Game-Level AI Advantages

[mkorin]

As a CIO, I question anyone who says they've gone through every line of code of a multi-million line program. If any of Firaxisis's programmers want to weigh in, I'll believe I've read the code arguement. Until then, I'll postulate a challenge for anyone who wishes to take the time. Start a duel, go to war. Open worldbuilder and put down two of the exact same armies on the same terrain. Check the % for your attack, log it, let the AI attack you and check the log for it's percentage. I've always maintained that the %s are correct and posted a pretty convincing statistical proof when IV came out. if these %s are the same, I'll be happy to admit I'm wrong.

Mike

[mkorin]

OK, that's a tougher test than I thought. You can't get the AI to attack you when it is fairly even odds. However, I did finally get a knight to attack my knight (no terrain bonus) The AI knight had Flanking 1 & 2 so I gave it to my knight. My attack % was 27.2. The AI's when it attacked me was also 27.20, but it had an extra +20% bonus in the combat log. Unless someone can explain that 20% bonus, I maintain that it was an AI bonus.

Yes, I know 1 test is not conclusive, but it is a start.

Mike

[ColdPhoenix]

Originally posted by Garth Vader
For future Civs they need combat stats. All you'd need for this discussion is a difference number. So if you win at 50% it would add .5 to your difference. If you lose it would subtract .5. If you win at 90% it would add 0.1, and so on. Take that difference number and divide by number of combats to get a combat number for a game. Hell, you could keep track of that for every game and keep the grand total in the options screen. Maybe that would finally put all these discussions to rest.

It would be interesting to have but then people would complain that the stats weren't working properly either.

I don't think these discussion will ever go away.

[Blaupanzer]

Evaluated the recommended file. It distinctly provides for advantages related to barbarian animals and soldiers. For the player, the advantages go down as the level goes up. For the AI, these advantages go up as the game level rises.

At the higher levels (monarch up) the AI also gets free units, lower maintenance costs (distance and # of cities), substantially lower upgrade costs and other things. Nothing is explicitly noted relating to AI vs player combat.

As to mkorin's example, note that certain promotions only apply to the defense or attack. Thus, if both had charge or city raider only the attacker would gain it. Also note that some "bonuses" to unit 1 only show up in degradation of unit 2's base number. If the attack %was 27.2 in both cases, then obviously both units got the same bonuses in the attack/defense.

Most designers, developers, and coders do pretty rigorous testing to assure that the game does what they say it does in actual performance. The AI may be programmed not to attack below a certain attack percentage, but, if it did, that percentage would hold true over the 10,000 mythical repetitions of those exact conditions. If the AI gains combat advantages at the higher levels, regular players at that level would see it, and this has not been reported. Also, the coders, et al. would tell everyone how they created such a "great improvement" -- not try and "sneak" it into the game.

Paranoia is a terrible burden. Firaxis is not out to "get" you.

[Kuciwalker]

Originally posted by mkorin
As a CIO, I question anyone who says they've gone through every line of code of a multi-million line program. If any of Firaxisis's programmers want to weigh in, I'll believe I've read the code arguement. Until then, I'll postulate a challenge for anyone who wishes to take the time. Start a duel, go to war. Open worldbuilder and put down two of the exact same armies on the same terrain. Check the % for your attack, log it, let the AI attack you and check the log for it's percentage. I've always maintained that the %s are correct and posted a pretty convincing statistical proof when IV came out. if these %s are the same, I'll be happy to admit I'm wrong.

Mike

PEOPLE HAVE DONE THAT.

The answer was - surprise! - the AI does not get an advantage.

Moreover, it's NOT HARD to go through the lines of code
that deal with combat, and see that the AI gets NO BONUS.

Idiot.

[Blaupanzer]

"Idiot" is a harsh response to paranoia. Lighten up.

[Jaybe]

Kuci, could you yell that a little louder please?

While the "AI combat advantage" paranoia has been gone over several times throughout Civ4 history, and we DO get weary of it after a while, ...
it is probably good that the issue is periodically reviewed, particularly for the benefit of newbs. Periodically verified, even.

[rah]

As long as we mentioned the odds posted, is a true thing that first strike and such is not used in calcing the displayed odds. If yes, is anything else not considered?

[Kuciwalker]

Originally posted by Blaupanzer
"Idiot" is a harsh response to paranoia. Lighten up.

This topic comes up all the time, and is constantly disproved.

[rah]

But if a different person brings it up, they don't know about "all the time".

So answer my question instead of ranting

[Solver]

Look, Kuci, I know you have suffered severe moral trauma in the course of your life, but will you ever learn to behave, for Hoffman's sake?

[Kuciwalker]

Originally posted by rah
But if a different person brings it up, they don't know about "all the time".

So answer my question instead of ranting

I don't know. I remember the question coming up in the past, but I think the issue was fixed a long time ago.

[Kuciwalker]

Originally posted by Solver
Look, Kuci, I know you have suffered severe moral trauma in the course of your life, but will you ever learn to behave, for Hoffman's sake?

Who

[Solver]

Hoffman is Hoffman, is the youth so uneducated these days?

[Kuciwalker]

Winners don't do drugs