AI cheats

from Sins Forums

I played SOASE a few years ago and always found it compulsive gaming. I came back to it recently and still found it addictive, but there were traits that left me disappointed. Those traits were AI cheats. These are used to defeat the player and prolong the game. No matter what level of difficulty is played, the AI always has the upper hand and tricks on hand that are not available to the player. Take that pesky little AI scout, which pops up everywhere like a flea just to irritate the player with no warning at all (no red line), and which can determine the victor of a battle or even game.When the humble AI scout pops up from nowhere it is generally to delay a jump at the enemy and it can take a massive fleet already present on the planet sometime to eliminate it, while the program is deciding how to answer the player's threat. This ruse is not available to the player, as the scout is eliminated in the middle and late game within seconds, even with the jump drive upgrade, so there is little current intel available on the AI's forces. The player can use, say, a Cobalt for same reason, but this usually has the same fate and can be an expensive option when the action has to be repeated many times to know what the current state of play might be with the enemy. The delaying tactic used by the scout is compounded by a faulty team jump. On one game  I team jumped to the enemy and, instead of sending the team, the program first sent one Robotic, which was eliminated with in seconds and then sent my Dunov ( which was supposed to jump with the Kol) and was annihilated with seconds before the rest of the team jumped. I lost the battle though I had calculated a slim victory. I suspect that the program had also calculated a victory for the player, so it sent in forces with no support first in order the player would lose. Timing also seems arbitrary in favor of the AI in production, research and jumps. The AI can jump without a problem when the player has, say, sent a Cobalt to the planet to delay the jump. When the player wants to jump and is delayed by a scout, the player cannot choose one unit to deal with the scout and jump with the rest of his team, but the whole fleet has to be occupied to crack a walnut with a sledgehammer. The AI can produce a cap ship in battle within 5 seconds if necessary, but the player takes 35 seconds to produce the equivalent. The balance in this game, as in many other good computer games, can be tedious. The equations are so exact that you can guarantee your capital ship ( or the equivalent in support) will be taken down, if the player has defeated the enemy's cap ship, even though the forces opposing it are weak. A strong ship (in hulls, armor, ballistics, antimatter etc) is just eliminated within seconds. In some cases the enemy is just not affected at all by the opposing forces.  The question arises, how do you determine a winner, when two equal forces are opposing each other and doing the right thing? It is a difficult question to answer and I know nothing about programming. You might try to exploit a weakness, but the intel is not available to the player to do this. I played two games with no outcome, no winner. These were 1v2 factions. I ignored the diplomatic missions and defeated the two enemies, but the pirates would not go down despite superior forces. I built three Novaliths which had no affect. I had to give up the games. I suspect that, because I ignored the diplomatic missions ( though some were achieved by default), no victory was offered.

I hope this thread is not too long-winded. I may buy the beta of Rebellion from Steam, but I wonder if the scenario above has been improved?

20,155 views 9 replies
Reply #1 from Sins Forums Top

michaelwalters,

Buy the beta. It offers a lot of upgrades, revisions, balances and otherwise. I find the best strategy to beating the AI is patience, overwhelming force and EVEN MORE PATIENCE. However, don't worry about calculations. Be willing to work on things by trial and error until you find your own groove.

Reply #2 from Sins Forums Top

The AI only gets cheats above the normal difficulty level. And I really have no idea what you are talking about with the scout, the only thing the AI can do with them players can't is to let other units use the scout ability. It is quite easy to scout the AI, especially with the probe upgrade for the TEC scout and the lingering visions upgrade on the advent scout.

Galactic Civilizations III
Reply #3 from Sins Forums Top

First off, go through your wall of words and make it legible by adding paragraphs in any future posts.  I'll respond in blue.

 


I played SOASE a few years ago and always found it compulsive gaming. I came back to it recently and still found it addictive, but there were traits that left me disappointed. Those traits were AI cheats. These are used to defeat the player and prolong the game. No matter what level of difficulty is played, the AI always has the upper hand and tricks on hand that are not available to the player.

Wrong.  Only Hard and above cheat.  Easy actually is gimped compared to the player.  If you lose to an easy AI, you're losing to a computer with half the resources you do that can't cheat.  That's inability.  That's your fault, not the game.

Take that pesky little AI scout, which pops up everywhere like a flea just to irritate the player with no warning at all (no red line), and which can determine the victor of a battle or even game.  When the humble AI scout pops up from nowhere it is generally to delay a jump at the enemy and it can take a massive fleet already present on the planet sometime to eliminate it, while the program is deciding how to answer the player's threat.

What on earth are you talking about?  Scouts have almost no military presence unless you're talking about them en mass, but you're talking about solitary scouts.  Just ignore scouts or let your strike craft handle them.  If your enemy has a fleet large enough to wipe out your scouts, you ought to have an equal sized fleet at minimum.

This ruse is not available to the player, as the scout is eliminated in the middle and late game within seconds, even with the jump drive upgrade, so there is little current intel available on the AI's forces. The player can use, say, a Cobalt for same reason, but this usually has the same fate and can be an expensive option when the action has to be repeated many times to know what the current state of play might be with the enemy.

See above.  You can do it, and if you fail at it, that's your fault.

The delaying tactic used by the scout is compounded by a faulty team jump. On one game  I team jumped to the enemy and, instead of sending the team, the program first sent one Robotic, which was eliminated within seconds and then sent my Dunov (which was supposed to jump with the Kol) and was annihilated with seconds before the rest of the team jumped. I lost the battle though I had calculated a slim victory.

If this is a concern, simply set your ships to jump simultaneously.  Once again, your fault, not the computer.

I suspect that the program had also calculated a victory for the player, so it sent in forces with no support first in order the player would lose.

Entirely possible.

Timing also seems arbitrary in favor of the AI in production, research and jumps. The AI can jump without a problem when the player has, say, sent a Cobalt to the planet to delay the jump. When the player wants to jump and is delayed by a scout, the player cannot choose one unit to deal with the scout and jump with the rest of his team, but the whole fleet has to be occupied to crack a walnut with a sledgehammer.

If you're talking about the desire of fleets to auto-target anything that jumps into a gravity well, just re-order them to jump.

The AI can produce a cap ship in battle within 5 seconds if necessary, but the player takes 35 seconds to produce the equivalent. The balance in this game, as in many other good computer games, can be tedious. The equations are so exact that you can guarantee your capital ship ( or the equivalent in support) will be taken down, if the player has defeated the enemy's cap ship, even though the forces opposing it are weak.

Harder AI's do actually cheat in income, build times, and capital training, but not that much.  Even Vicious difficulty AI's can't put out a capital in five seconds.  Also, you're severely overestimating the power of capitals.  Capitals are never as efficient per cost at putting out damage as frigates are.  Capitals are only useful ultimately because of abilities.  Killing a capital in no way guarantees victory.

A strong ship (in hulls, armor, ballistics, antimatter etc) is just eliminated within seconds. In some cases the enemy is just not affected at all by the opposing forces.  The question arises, how do you determine a winner, when two equal forces are opposing each other and doing the right thing?

If no one is winning, no one is doing the right thing.  As mentioned above, use scouts.

It is a difficult question to answer and I know nothing about programming. You might try to exploit a weakness, but the intel is not available to the player to do this.

Programming isn't relevant here.  That said, you should exploit weaknesses.  Check the stickied things in the strategy section of the forums to learn what counters what.  Once again, the intel is available if you actually bother to get it.

I played two games with no outcome, no winner. These were 1v2 factions. I ignored the diplomatic missions and defeated the two enemies, but the pirates would not go down despite superior forces.

Don't try to go against pirates and their defenses at the same time.  Draw their fleet out first and defeat it before going after the turrets.

I built three Novaliths which had no affect. I had to give up the games. I suspect that, because I ignored the diplomatic missions ( though some were achieved by default), no victory was offered.

You can win by wiping out the enemy.  Apparently you missed something.

I hope this thread is not too long-winded. I may buy the beta of Rebellion from Steam, but I wonder if the scenario above has been improved?

End of quote

Reply #4 from Sins Forums Top

Thanks for your replies.

Hemomancer: I probably will buy Rebellion. I am weak! I like the game too much not to. Thanks for your advice. If I lose a game, I want to know why. I have looked at the forums extensively and there is much contradictory information out there so, as you say, success has to won by experience. This is hugely time consuming as some games have lasted 8 or so hours. This is one of the things I like about it, it has depth, and the other strength of the game is bitter/sweet decision making. 

 

Goafan77 : I should have made it clear I have only played TEC.It is something I am trying to master before using another faction. The sensor probe takes a snapshot of the planet, but this lasts only a minute or two, which may not be enough time to fortify or attack, depending on what info (intel) is provided. The probe does not guarantee survival of the scout and in the middle to late game it is eliminated within seconds, unlike the enemy scout. I have only played easy and normal difficulty. I have just gone back to an ongoing game ( I am using Trinity) and everything I said still applies.

 

Volt_Cruelerz : Thanks for your detailed reply. I have only played normal and easy with TEC. I have won quite a few times on a small map, but have lost on a small map with a different scenario and even on the same map. You have misunderstood my point about the AI scout. It always pops up from nowhere when I intend to jump the enemy and, mousing over my fleet at this point, shows they are all occupied in fighting the scout. This delays my jump so the enemy can fortify or provide an answer. In some cases, even when the scout has been defeated or left the planet, my fleet reports 'gathering information' for a further delay. Sometimes I have to repeat the order and guess what, another AI scout shows up. A whole fleet to defeat a scout? It cannot be ignored and I have no choice in the matter. I can't choose, say, a Cobalt or a scout only to fight the AI scout and then jump with the rest of my team. When there are only two opposing scouts on a planet they also fight each other.

When I team jump I have set all the units to jump simultaneously, by right clicking on the move button. I have tried re-ordering the fleet to jump (which I should not have to do), but they will not jump until the scout or other weak single unit has been eliminated. This can take up to 30 or so seconds to happen.

Don't you think it is cheating, when I have ordered a team jump, for the program to send some strong units by themselves, before team jumping the remaining fleet?

Why do you think I have overestimated the importance of the capital ships? In most cases I have only four capital ships in the late game on a small map with a strong support fleet such as Ogrovs, repair Robotics, siege frigates etc., unless I have discovered the enemy is building more capital ships. This may be a weakness or a strength. The game has to balance the equations, no matter how unjustified this may be. If an enemy capital ship is destroyed, you can guarantee yours will follow automatically or the equivalent in support fleets, even when the opposing forces are weak. One tactic I have used is to move my capital ship when the shields are down, which buys some time. Even on easy or normal difficulty, my fleets can be eliminated by enemy forces not capable of doing this. I may have got the precise times wrong, but production of fleets by the enemy AI is far quicker than for the player in battle.

When you say 'If no one is winning, no one is doing the right thing' is this really true? Doesn't the AI do the right thing for the faction it is representing? I repeat my original question : If two opposing forces are doing the right thing, then how is a winner determined? Most of the SOASE games I have played have been wars of attrition. There is no quick fix here. I was simply wondering if the mathematical algorithms used allow for errors in calculations so one side or the other can exploit them. It must have error built in. So far as I can see, so called 'strategy' computer games must cheat to be competitive. 

You may be right about the pirates, but why did the Novalith cannons have no affect on the pirates planet?

Nothing you have said changes my view in my original posting.

Thanks again everyone. 

 

 

Reply #5 from Sins Forums Top

What you're referring to regarding the scout is auto-target priorities.  For whatever reason, the fleet is known to do this.  You just have to re-order the jump.  The reason that it doesn't doing anything to the AI is because they'll just reorder the jump.  There are effectively two levels of AI.  Tactical (basically, this is auto-cast) and Strategic (the ones you actually play against).  Sometimes the tactical AI your ships use is dumb.  This is just one of those instances.

Regarding fleet jumping not working correctly, that's an issue with rotation.  I always order my fleet to do one final forward movement command towards the target well before jumping so that they're in roughly the same direction.  This speeds up jump time and makes jumping more uniform.

I may have misunderstood your OP then if you only build four capitals.  No offense, but it's very difficult to follow that block of text.  What you need to know is that humans are significantly better than AI's, meaning that you should be able to micromanage your units more effectively than the AI can for better results.  I was playing against a turtling TEC yesterday and attacked with one third the units as Advent and suffered only minor losses.  It's about how you use your ships and abilities that allows humans to win on higher difficulties.  Vicious AI's have something like 8x the income human players do for instance, so the only advantage humans can have over AI's is intelligence and critical thinking.

The AI often times does stupid things.  Humans do as well, but a human can learn from one game to the next whereas the AI doesn't.  There often times is no "right" way to play.  There is only action and response.  For every act that you do, your enemy will react and you must then react to that.  The way you act and react determines who wins.

Reply #6 from Sins Forums Top

Volt_Cruelerz : Thanks for your reply and advice.

I will try next time to move forward towards the enemy first, so the jump may be more uniform. This still leaves the problem of the enemy scout popping up from nowhere to prevent the jump. I have not played against a human player before, as my knowledge of opposing factions and what is best to oppose what in battle, is still only limited to TEC. Mostly I have found it unnecessary when playing the computer to micromanage the battle. I do hot key the capital ships in order to move them about out of the heat of the battle when necessary and place the support fleets on hold, particularly when I have placed a jump drive inhibitor -- can be an effective strategy. On one occasion I misplaced the jump drive inhibitor, only to find the enemy jumping at 45 degree angle and so avoiding the inhibitor.

Diplomatic games remain a problem. In the few games I have played, the missions given by the other factions are almost impossible to complete without heavy losses the other faction is then able to exploit. Usually both factions (1v2) give missions at the same time late in the early game. On most occasions the faction I am supposed to fight is three or four planets away so by the time I have expanded further, gathered credits and resources and researched, fortified all of the exposed planets, the time to complete the mission runs out, and I am attacked by both factions. If I don't fortify either faction will jump in, even if they have a margin of 'friendliness.' At first I thought the faction giving the mission would not attack while I am trying to fight the other, but this is not true. I need to look at it more closely later, to see how this catch22 situation can be averted.

Reply #7 from Sins Forums Top

Some abilities should really be micro'd.  All abilities on the Kol for instance drain so much AM that you need to micro them for instance.

You can always reject the mission of one to attempt to perform the other.

Reply #8 from Sins Forums Top

Quoting michaelwalters, reply 4
The sensor probe takes a snapshot of the planet, but this lasts only a minute or two, which may not be enough time to fortify or attack, depending on what info (intel) is provided.
End of michaelwalters's quote

FYI the probe last 10/15 minutes, which is quite a long time.

Reply #9 from Sins Forums Top

GoaFan77 : I am sure your right about the probe. Time moves fast playing this game and what might seem a few minutes may, in reality, be much longer. I have only used the probe upgrade once. The scout in the middle to late game hardly ever gets through, but the enemy scout always does even with jump inhibitors in place. I rarely, if ever, get to recon the enemy's home planet, as the scout usually has to pass through enemy lines before it reaches the enemy's home planet with no short circuits available.

I will probably download the Rebellion beta tonight on Steam. I am a bit concerned about whether my computer can handle it, as the recommended specs are more than double what is required for Trinity. I have an AMD 6850 graphics card, 4gigs of DDR 3, and an AMD 4x 955 processor. In the late game it may falter with so many units and fleets available.

 

Volt_Cruelerz : I'll have a further look at the abilities of the Kol and try to decide what is vital and what is not and how that will affect the progress of the battle. It is useful to know that the type of game can be changed in-game, though I wanted to try my hand at the diplomatic game.

I have just won a military victory after 6 hours of play and there were some oddities in the game. When a fully maxed out Kol, in ballistics, armor, shields etc., jumped onto the enemy planet, it was immediately reported that the 'shields are down.' They are a wicked lot at Ironclad as the Kol was fully intact. (I had the right cap ship) This creates tension along with other commentaries, delays and even the cheats.

I tried the forward movement of my fleet to see if it speeded up my jump. This was a star, so the fleet moving towards the phase line direct to the enemy crawled around the rim for what seemed ages. I had to give the synchronize movement four times before the jump happened. When it finally happened, four Robotics were sent ahead. Robust as these may be, they still need support, but happily the rest of the fleet jumped soon after. I was on the verge of victory and I doubt if the game would have been so kind earlier in the match.

I will probably download the Rebellion beta tonight to see what improvements have been made. I wonder if it is more of the same with tweaks and bells and whistles?