Citizen Smacklefunky: I was just about to post the same.
The hypothesis (AI has foreknowledge) can easily be tested using the method you describe. I have done the same, and come to the same conclusion.

The AI, on less than intelligent settings, doesn't cheat (as in, doing things humanly impossible) in any discernable way.

The evidence is conclusive and easily confirmed, those continuing to quarle about it waste their time.

But there could other explanation: - the Arceans flagsip enters a wormhole and ended unnotice in the player space. Then it manage to escape unnoticed through another hole - the AI had send a colony ship blindly. The main problem is how to exactly know the precise destination set for the ship

Theory 1 is possible - just highly, highly unlikely. Remember, I had this planet continually under surveillence from turn 2. They would've had to trip over the wormhole in their first move, AND still managed to get the flagship out of my bouy's sensor range, all on turn 1, so that I wouldn't see it when I launched my sensor bouy on turn 2.

Theory 2 is also highly unlikely. 80% of the stars on that map had planets, but no colonizable ones. The probability that they just happened to blindly set a colony ship's course for the single star in my region that had the only habitable planet (out of about 10 stars in the region I had my sensor net deplyed over) is pretty low.

Of course, if my supposition is correct, then it should be repeatable. I'll have to try it again tonight - especially so I can get a history of save games like Peace Pheonix suggested. (That didn't even occur to me. Nice idea!)

There's also another theory - at one point, i did have to save my game to run off and do some errands, and then reloaded to continue the experiment. It was after then that the colony ship appeared. If there were a bug in the save game-load game that caused AI's to suddenly know about planets it didn't know before, then that could account for it. Either way, more tests are always helpful.

crickel

About autosurvey, sometimes my scouts turn off the autosurvey mode while there is still anomalies in the map. Sometimes it happens when there is a anomaly right at its side.

I've noticed the AI DOES seem to KNOW where the resource places are as soon as you yourself discover them.

Even though Brad revelaed "the code," it doesn't show the subroutine and def calls he's making in detail- I think there could be a bug in the functions that define things seen by ships- seems like the AI sees things seen by the PLAYER'S ships.

Not that I care too much since the Ai needs help in a lot of other ways:
over valuing the "strength" of its military so you're crushing its navy but it doesn't see that- doesn't get that it's weak fleet adds up to 0 since NONE OF HIPS SHIPS can beat any of yours at ALL...

doesn't plot uses for its ships all the time- since a lot of Ai ships sit around doing nothing for many turns when OBVIOUS uses are out there. I once gave three VERY POWERFUL ships to the Torians who were losing a war right BY their enemy fleet. The ships sat there and sat there and then wandered toward home when they could have shattered the enemy fleet's backbone- destroyed them in fact... sad.

etc...

So, I ran another test!

1. Started a new game.
2. Proceeded to Sensor net my sectors. Found one habitable planet in the vicinity on turn 2, had it under constant surveillence by turn 3.
3. Destroyed all scouts with fast interdictors before they got anywhere close to my home system or the nearby system.
4. A Korx colony ship bee-lined directly to the habitable planet.

Notes: I gimmicked my ships a little so that I could toss out a Sensor Bouy every turn with max military spending. This means they only had a sensor range of 10, but they moved at a rate of 6. They still detected the second planet on turn 2, had it under total observation within standard scout range by turn 3. My interdictors had a single laser and moved at 12. I used both the sensor bouys and the interdictors to expand my sensor net much more quickly and much farther than my first trial.

A backup of debug.err is stored with the save game files.

I saved regularly thoughout the game this time. Going back through the saved games and using the cheats to view the events, I can verify that this planet never came under observation by scouts, wormholed flagships or any other method, and that the colony ship was sent on a direct course.

You can download the saved games, if you like, here: Link

crickel

time to put this to rest!!! (although as I type this I know that people are still going to argue after this proof is given. Here is an EASY test that anyone can perform. a) Run galciv2.exe with the command line paramter "cheat". There are a few ways of doing this, 1) create a file (with notepad if you want) galciv2.bat which contains the line "galciv2.exe cheat" in the same directory as the game is installed 2) start->run->cmd, change to the correct directory and run "galciv2.exe cheat" 3) Other ways I can't be bothered mentioned, you get the idea Start a game. Low habitable planets and lots of stars or whatever you want. c) Press Ctrl-U. (effectively lift fog of war) Now you can see where ALL objects are You will instantly meet all other races in the next turn. But this is because you now see them... (if you are wondering that because we cheated in seeing all activity, then so can they... I disagree due to the results of this experiment seems to show they can't see). d) Sit back and relax and press "Turn" alot, watch what the AI does. 2 things I notice. Firstly, the Flagship B-lines to all anamolies, seen or not. (Which they shouldn't,. no way they should know...) But we can do that also, so its fair. Secondly (and more importantly) 2 or 3 scouts ships explore the universe to all planets, habitable or not! and well before any colony ships. Thirdly, they also keep ships near the other races areas so they can "spy"... sneaky little devils So there... feel free to argue all you want, only those who read this post and try it out can see for real what the enemy is doing. (To the original poster, try your experiments in the same manner as before but with this cheat, and you can see how they managed to deceive you all this time). HAPPY SPACING

THANK YOU.

This issue is not hard to resolve because of what he says.  Just use the cheat mode, unhide fog of war and watch.

You will see the AI send scouts around FIRST.

Another question I'd ask is WHICH player are we talking about here? Different computer players have different AIs.

This issue is not hard to resolve because of what he says. Just use the cheat mode, unhide fog of war and watch. You will see the AI send scouts around FIRST.

Which, in my test games, I shot down. Every single last scout that came within my hugely extended sensors fell before my interdictors. Their sensors had no chance to detect the planets they ended up sending a colony ship to.

I didn't just passively sit back and watch the AI run. I actively shot down their scouts so they could have no information about what lay in those sectors so I could see if it was cheating or not.

Another question I'd ask is WHICH player are we talking about here? Different computer players have different AIs.

I was playing the Terrans, and it was the Dominion of Korx that sent the first scouts I shot down, and then the colony ship directly to the non-scouted planet.

crickel

PAR:

I accept your sentiments, and I'll apologize too. I was overly harsh; I understand that unless one couches English carefully, it can be interpreted to mean unintended things.

You can download the saved games, if you like

Excellent, out of curiosity I will be looking at these as soon as I get home from work... I will report on my findings. (I think there is more fun trying to break/understand this game then there is playing it... Which is a lot, considering it is hella fun to play).

Brad said on IRC tonight that he ran the AI scouting stuff through the debugger; it's working fine. The AI *MUST* scout before sending any colony ships. He still has yet to look into the resource-rushing though, that may be a real problem.

All of these tests are based on the hypothesis that the AI didn't see the planets. To validate that test it is claimed that the player saw ALL of their scouts. It is flawed. Sorry but that's the simple fact, there is just no way you can prove that you saw everything at all times. You can't "see" what their scouts saw. Further to that, yes, the AI can see all planets and send a constructor straight at a planet.... it's got Stellar Cartography. Once you have got Stellar Cartography you can do the same thing too. It's much harder for a human to click directly on a planet on the mini-map, where a computer will send it directly to the right spot.

My obeservations in game (not tests, just observations) tend to break this theory. I have seen countless errors by the AI in settling new colonies. Errors where they have missed higher PQ worlds just a few turns away, errors where I have just settled the planet they were heading to and have then seen them turn round and go back, even though theres another planet just a few parsecs further into the fog of war.

I am afraid I can hardly accept that any of the tests above are conclusive evidence that the AI knows where colonisable planets are. I have no proof against it either, but I am more likely to believe the developer who can run a debugger situation than a player who could have missed any number of factors.

To validate that test it is claimed that the player saw ALL of their scouts. It is flawed.

In order to see the planet in the first trial, the AI would have had to have started out with a scout on turn 1 and flew it past the system, then back outside of my sensor range. This is impossible, since you don't start with scouts.

In order to see the planet in the second trial, the AI would have had to have done the above on turn 2. Since I stored the saved games this time around, you can plainly see by engaging the cheat codes, no scout ship on turn 2 anywhere in the vicinity.

I may not be able to see all things at all times without cheat codes, being a mere player, but I can tell you that no scout came within 8 sectors of that planet over the duration of the game after turn 3. Seeing as their scouts have a sensor range of 2... I'd darn well say that it was impossible for them to know it was there among all the other trash planets.

crickel

Hinriad,

Thanks for the advice! I´ll try it my next game Btw, Now that I found the buy button things go a little bit smoother

But even tho I get I perfect start I always end up way behind in tech, planets and military. Techtrading would explain the advantage in tech but I cannot explain how they are able to get so far ahead of me in military. I am almost always 1st in manufacturing. I doubt they can by 20 heavy fighters so early in the game. I can afford around 5. Espcially as the must be buying numerous Colonization ships.

Makes me wonder if the AI get a "boost" in the start depending on difficulty ala Civ? I don´t mind that but it should be documented somewhere if thats the case.

THANK YOU. This issue is not hard to resolve because of what he says. Just use the cheat mode, unhide fog of war and watch. You will see the AI send scouts around FIRST.

Frogboy, I hate to have to tell you so, but having something that works correctly with cheat codes enabled and FOW lifted doesn't give a 100% guarantee that it works the same way with no cheat enabled.

BTW, have you looked to the savegames posted by crickel ?

And it would be nice if when using the CTRL + SHIFT + Z cheat, the FOW will see corresponds to the one of the AI whose control hae been taken. Same thing for the F1/F2 key. It would really easier things to allow us to see what the AI truly knows.

Anyone else wonder about the title of this?

AI isn't very smart? I have always seen the AI send scouts out first. But let's say he's right and there's some bugaboo that lets colony ships get to planets without exploring.

Does that make the AI not smart? There's a lot more to the game than colonizing. the first GalCiv was considered exceptional for its AI and in that the AI knew where the planets were.

To original poster - Isn't it clearly stated in the manual that all other races except humans start with maps of the universe and humans, because they are new, must yet explore? There is even written "start the game as another race and watch AI humans struggle with exploration". did you try that?

To original poster - Isn't it clearly stated in the manual that all other races except humans start with maps of the universe and humans, because they are new, must yet explore? There is even written "start the game as another race and watch AI humans struggle with exploration". did you try that?

Well, this means that some races starts with stellar cartography, but this tech doesn't give you location of the GOOD planets. It does indicate where are the planets orbitings stars on the minimap. And Frogboy have stated numerous times that the AI must scout first.

So an important question is: Is there currently a flaw in the AI procedure, that would allow him to target good planet with colony ships that haven't been explored yet by the AI.

So I did my own testing as well.

Normal game
Size: Tiny (don't know why people go for large maps.. so many things could go wrong) and ctrl+n till i have a star behind my back that the computer can't reach.
habitable planets: abundant
number of planets: abundant
number of stars: rare
star density: tight cluster
anomalies: rare (don't want their anomaly ship to get into a wormhole)
tech rate: very fast

I used a custom race.
+10% speed (so my ship can catch up to scout and destroy them)
+2% sensor
+30% soldiering (oops.. forgot to get some other bonus instead)

party: technologist
research: +20
sensor: +1

tech: xeno communications, space militarization, galactic warfare, xeno engineering, stellar cartography, xeno research

Lastly, i set the opponent to 1 (terran) on bright. Difficulty challenging


Now here might be a bug.. actually. My home star was called Custom Star.
Home planet: Crescent Moon just as I named it PQ 10
Custom Star II: PQ 0
Custom Star III: PQ 11
Custom Star IV: PQ 0
Custom Star V: PQ 11

two planet with PQ 11 in my home star, joy.

I am in the middle, terran up north, a system up north east and a system down south.

Naturally, I headed my survey ship toward the terran while grabbing one of the PQ 11 planet.

Then I researched for beam weapon and some propulsion tech.

I build myself a ship called watcher, using cargo hull. I didn't think I would have to take down anything other than an scout. I also figured that I could cram in more sensors and such this way. In the end I didn't fill it all the way up for sake of saving money and there is no need a tiny map.
Watcher's stats: beam 1
hp 1
speed 3/pc week
sensor 3pc

With my racial bonus though
the speed is 5
sensor is 6

Meanwhile, my survey ship has COMPLETELY missed its survey ship while going north east.
Luckily, when my watcher goes north west (nothing around there), I found the survey ship and blasted it to smithereen.

With an additional watcher, I headed toward the terran system, blasting any scout that comes my way. Nothing escaped.


NOW is the moment you have been waiting for, the bug.

The AI kept on building scout after scout.

It wasn't till 8 Dec 2226 that the computer decided to build a constructor instead of building scouts that only get blown up by the sad sad sad 1 hp Watcher. The computer set up an influence starbase with it.

Then in 1 Feb 2227 (might be earlier) the computer finally decided to build a scout.


So having watched the AI build scout over and over, I decided to let one on the loose and see if the AI would somehow know the planet being there while being relatively close (cause there might be program to AI sensor).

I destroyed the ship right before it sees the star (lucky me, actually I was just about to let it see it.. but oh well).

The AI wasn't able to see the star, it continued back to its old routine of building scouts.

Come 22 March 2227. The AI finally decided to build a heavy fighter and thus here i am writing my report.




If anyone wants to see my save, i'll post them up. I have about 9 saves.

I just figured that this is much easier to replicate. Once you take down the survey ship, you can pretty much trap the AI in the corner and watch it build scout over and over. Of course, one could say the AI might act diffrently on larger maps.. but I felt this is a better experiment because it is closer to a controlled experiment.


Now, perhaps I should try a harder AI before asking devs to address the 'bug' of AI building scouts over and over only to have it destroyed.

Lastly, i set the opponent to 1 (terran) on bright. Difficulty challenging

Well, try another race, one that the stellar cartography tech

My home star was called Custom Star

Have you changed it when creating your race? Or was it correct the first time and get screwed woith CTRL+N?

Crickel, one question... what difficulty where you playing at?

ok, Hi its me again...

Now, I more than anyone else think this thread should die... that is until now... now it gets interesting

Some may remember earlier I said:

Excellent, out of curiosity I will be looking at these as soon as I get home from work... I will report on my findings.

Well, I kept to my word and spend 7 LONG HOURS trying to get something out of it.

The only thing that kept me going was that there was obviously Something going on. Either the AI indeed did have a bug, or the people who said it did where missing something big.

So, I grabbed the saved games, and came up with A DEFINITIVE ANSWER. Of course by definitve I mean definitive until it is disproven. But the procedure is solid and well re-producible, so it actually may stick to be the final conclusion.

Anyway I won't spoil the ending, here is my report, full of spelling and gramatical errors (and pictures):

Link

I think anyone who has shown an interested in this thread (be it for, against, REALLY REALLY against, mildly interested or accidently clicked on the wrong thread) should read this, and try it out.

SmackleFunky - That was very interesting. I'm disappointed that Brad hasn't found this to also be true...hopefully they will reproduce this in-house (or cover up the scandal somehow, heh) and find a fix so that we can play a fair game.

Hm, nice work, SmackleFunky, BUT:

Your conclusion is only valid under the assumption that the cheat removes the FoW for the human player only.
Since we don't know how the Ctrl-U-cheat is implemented, we can't be sure. It Ctrl-U removes the FoW for all players, human and AI alike, then your test proves nothing.

*sigh* It is simple to prove the FOW doesn't effect the AI since if you play a normal game where the AI has planets to colonize with FOW cheat everything happens as you expect... AND FrogBoy/Brad says when the FOW is lifted it doesn't effect the AI... AND all the tests that proved the AI didn't cheat used the FoW cheat. People used the "FoW changes AI behaviour" to argue the other way around!! you can't use the same excuse for both sides of the argument...

Anyway, the main point of me writing this was so FrogBoy sees where to look, and can confirm or deny my theory, so any other hypotheseses are irrelevent