Any results with the "bug" test?

Well that's certainly not true. There's been many times when I've colonized a planet only to see an AI ship come by a few turns later hoping to grab it for itself. Then it will hang around for a few turns more and finally go somewhere else.

Yeah so? I do too, at least at the start of the game. I used to send my scouts out first until more experienced players pointed out that it makes more sense to send out colony ships directly. They're faster with the same if not greater sensor range, and if they discover a habitable planet they can colonize it right away without having to wait for another ship to come from the home planet.

If you'll notice, the AIs tend to make Stellar Cartography one of their first techs. This allows them to see planets on the mini-map as well as stars. If there's a system with 5 planets around the star, there's a reasonably good chance that at least one of them will be habitable so they'll make a beeline for that one first. I do exactly the same thing. There's no mystery or conspiracy involved, it's just smart tactics.

Belanos, you are just repeating arguments that have been made before in this thread. And which have been rendered irrelevant by the tests made so far. No offense.

It's interesting... many people (myself included) have the distinct feeling that the AIs are cheating heavily when it comes to finding suitable planets and it can be "proven" by several tests. Yet at the same time Kryo makes a test more or less proving the opposite and Brad always vehemently claims that the AI doesn't cheat in this way.
Since I believe what the devs say (they have no reason to lie to us) but also see the other results, I wonder what is going on. Sometimes it seems we aren't playing the same game.

Is it possible that the Thalans get a small kind of bonus cheat at the beginning to make their colony ship worthwhile since they can't colonize a second planet in their home system? So their first colony ship is allowed to go straight to the nearest habitable planet instead?

I'm looking forward to see Kryo's results if he makes some more tests.

Being curious, I did some tests myself, using the "reveal map" cheat after a few (3-6) turns and looking at what the opponent did so far.

The first thing I noticed is that the AI's scouts always have a habitable planet as destination, often way out of sensor range. First I thought that was an indication for the AI cheating, but I found out that the AI immediately recalculates the destations after you reveal the map. How I found out? I saw several scouts which evidently moved in one direction from their homeworld while the map was not yet revealed and after looking at them, their destination was in the opposite direction of their homeworld. So you can't just click on them and think that the line is where they were going to travel to originally.
The same applies to colony ships, of course.

So I started about a dozen games with different opponents and tried to figure out if it was possible for the AI to have its ships where they were without cheating.
Unfortunately, I didn't get a result. The colony ships always were flying straight to the next habitable planet, but there always was a way they could have seen the planet before. I still think they are cheating, though. I'd expect them to at least once fly in the wrong direction. Will test some more later and post if I find anything conclusive.

Hi!

IMO there's an (unwanted) glitch in the code, that alows the AI to see what I've revealed. That would explain my (quite rare though) experience, when I found in a late early game a galactic resource in far "backyard" of one AI, and sent there my constructor to claim it. But it stopped at a half-way out there, since the AI, that was ignoring it for a whole game year claimed it. Quite the same thing I noticed with some far-away planets.

Well, my proof is just an anecdotical one, but too many anectodes about the same issue do stop making it one.

BR, Iztok

Yeah, but I didn't reveal it. Only after a few turns, I revealed the map and looked at where they went so far.
Nonetheless, you may have a point there and that would explain many things. Including why sometimes the AI seems to know exactly where to go and sometimes it doesn't have a clue. And why there are tests which hint at the AI cheating as well as those that hint at the opposite.
The question is: Does the AI only see what *you* reveal or also what the other AIs reveal?

Alright, did some more testing and just saw something interesting: I revealed the map after 4 turns and the Arceans had colonized their second home planet. The next useable planet was about 1-2 sectors away. They had built a second colony ship but it was just orbiting their home planet. So I thought they probably just finished building the colony ship and it will start towards the planet the next turn.
But it didn't. Only another turn later, after their flagship got into range of the planet, they started to move their colony ship.
At first glance it looks like they waited until they could see the planet. Which would be fine. But from what we gathered so far, they should have seen it the moment I revealed the map. Yet they only moved two turns later. I conclude that it's just as weird as I thought.

If that's true it explains a lot. And I didn't want to get in an argument with the devs just because the tests I run seem to come up with different results than them. I really don't care if the AI "cheats" in this way, but I was able to isolate the behavior.

Perhaps the problem with my test was those dozen sensor range 15 cargo hulls. They revealed the whole map to me, so thanks to the bug revealed the whole map to my enemy?

I wonder if that bug shows up in their debugger. Perhaps that explains the discrepency between mine and Kryo's test results. I absolutely believe Kryo... at the same time, I know what my test games showed.

All in all, not a big deal to me. But some people seem to care.

I did not use a debugger, but the nature of the test I did would mean any such bug would not have any effect. More conclusive testing could only be done by the devs.

I think it's mostly a matter of "honesty" if you will. It's only really an issue for the AI when it is playing the Humans, since they don't start with Stellar Cartography - and even that is outdated and nonsensical, since even the Thalans, who have been in the galaxy for only a year or so, have it, and heck, even all the other non-original races *and* the Akilians have it too...

Sorry, double post

I think there might be a way to be a bit safe from a colony rush, what i did for one of my games is that instead of getting my scouts or colony ships to explore all over, after i colonized like 6 or 7 planets, then i stopped all contacts with anyone. i kept to myself, while the AI players war it out amongst themselves. otherwise, this might be a good strategy because this way, The has less of an opponent to pick a fight with. let me know if such similar strategy works with anyone.

Also, is not only colony rushing that the AI is very good at, it is like their tech development at normal level goes by pretty fast. I do notice that when researching techs, after research certain techs, it shorten the time for all your other techs, but then after develping doan another line of techs, the rest of the other techs would take a bit longer. I think the AI would know ahead of tiemn which techs to research in sequence so that it will be able to keep at an technological advantage. WQell at least for one of the leading race, whichever one that will be during a game.

We aren't really talking about colony rush strategies, but what the AI does during a colony rush. Is it behaving as intended? Whether it's behavior is fair or not, makes them game harder or easier, etc. is really a secondary point. If the AI is working as intended (i.e. not "cheating") or whether it is somehow getting sensor information it shouldn't have.

The ai in this game rocks they really think

I don't generally read through the whole thread before I make a comment, I take a more spontaneous approach. If something comes along while I'm reading that interests me I'll mention something. I realized after reading all the posts that my statements were redundant and for that I apologize. It's just the way I read and respond to forums.

No problem, Belanos.

So, we still don't know whether the AIs "cheat" regarding their colonization... still looks like it to me, though. The problem I see here is that the initial colony rush is extremely crucial. And it's one thing to lose out because of bad luck or because you made a wrong decision but it's completely different when you don't even have a chance due to the cheating AI.
I stopped playing GalCiv 1 precisely because of this AI cheating. It's not as bad in GalCiv 2 (or rather, the consequences aren't as bad if you get a few planets less) but I'd still like it to be fixed. Can't you devs devote a few hours to this and check what exactly is happening?

Would it possible for you to do the test with Thalans? (The same test you did with Korath) And if it does not show abnormal behaviour then could you do the test again but without demolishing other ships? The test should not take long and may give some answers...

I think it is worth testing. If there really is a "bug" then it is a big one. Initial colony rush stage has a huge impact on the game. As someone (I believe Wyndstar) pointed out in another thread: games are won or lost in the first year.

I've got a chance to do additional testing. Here's what I found:

The behavior seen by the Thalans seems to be a very small and simple oversight, applying only to them. The reason for this appears to be that the AI will automatically use its first colony ship on the nearest colonizable planet (expected to be the other in their home system). But the Thalans don't have a second usable planet in their home system as all the other races do, so this results in their first colony ship potentially beelining for a planet halfway across the galaxy.

Assumably this behavior will also apply to any modded races using single-habitable home systems. It does not apply to any colony ships beyond that first, and all behavior afterwards appears completely normal. Scouts are built and tasked to planets seen on the AI's minimap (I have seen it task scouts to all varieties of planets, be they usable, special environments not yet usable, and class 0 planets all the same) without any discrimination indicating prior knolwedge of planet class/type. Subsequent colony ships, or even cheat-clones of the starting ship, are parked until the scouts find additional suitable locations to colonize.

One other quirk I did notice is that the Terran AI seems to scout identically to the others (tasking scouts on the exact locations of planets) despite not having Stellar Cartography.

This thread is more than a year old but I think it is better to “reincarnate” it rather than to start a new thread. This is a long post: it seems better to err on the side of details then risk misunderstandings.

I was satisfied with kryo's answer for long time. It explained nicely Thalan uncanny ability to find habitable planets. However, AIs build much less scouts in ToA (probably because small hulls are not immediately available). I have started a number of ToA games by now (none of them with Thalans) and AIs were just incredibly lucky with their colony ships in each of them. It became harder and harder to believe that AIs do not have prior knowledge of habitable planets.

I decided to run a little test. The test had the following parameters. Size: medium, scenario: normal, all victory conditions checked except alliance, all game options checked except blind explorations (note that according to Stardock AI explores blind regardless of whether this option is checked), Habitable planets: rare, Number of planets: abundant; number of stars: abundant; star density: tight clusters, anomalies: rare, number of asteroids: rare, tech-rate: very fast; number of minors: 0, my race: Altaria, AI opponents: Terrans, Drengin, Korath, Arcean, Torian; difficulty: tough; all AIs set to max CPU;

I started close to one corner. There were several star systems approximately behind me in the corner. All AIs were relatively far away. I scouted my corner and discovered two planets habitable without a tech and two anomalies. Mindful that AI may see what human player can see (a suspected glitch) I reloaded the game. (The save I made the first turn before making any moves). This time I was careful not to reveal the habitable planets. I "harvested" the anomalies to avoid drawing AI survey ships to the area. Then I set up a perimeter so that I could see any ships moving to "my" corner. Since I took max sensors as my racial ability and used space miner and colony ship for scouting I was able to set the basic perimeter up in three turns. Then I built some warships which I used later to destroy "invading" AI scouts. I colonized Wisp (the other planet in my home system). Then I waited. AIs colonized all habitable planets near my perimeter. Not very long after that Terran and Torian AIs sent a colony ship almost simultaneously towards "my" corner. I did not interfere. Both ships went straight for the habitable planets and colonized them. I reloaded the game (the save I made when the first AI colony ship made the appearance). This time I colonized the two planets before the AI got near. AIs made one full move towards the planets but then turned sharply and left the perimeter.

Conclusion. The only way I can explain AI behavior is that AIs know the location of habitable planets without scouting. Do not get me wrong: I am not accusing anyone of lying or whatever. I do believe that the AI code is designed to scout. However, my test leads me to believe that I have discovered a major bug in the code. The implications of AI knowing the exact location of habitable planets is profound: inter alia it results in much more intensive and luck based colony rush for human player and decreases the diversity of the game. I very much hope that Stardock can fix it unless AIs prove to be hopeless without it.

P.S. I have the two saves available if anyone is interested. (The one at the start of the game and the one when the first AI colony ship appears)

P.P.S. There is a very slim chance that an AI's survey ship jumped to my corner through a wormhole and left before I could set up the perimeter. However, the chances that two AIs survey ships (Torian and Terran) managed the same is virtually impossible. This can be further tested by destroying all Terran and Torian colony ships and observing whether other AIs would send theirs.