Every so often someone, usually someone who just got beaten, will start screaming that the game "cheated!".  It's a lot like the neighbor kid who screams "cheater" every time he loses at a game.

One of the reasons people like Galactic Civilizations II's AI is because it doesn't cheat.  But that generalization is only true to a point -- it doesn't cheat until AFTER "intelligent" (aka "Tough" in the overall difficulty slider).  Few players can beat it at that difficulty level and that's where meticulous care was put in to make sure that it's playing exactly the same game as the player.  Beyond tough, all bets are off.  At that point, we're talking about players who have either found an exploit or are simply incredibly good at strategy games and our goal then is to make sure they have a challenge. 

At Intelligent, it's playing its best game, same rules, same money.  At genius, it gets a 125% modifier to its bonuses (economically mainly and ability bonuses).  In practice, it just makes them a little bit spunkier as no AI can be as efficient as a human in a crunch so this is the level I recommend for people who are really good at the game who can beat intelligent.  And the last one is called "Incredible" where we're doing everything we can to beat the player.  200% modifier to bonuses, the intuition algorithms are set to be very lucky, etc. Basically, the player's a god at this point and we're doing all we can do beat the player, no holds barred.

The difficulty levels (from the manual) are:

1. Fool.
2. Beginner
3. Sub-Normal
4. Normal
5. Bright
6. Intelligent (this is where the AI is playing the same game you are with no bonuses or penalties with its best algorithms)
7. Genius
8. Incredible

In most games, the AI begins cheating right away or at such a low level that even casual strategy gamers have to raise the level well beyond anything that is remotely "fair" in order to be challenged.  In Galactic Civilizations II, it's not until near the end that the game is tilted against the player.  Again, this is in the user manual so this isn't some state secret. 

In the development journals, we talked about the goal of having the "intelligent" AI being able to beat 95% of players.  I think we succeeded.  In v1.1, our goal is to maintain that 95% (as people get better at the game some will eventually be able to beat the "intelligent" AI so we have to evolve with them).

So where do people scream foul and why?

1. Claim: "The AI knows where the good planets are".  This is largely a carry over from Galactic Civilizations 1 where they did know where the good planets are. As any experienced GalCiv 1 player can tell you, that advantage was really pretty minor in the bigger scheme of things.  A good player can colonize as fast as the AI in GalCiv 1.  In GalCiv II, it's even easier because if you have Stellar Cartography you can see the planets on the mini-map. 
   
   However, we knew that many people used the GalCiv 1 AI advantage to say "the AI isn't totally uncheating" so a lot of work was put into having the AI design, build, and scout out planets. And I mean a lot of work.  To a user, scouting is easy, they "look" at the mini map. But having the AI do it intelligently took some work.  Even things like how many scouts to build, when should it build them, where should it send them, involved a lot of work to do it well because every move at the start of the game is crucial.
   
   In 1.0X of the game, we even put in additional checks to make sure that this was the case since there were cases where the bIsKnown flag could be uninitialized and have some junk in it (and as any C/C++ coder can tell you, if you're doing a if(bIsKnown) and there's garbage in there it'll return true.
   
   But this scenario has been played out over and over again, tested, verified by us, other players, etc.  Anyone with the demo can test it for themselves too. It's very easy to verify that the AI does not know where the good planets are.  Now, at "Incredible", it's hard to say.  The ability bonuses and some of the simulated intuition stuff gets murky when you're multiplying everything by 200.  But that's not material to our discussion since the AI at Genius and Incredible is getting advantages over the player and has always been stated as such.
   
   For instance, at 200 bonus, the sensors are probably so powerful that they have no real fog of war. 
    
2. Claim: "The AI builds too fast."  This part almost always rises out of, sorry, "newbie" players who don't realize you can quick buy things. Decent human players do this, and you can bet the AI does this when it can.  The more money it has, the more often it can quick buy things.  On the planet screen, that's what the "Buy" button means.  A player starts out with 5,000bc.  You can literally build something every turn for dozens of turns without any special advantage.  At the highest levels, where the AI gets money, it can go even longer. 
   
   This particular claim I find especially annoying because it sounds a lot like the neighbor kid screaming "cheater" because they don't know how to play the game effectively yet.  It would be much better for the person to simply ask on a forum and then someone could tell them how it's being done (or read the manual).
    
3. Claim: "The AI trades with each other too much."  This is another one of those "I'm losing so the game must be cheating" type things.  In 1.0 to 1.0X the AI literally does a ulPlayerIDToTradeWith = rand()%(ulNumberOfActivePlayers);.  IF it's the human player, it pops up a dialog and offers to trade X for Y. ELSE it just simulates the same thing with the other player deciding whether they want to make the trade or not.  In 1.1, it still does much the same thing but I made it a lot less likely for the military techs to get traded if they're expensive since it was very annoying having each player having the same military techs. 
   
   The point being, the AI isn't making a distinction between the human player and another computer player.

Now this isn't to say there isn't some bug or something.  But I haven't seen it and I haven't seen one other than the bIsKnown initialization issue and even that would likely only show up on a saved game and was fixed weeks ago in 1.0X (and I was never able to reproduce it, I just saw that it wasn't being properly initialized and that the variable wasn't being loaded correctly from saved games).  So it's possible that there is some bug. 

For example, I haven't looked into this because I didn't write the code but the AI may know where galactic resources are.  I say that because during the beta, galactic resources showed up on the mini-map.  IF that's the case, we'll likely have galactic resources show up on the mini-map again (assuming they don't already) because there really shouldn't be a "Race" for them, they should just be known IMO.  But as a practical advantage, it's pretty minor. By the time you've found your planets, you've probably also found the galactic resources.

Consider some of the things the AI has to do, in order to play by the rules and still be effective:

• It has to design its own ships. You think you have trouble deciding whether to put an extra sensor or life support unit or weapon on a ship? Try being a computer player trying to fit every last thing on the ship without cheating. Think of all the possible combinations of fitting stuff onto a ship to use up every last bit of space.  Think about what weapons you would put on and what defenses you would put on.
• Similar to the first point, having to figure out which weapons and defenses to focus on. Not only does it have to focus on them and research them but it has to figure out when to research them and do so without being scripted.  It has to know how much offense and defense to put on.
• It has to know whether it should be focusing on range or speed based on different galaxy sizes, different numbers of players, etc.
• It has to know what to build on a planet and when.  Should it use that special food tile for a farm or put a factory on it anyway because it needs the production more than it needs the money?
• It has to know when to know when it should be building colony ships as opposed to freighters as opposed to constructors.  In Civilization IV, my favorite TBS, the player might build a dozen or so "workers" throughout the game. Workers don't get consumed when they're used, they stick around the whole game.  But in GalCiv, you have constructors which are consumed and you have freighters, which are consumed, and you have transports, which are consumed which means a LOT more decisions have to be made on what to build on a planet.  And some of these units are expensive.
• It has to do a lot of work on the diplomacy, there's a lot of factors to take into consideration.  Because the game has no political borders which in other games prevent players from doing a sneak attack, GalCiv's AI has to be on the look out for all kinds of threat patterns and do so intelligently.  Plus, you don't want players going to war with you one turn and then asking for peace the next. There has to be a good, logical, intuitive reason why it goes to war.
• The game has fleets AND those fleets are bound by logistics. In other games you either have armies/fleets or you don't.  If you have them, there's no practical limit to how much stuff you can cram in them so they can just go around wreaking havoc.  In the games that don't there's no effort in having to put together such an army.  But in GalCiv II, the AI has to decide how big of a fleet to make, when it should be used, what ships to put in it.  Plus it has to research techs to increase those logistics abilities. When should it research them? Why?
• The tactical battle AI could take pages to explain.  In most space games, they don't have an open surface so there's plenty of choke points and a very limited # of places a fleet can go.  In most land based games, you have mountains and oceans which create choke points as well.  But in GalCiv II, it's open space which means that it's very very easy for a human player to outflank or use mobility to win the day.  The AI has to deal with that kind of thing. It has to know when to flee or avoid a target, to circumvent an opponent to go to a more valued asset.
• The AI has to contend with special invasion ships. In other games, any unit can invade. In other games, one can build a transport and send it over and not worry about having to escort it.  In GalCiv, the AI has to decide when and how to escort a ship.  Escort too much and you lose the initiative. Escort too little and you lose billions of troops in space.

And it goes on and on and on.  So when you look at the claims people have made above, they're such small potatoes compared to what is involved in writing a computer AI that is playing by the same rules. 

If I were going to have the AI cheat, I'd do things like give it more space on its ships. That would be a wonderful advantage.  Or I would teleport ships outside the fog of war so that it would be easier to coordinate attacks.  Or I'd bump up its logistics ability so it could fit just a bit more in the fleet.  Or maybe I'd sprinkle its worlds with just a few more special tiles.  But I don't.  Because the AI doesn't cheat.  But it would be so much easier if it did.  And so you can probably see why I get so annoyed when some new user or hack reviewer tries to build their own credentials by making cheating allegations.  Because I could have saved a ton of work if I had had the AI cheat.

And claims like "The AI colony ships knows where good planets are" are so annoying because first, it's not true and secondly, even if it was true, it's such a trivial advantage compared to the real tough stuff that players take for granted.  As most veteran GalCiv II players can tell you, they typically out-colonize the computer AI because human intuition is better than simulated intuition anyway and that would still be true even if it knew where the best worlds were.

But let's say that the game cheats. Let's say it cheats like crazy.  That would put it on par with pretty much every other strategy game out there.  Hearts of Iron 2 is a really good strategy game isn't it? Very popular in Germany?  Yet its AI isn't even remotely playing the same game as the human player. Does that make it a bad game? No. 

The reason Galactic Civilizations II's AI doesn't cheat isn't because it makes it more fun or it's a good marketing tool, it doesn't cheat because for me, as a developer, the whole point of the exercise of writing a strategy game is to write the AI.  That's why our graphics usually suck in our games because I tend to only care about the computer AI.  

The point being that there's no real business justification for me to be coding on the game other than I love coding computer AI.  It's what I love to do.  I love hearing about other people's strategies. I love making tweaks and enhancements to the AI.  I love adding in new features.  But I also get very annoyed when some user or "reviewer" makes "it cheats" claims without even talking to us.  Odds are, they either misunderstood something.  But let's say they found a bug (which isn't likely but for the sake of argument) unlike the AI, we're only human. Who knows, I could make a mistake.  The AI doesn't escort its ships as well as I try to code it, it's possible that I missed something.