Battle odds conspiracy

[lange17]

Am I the only one who feels like he loses battles in the 80-90% range on the odds calculator incredibly often? They feel like 50-50 gambles, and it seems like the rare game when I can get around and visit a few huts and make it home with my initial unit without getting him killed on good defensive terrain. I've wanted to post this before, but I just lost a game on Prince when an unpromoted Barbarian warrior attacked across a river into my 40% culture capital and killed my fully-fortified Combat I warrior. The combat log says he was 0.5% to win the battle, but win it he did.

I realize that these kind of things happen if you fight enough battles but... why me?!? The game hates me, its the only explanation. I demand an audit, or a written apology from Sid, or something...

Ok, thanks for listening. Time to load up another one.

[YC4B4U]

I know exactly how you feel, happens to me constantly. I have often thought about recording my win losses based on percentage chance of winning.

My gut feel is:

<50% Don't bother fighting - not likely to even chip the paint (although I must be fair and admit I did win one battle at 37%
50% - 65% Good chance for a little bit of damage, better just to wait one more turn to bring down the city defence or send in a catapult etc
65% - 80% Might get lucky and score the occaisional win - have lots of units in reserve just in case you get an unlucky streak
80% - 95% The frustration zone - you know you should win, but.... only 50/50. Ensure you have twice as many units as your oponent
95%-99% OMG I can't believe that just happened zone - Should only lose 1 in 20 battles (isn't that right?) but seem to lose 1 in 4.
>99% Warlord death zone. I have lost over a dozen battles with Warlords in this zone. Just trying to get an easy experince point with little risk, or so I thought.

Having said that though - I will have to keep statistics, as I know that I remember that losses more than the wins...

[sjm]

Not wishing to sound too harsh, "a bad workman always blames his tools".

But to be fair, humans are very bad at appreciating "randomness" - if you ask a human to plot 1,000 dots on a piece of paper, the dots will never be random; people will feel the need to spread out the dots fairly "evenly". This appears to be more "correct" to us, but if you ask a computer to plot the same thing using a series of random numbers, you'll see much more clustering.

Another example: my statistics and probablity lecturer asked us to go home and write down a series of "random" 1s and 0s. We were to generate two sequences. One by flipping a coin, and one by writing down a series that we "felt" was random. He claimed he would be able to tell which was really random, and which we had made up. How? by counting the longest series of identical digits in the two sequences; the sequence with the longest uninterrupted series of 1s or 0s would be the "random" one, as we humans would overcompensate, and tend to write a "1" after 3 or 4 "0"s. I must admit, I cheated and purposefully wrote a longer series of 1s in my own number, after generating the random one using a coin.

Add that to the fact, that we only ever notice that we lost a 95% victory when we lose, and don't think much of the 19 victories, and the fact that a human is unlikely to attack very often at a 5% chance (and thus never be on the winning side of a losing battle), our perception is horribly skewed.

That said, I've won more than a few 0-5% battles using siege weapons in my time!

[trev]

The calculated odds are pretty good usually, but some multiple move units are calculated incorrectly when attacking across rivers, this includes single move units using roads. So allow for decreased odds if the 25% river defence will come into play, but otherwise trust the odds generated.

[Nubclear]

Indeed. I've won too many 0-10-20% battles that even a string of other battles is fine by me.

[Thedrin]

I realize that these kind of things happen if you fight enough battles but... why me?!? The game hates me, its the only explanation. I demand an audit, or a written apology from Sid, or something...

Enter Worldbuilder. Create 100 of your own units [of the one type on the one tile] next to 100 barbarian units [of the one type on the one tile]. Exit worldbuilder. Check the odds of one of your units beating one of the barbarian units. Perform 100 attacks.

Everytime I question the combat engine's random numbers I do that. The results always remove doubt about its integrity.

[couerdelion]

I know how you feel but I think that there is no conspiracy. I happened to get into another war yesterday and noticed that I was getting hit on some poor odds. If I was unlucky then it was just my turn to be unlucky.

I was certainly a little peeved when my Infantry failed to kill enemy cavalry. And I think I had at least one city raid combat with odds of 98%+ which I lost. The really galling one, though was my Combat I Frigate attempting to sink a Galleon loaded with settler and longbow. Not having a particularly large navy it was quite distressing to see my boat go down with all hands. I can tell you that I was very nervous by the time a second frigate had arrived to sink the enemy.

I think it does all fall down to selective memory.

[Mr Justice]

also note that if you reload a save and do the combat again it will give same result, so if you try doing this it gives the Apprearance of calculating combat again but infact it doesnt and will give same result, makeing it look even more skewed in the AI's favor.

[Diadem]

In fact the random seed is saved in your save game. So if the next random figure that will be created is a bad one, then reloading won't change that. Any combat you then try will have bad results. If you lost a combat of 70%, reload and try a combat with 60% with a different unit, you will loose that one as well. Guaranteed. And if you try an 80% battle after reloading again, you will suffer very heavy damage in the best case scenario, but more likely you'll just loose.

[couerdelion]

So what you could do is set up 100 combat situations each with a different win probability and save from the start. Start with the weakest and register result. Reload and try the second. Reload and….

There a simple way of getting a more accurate idea of the number generated. Once you have a victory you have a small range for the random number and can replace the unit the won and go through the whole thing again

You’ll have a stream of about 2 random numbers in no time at all this way

[lange17]

To be clear, I know in my brain that there is (probably) no conspiracy. My frustration at losing a whopper just gets to me sometimes and I wanted to vent to some people who would understand my pain rather than just yelling at the computer screen like usual.

I will admit to twice starting a spreadsheet to track battle outcomes based on slotting them into "% expected outcome" bins. Of course, I didn't stick with it since the game detected my spreadsheet and decided to be nice for a few days until I gave up tracking it.

I guess part of the trouble is that a major advantage I need to use as a human player is being smart enough to use mobility to make up for sheer numbers of troops, this projecting more military power with less expenditure of hammers and money. If I choose to do that by defending each of my cities to a 99% win chance, all is well until I get the one unlucky dice roll (or series of rolls) and strike out. I just have to learn to spread myself as thin as possible, and no thinner.

And restart that spreadsheet of course. Trust but verify!

[zeace]

Anytime I lose a particularly important and heavily favored battle I like to stop, breathe, and imagine the headlines on the newspapers the next morning. "Miraculous Spearman Destroys Battleship" or some such. Imagine the people in the streets partying at their good fortune. Then imagine them fleeing from the party as my next wave of troops hits the city.

There have been some great battles that have not gone as they should have, and we remember them as the great turning points in wars. I like to play civ the same way.

[Calvin Vu]

Originally posted by couerdelion
I think it does all fall down to selective memory.

How about selective odds ?

I don't think the odd is stacked against me at the beginning of the game. However, near the end of the game when I am way ahead and just want to roll over the AIs to wrap the game up then it seems to be much harder to win and I have the impression that an 85% chance only gives me a 50% of winning.

BTW, after the new Warlords patch, the odd of capturing a worker is only 99.9%. Have anybody lost when attacking a worker ?

[kiky]

The other day I lost a maceman, while he attack with 99.9% of winning ....
That was few turns after another unit of mine died on 95%+ winning..


anyway, back to topic
The percentage is calculated without using HealthPoint. So even if it 95%+ but this unit has very low health point it might die. One strike from enemy went through and he will die.

[Welshie]

Originally posted by Diadem
In fact the random seed is saved in your save game. So if the next random figure that will be created is a bad one, then reloading won't change that. Any combat you then try will have bad results. If you lost a combat of 70%, reload and try a combat with 60% with a different unit, you will loose that one as well. Guaranteed. And if you try an 80% battle after reloading again, you will suffer very heavy damage in the best case scenario, but more likely you'll just loose.

I'm fairly sure that this isn't true. Fight a different battle with lower odds and you could just as well win, as long as the circumstances of the battle are different.