Please, fix the combat odds!

[parachute4u]

I simply quick save. I don't reload lost 80% battles. I don't reload sea battles, cause I know I will always lose them. But If I lose a critical >92% battle (spear vs horse, CR infantry vs unpromoted longbow), I reload. I can live with losing >92% battles on noble/prince. On monarch and above I do not think I really need to fight vs big AI-advantage and bad luck at the same time.

I mean, after suiciding emergency-upgraded cannons into a stack, railroading a couple of tiles to be able to attack with the anti-whatever unit, softening the defense of the strongest defender with suiciding unpromoted units, and finally getting odds of >92%, I am NOT supposed to lose. If I lose these battles, it is pointless to have any kind of attack plan. If it is a game of luck, I should simply be building warriors in my first city and hope for the 10% of taking the enemy capital.

I hated the RNG in Civ3. CIV has been better until Warlords. But I still hate the randomness of anything random in a game that is a STRATEGY game. It's a long term strategy game. There is no room for short-term luck, IMO.

 

[alpha wolf 64]

I lost 2 100% battles in the same war. I dont remember how I had the infantry promoted but a stupid pike took out one then a few turns later took out a second. Finally took him out with a under 50% maceman.

 

[Hey Joni]

one thing i've noticed / suspected is that the way the random number generator works in warlords, if it's decided that your 96.7% odds with a maceman are going to result in failure, it'll result in failure for all of your guys. in other words, if it has decided that you're in for a run of bad luck, you're gonna have a serious run of bad luck. it's gotten to the point where, if if i see a couple units die on high odds, i wait for the next turn and try again.

Exactly! This means that combat results are not random at all, they are more like randomly predefined on each turn. The same goes for the exploit I described - in the begining of the turn it is randomly defined that all goody huts will give techs. So if you have 10 units next to 10 different goody huts, you are a lucky guy.

This is not the way it should work! Sure, you wait for the next turn, when things are back to normal but why slow your progress just because something is not made right?

 

[MrCynical]

I mean, after suiciding emergency-upgraded cannons into a stack, railroading a couple of tiles to be able to attack with the anti-whatever unit, softening the defense of the strongest defender with suiciding unpromoted units, and finally getting odds of >92%, I am NOT supposed to lose.

Yes you are. The law of averages states you should lose up to 8 times out of 100, which given the number of combats in a Civ game is going to happen on plenty of occasions in a single game. Reloading is just cheating, plain and simple, and what you did to get a combat with 92% odds of winning is of no relevance to how likely you are to win a 92% combat.

Exactly! This means that combat results are not random at all, they are more like randomly predefined on each turn. The same goes for the exploit I described - in the begining of the turn it is randomly defined that all goody huts will give techs. So if you have 10 units next to 10 different goody huts, you are a lucky guy.

Please actually try this before proclaiming you've found some great flaw. It's trivially easy to demonstrate in worldbuilder that not all goody huts give the same result on the same turn, or even anything remotely resembling it. Similarly the combat odds as stated are very reliable if you actually do a proper test. 90%+ odds of success does not equal certainty, and a single Civ game can contain thousands of combats.

The same complaints have been leveled at the RNG in every generation of Civ. They're caused by a combination of people expecting that high odds means they will always win, and selective memory. People never remember the times when they've won at tiny odds. This is partly because a human very rarely attacks at very low odds of success, whereas the AI is much less choosy, creating the illusion that the AI has better luck. In Civ 4 though, take a minute to think about suicide catapults. Ever had one actually win? It happens more than you might think, and I can think of a number of occasions where I've had catapults win at less than 0.3% odds.

 

[ainwood]

How do you know what the odds were that the your macemen (defenders) would win against the horse archers (the attackers)?

A 96% chance of you winning when you attack does not equate to a 4% chance of them winning when they attack. The attack & defense bonuses are calcualted separately - for example, if you have a city raider promotion, it does nothing for you on defense.

 

[DemonMaster]

Okey I write this without knowning how the AI calculate this odds, but cannot be so that the AI doesn't take "First Strike" in the calculation, that would explain at some of the situation because 92% win only stands correct if their is no "First Strike", but a Longbowman can get lucky with his 1+ "First Strike" and then your super unit get injured badly before he could fire back.

Just a thought.

 

[Hey Joni]

Please actually try this before proclaiming you've found some great flaw. It's trivially easy to demonstrate in worldbuilder that not all goody huts give the same result on the same turn, or even anything remotely resembling it. Similarly the combat odds as stated are very reliable if you actually do a proper test. 90%+ odds of success does not equal certainty, and a single Civ game can contain thousands of combats.

The same complaints have been leveled at the RNG in every generation of Civ. They're caused by a combination of people expecting that high odds means they will always win, and selective memory. People never remember the times when they've won at tiny odds. This is partly because a human very rarely attacks at very low odds of success, whereas the AI is much less choosy, creating the illusion that the AI has better luck. In Civ 4 though, take a minute to think about suicide catapults. Ever had one actually win? It happens more than you might think, and I can think of a number of occasions where I've had catapults win at less than 0.3% odds.

Firstly, I admit that I have not experimented with the world builder, but I am sure that there was never an exception to this rule during actual gameplay. EVERY time I used this exploit with two or three units I got the respective number of techs if entering goody huts on the same turn! Also, it is possible that the exploit is fixed, as I have not tried it for a long time...

As for your second point - it proves nothing! Winning often at tiny odds should not compensate you for losing often at great odds! Though I agree that I win on 0.3% a lot more often than I should, this doesn't mean that the odds are displayed correctly. In fact, this is just another illustration of my point - the odds are not the measure for victory.

You can say that I have selective memory for the long-term. However, the example I described is in the time span of a few turns, I had no time to "forget" anything...

 

[MrCynical]

As for your second point - it proves nothing! Winning often at tiny odds should not compensate you for losing often at great odds! Though I agree that I win on 0.3% a lot more often than I should, this doesn't mean that the odds are displayed correctly. In fact, this is just another illustration of my point - the odds are not the measure for victory.

This isn't the point I'm making in the post above, so it's not remotely surprising it proves nothing. You claim I'm saying that you get extremes of luck at both ends, but this bears no resemblance to what I've said. The oddds presented are reliable, whether they are at 99%, 0.3% or 50%. On each occasion the combat odds have been tested, they have been found to be very reliable. This is data, and of value. You are presenting anecdotes, which are not.

You can say that I have selective memory for the long-term. However, the example I described is in the time span of a few turns, I had no time to "forget" anything...

The combat results of a few turns in one game by one player are of no real relevance. It would be surprising if such extremes of probability did not occur given the volumes of games played.

 

[Hey Joni]

How do you know what the odds were that the your macemen (defenders) would win against the horse archers (the attackers)?

A 96% chance of you winning when you attack does not equate to a 4% chance of them winning when they attack. The attack & defense bonuses are calcualted separately - for example, if you have a city raider promotion, it does nothing for you on defense.

I understand this. It's in the first post. The maceman had combat1 instead of CR, I tend to promote a few like that in every army. The spearman had combat2, both units were fully healed. Horse archers, on the other hand had combat1, but a penalty to city attack. I guess that their chances of winning were higher than 10%, I should try this with the world builder when I get home. It was much worse when the one remaining HA on 0/6 strenght defeated the full strenght chariot (combat1, this is all actually repeating my first post ) in the city! So the computer had great luck a few turns ago, when defending from my macemen with collaterally damaged archers and an axeman and, once again, when attacking my city with horse archers, again at low odds... This must be the poorest campaign in history

EDIT: Oh yeah, the units defending the city were fortified for a while. There were no bonuses from terrain for defense.

 

[Hey Joni]

On each occasion the combat odds have been tested, they have been found to be very reliable. This is data, and of value. You are presenting anecdotes, which are not.

The combat results of a few turns in one game by one player are of no real relevance. It would be surprising if such extremes of probability did not occur given the volumes of games played.

The "anecdote" I presented was to illustrate the problem. I know what data is, I took two courses in statistics at university, I took a psychology course too (though statistics were not my favourite ). It has happened to me much more often, I can tell you at least a dosen similar stories. Most of them are related to losing at great odds in the same turn (I don't mind even that, cause I'm used to it). What makes this one different, is that it happened twice, three times even, in only a few turns (after all the struggle to get this enormous tech lead on Monarch) and annoyed me more than usual.

The combat results from one game by one player are all that should matter. I described just a few turns, but I had this in my first war, and I am pretty sure these were not the last annoying combat results I saw in that game. I'm sorry I did not prepare a survey, I'll try to do that the next time I complain.

BTW, you seem pretty confident, have you by chance participated in the testing? I'm not trying to insult you, or something, just asking.

 

[Koelle]

My warrior killed a Gunship once. It was a war that 10 Civs involved, between me and 3 vassals vs. Alex & Co. so i didn't know the status of Alex's Gunship before the attack, it just came out of nowhere and attacked my almost undefended city. But even if the Gunship was badly wounded before, it's hard to imagine how a Stone Age unit could kill a Gunship. Throw stone at it?

 

[parachute4u]

The same complaints have been leveled at the RNG in every generation of Civ. They're caused by a combination of people expecting that high odds means they will always win, and selective memory. People never remember the times when they've won at tiny odds. This is partly because a human very rarely attacks at very low odds of success, whereas the AI is much less choosy, creating the illusion that the AI has better luck. In Civ 4 though, take a minute to think about suicide catapults. Ever had one actually win? It happens more than you might think, and I can think of a number of occasions where I've had catapults win at less than 0.3% odds.

If my suicide cat wins at 3% I hate this just as much as losing at 97%. If I take a city winning a 30% and several 70% fights I feel bad, cause I don't like to win the game because of luck. -- That's why I reload. I don't like the luck element in this particular game.

 

[Veritass]

Make a little grid on a piece of paper, three columns by 20 rows (Excel is good for creating such a thing). Number the rows 0-5%, 6-10%, ....96-100%. Use tally marks to tally wins in column two and losses in column three every time you attack. It shouldn't take more than a few hours of battles to get an idea of the "real" odds.

Note: I have been tempted to do this just out of curiosity, but I have never taken the time. I would be interested if anyone else has done something similar.

 

[Mr. Do]

As for your second point - it proves nothing! Winning often at tiny odds should not compensate you for losing often at great odds! Though I agree that I win on 0.3% a lot more often than I should, this doesn't mean that the odds are displayed correctly. In fact, this is just another illustration of my point - the odds are not the measure for victory.

This is an excellent point- too often when people comment that they think there might be a problem with the RNG or odds we're told it's due to selective remembering or being sore losers- but the simple fact is that the complaint lies that far too many low odds battles are won; we don't care whether it's us winning or the AI winning, it's that the odds are so contrary that concerns us! I once posted a thread telling of how I had fooled the RNG into giving me easy -wins- when I should lose by simply reloading and instantly enacting the attack, but still I had people telling me I was complaining about the RNG because I was a bad loser/ player/ person

MrCynical's post was, unsurprinsigly, cynical, but it made me think of one thing: thanks to the Great General experience meter you get in Warlords, you can now keep a rough track of the number of battles you engage in during a game. I can safely say that in my last game, I did not exceed 300 experience points, which means I had AT MOST 300 victorious battles (And more likely 200-300; this excludes Barbarians, which is a shame because I've had some shocking results, including my first 100% loss a few days ago when a Barbarian attacked a heavily fortified unit in a city and won). In that game I lost at least ten 90%+ battles. I also won about three 0.3% battles- and I certainly didn't engage in the 500 battles at those odds that would need to take place before I win that many low-odds fights!

It's entirely possible that the curse of selective memory is affecting me (I am getting old now, at 24 years old). So I shall take Veritass' advice, and next time I play, I shall record all high-end losses and extremely unlikely wins. While I'm not going to go the extent of recording every battle, with the GG experience bar, I'll know how many fights I had, and if the victories and losses experienced tally even at the most generous of assumptions.

 

[BlackJAC]

Unless the odds are 80% or more I expect to lose fights. Wouldnt surprise me if they are just "theoretical" odds, ans the real odds are boosted in favor of the computer player at higher levels.

It just happens too much to believe its random chance.

Of course it's not down to random chance. The fact the number is generated from a computer will always mean it's a pre determined number

 

[floppymoose]

FYI, it is possible to have a broken RNG. I used to play angband, which was open source. It had (still has as far as I know) a broken RNG that would cause statistically unlikely events to clump into runs. So you would be fighting something that pretty much can't hit you, and then it would hit you.... ten times in a row. At first I thought it was my "selective memory", but I played a lot so I had a pretty big sample size of experence. Then I looked at the rng in the code and realized it was crap. I replaced it with a real one and rebuilt it, and wala, problem solved.

I'm not saying that's the case with Civ4, but I wanted to point out that it has happened before.

 

[Mr. Do]

Well, the experiment is on. Currently I'm on at most 17 victorious battles. However in the last two turns I have lost a War Elephant and a CR2 Swordsman to one Archer in one city, at odds of 87% and 85.5% respectively. Not sounding good so far...

 

[magfo]

I'm not really sure why Firaxis keeps a huge chance-based model for their battles. The only reason given by proponents is that it gives an outplayed and outmanuevered civ a chance to win by getting lucky. I hate gameplay mechanics that are designed to give losing sides a chance to win due to luck. Something like a pure hitpoint based mechanic found in RTS games would be better, imo.

Yea, make it like real life. In real life a guy with a new gun ALWAYS wins over a guy with an old gun with 2 HP left.

*sigh*

Get real.


Nothing can be taken for granted. NOTHING. Real life IS chance. And you want the game not to be? Sheesh. BOOOOOOooooring.

 

[magfo]

I simply quick save. I don't reload lost 80% battles. I don't reload sea battles, cause I know I will always lose them. But If I lose a critical >92% battle (spear vs horse, CR infantry vs unpromoted longbow), I reload. I can live with losing >92% battles on noble/prince. On monarch and above I do not think I really need to fight vs big AI-advantage and bad luck at the same time.

I mean, after suiciding emergency-upgraded cannons into a stack, railroading a couple of tiles to be able to attack with the anti-whatever unit, softening the defense of the strongest defender with suiciding unpromoted units, and finally getting odds of >92%, I am NOT supposed to lose. If I lose these battles, it is pointless to have any kind of attack plan. If it is a game of luck, I should simply be building warriors in my first city and hope for the 10% of taking the enemy capital.

Yea, why not?

And I guess that you reload every time your last archer in your capital wins, altho he only had .9 health left?

No you don't. Never ever. 90% means that you WILL lose 1 out of 10. That means you will probably lose 20 or so in a normal game.

I'd recommend all of you to play some poker. Learn odds. Seems there's very few here who understand them. 92% does NOT mean you WILL win. It just means that you are more likely to. Seems that some people think that the 1 in a billion chance to win the lottery is about 1/3, and 92% is a guaranteed win. They're not.

Let's use this well educated man as a example:

Well, the experiment is on. Currently I'm on at most 17 victorious battles. However in the last two turns I have lost a War Elephant and a CR2 Swordsman to one Archer in one city, at odds of 87% and 85.5% respectively. Not sounding good so far...

No. Not sounding good at all. You win WAY too often. 17 victorious battles... and you lost two that are roughly 1/8 and 1/7. According to those odds you ought to have lost 1 for every 7.5 you've fought. You won 17 plus these 2. That makes 19. 19 battles and 2 losses. And you OUGHT to have lost about 1 for every 7.5. So you OUGHT to have lost 2.5 battles. But only have lost 2. That's... broken!

Make another example. Toss a coin. Write down the result. IF you get head-tails-head-tails-head for all eternity, then write a new post. However, if you by chance happen to get head-head-head-head-tails... that's bound to be a broken RNG in nature, right?

 

[Fosse]

Here:
New Random seed on reload.

100 Axes vs. 100 Spears
100 Swords vs. 100 Axes behind Rivers and on forests
100 Infantry vs. 100 Infantry
And a special treat.

I suggest turning combat animations off, and stack attack on.

Post you results here, and hopefully people will be able to link next week's "Combat Broke!" thread to this one.