake, and that fleet defends it according to some simple pre-set paramaters.  When you create the fleet zone you simply answer some questions to setup your defense and the AI would be able to do the same.

So you setup the area and assign certain ships to it.  You then are given a list of planets and installations in the area to which you can assign ships, along with an option to create a mobile response group.  Let's say you have a 15 factory production planet you want to protect so you assign 50% of your forces to that and select an option to either have them stay in orbit at all times or only engage enemy fleets that allow a return in the same turn, the next turn, 2 turns ect.  There would be other options to allow the system to automatically handle dispersing some ships on the borders for sensor coverage and making sure there was 1 or 2 or however many ships on each planet ect.

The last set of options simply determines how your forces react.  Do they attack anything that enters the area if within the allowed zone?  Do they only attack weaker forces?  Do they run from stronger forces?  If so where would they rally to?

This is something simple for the AI to do.  It simply makes an assessment of the relative values of the planets in a certain area and assigns its defensive forces accordingly.  If there is nothing of exception value and it is more of a border post most of your forces to a mobile force.  Everything is a simple yes/no force assessment.  They may not be as good as the player at it but it is still better than what exist.  If their mobile force was told not to engage anything it was weaker than and in the event it spotted such a force to rally to the most protected planet then at least the system is making a sound decisions, rather than just waiting to be blown to bits in pieces.

The programing, while challenging, is still a bunch of yes/no questions that the game already does on a global basis when deciding to declare war or not and so on.

Hi!

BigJKU, you obviously don't know that the AI stands for Aritficial Idiot.

Seriously, AI was programed mostly by one person (Brad Wardel) with the intention to give a casual player a good run for his money. If you're experienced strategist, AI simply isn't in your league.

Eh, you obviously haven't coded anything similar. The main problem to Yes/No answers is "what the question is". In GalCiv you don't play the open info game like chess, but a game, where uncertainty plays a big role. So IA needs to make  A LOT of assumptions, EACH of it needed its own code. AFAIK Brad loves coding AI, but he probably loves playing "Real LIfe" more. So until you come out with a very detailed block diagram of situation analysis and another one with detailed "answers", will AI remain an idiot. Currently only we humans can learn and adapt in such a game.

BR,  Iztok

  Great one Iztok!

Plus the fact that in the case of chess, the AI can simply calculate all possible scenarios in the next few steps + assess the "value" of each of those situations.

In an open world like GalCiv II this is not the case, so many variables, so many possibilities even more so since there are events that can spark wars unexpectedly...