Against all odds!

[Mao]

Originally posted by Xerxes712


Spank me if I'm wrong, I read there is a change to that; you are correct but....

The new patch has an option to turn on/of the random seed gernerator ability if you reload a savegame now.

That way you do not have to 'adjust' the odds to get a different result.

But of course, none of us would never admit that our super unit just got wasted by a lucky hit by a lowly inferior unit..

*spank spank*

...



Oh. He was right? No matter...

*spank spank* :P

On a more serious note: If units are stacked, do their strengths really get summed like that? If so, it seems to me that if you just stack up a massive number of units, then they'd be unbeatable, right?

[Aro]

Originally posted by Mao



On a more serious note: If units are stacked, do their strengths really get summed like that? If so, it seems to me that if you just stack up a massive number of units, then they'd be unbeatable, right?

Well... Nope. The strenghts aren't summed. It seems to be just a way to save clicks, or so, in combat. Never tried.

[Merzbow]

You need 5 turns fortified to gain the +25% bonus? I don't think so...

[Bhruic]

Fortification gives you +5% per turn up to a maximum of 25%. 25%/5% = 5, so yes, it would appear you do need 5 turns fortified.

Bh

[Saurus]

Uberloz , I fully undestand your pain! This kind of oddness has happened to me too after the patch.
Here is my story.

To clarify the problem:
I have 2 knights both with 54% change (odds) of winning a battle against my opponents knight - after the battles are over my opponent is standing there - still at 100% strenght!

(a compleatly another story is that at this point my desperate, severly damaged swordman whitout promotions attacked the 100%-strenght knight with 2 combat promotions from behind a river and WON receiving 19 experience points from the battle)

But back to the point,
The next turn the situation with the Knights plays out again at another place with 2 of my knights facing 1 enemy knight - according to odds a 50/50 situation for each battle.
Again, both my knights destroyed - the enemy knight this time around remains at 80%-health.


Conclusion:
Ok, let's forget about the swordman-thing - even at 0,9% odds it is possible to win from time to time.
But the fact that I lost four 50/50 fights while only managing to reduce 1 enemy knight to 80% strenght seems to me a far more strange thing than the swordman winning the single battle.
Battles are divided into turns and since all knights involved were at full strenght and the odds where 54-46 in my favor for the two first fights and 50-50 for the next two, It must mean the odds for each TURN is about 50-50 as well. Acording to the battle log, a knight att full strenght needed 5 hits to destroy another knight ... so calculated in battle turns this means that at 50% odds (for the sake of simplicity) for each combat round the A.I manage to hit me 20 times during the time I hit the A.I one time.
The combat log told me that during theese four battles A.I hit me 19 times in a row until I managed to hit him once ... then he finished me off.
The probablility for this to happen under theese cirmustances should be about 0,000019 %
(By all means, correct me if I'm wrong!)
- not so strange I feel there IS a problem here!

There were several of theese odditites in my latest game ... yet sometimes even in my favor.

But the problem is that there seems to be HUGE swings in how combats plays out.
If under EQUAL conditions the enemy is outnumbered 4-1 there should not be possible that when the dust settles, the enemy still stands there - almost compleatly unscratched. It takes the strategy out of the game when I cannot count on the strategic setting presented to me since the results plays out differently.

[Urban Ranger]

Originally posted by player1
As blake said...
Custom option, 1.52 patch only.

Actually, it has been in the ini file all along IIRC.

[uberloz]

Originally posted by Aro
uberloz, we live in a statistic universe. Everything is possible... Even if not likely.
And that includes Civ IV.

Thanks for that, I feel so much better now.

j/k, yeah I know that.

For some reason I keep expecting a real breakthrough in 'ai' programming that allows for an imitation of a real human opponent.

If any game series is harnessing some serious brain power for 'ai' design I would vote for this one. And for that assumption I probably expect more than current technology is able to give.

Oh well, the existence of these 'superhero' units is probable, but there are way too many for my taste.

[Diadem]

Reloading always produces the same result. This is done to make 'cheating' by reloading lost combats until you win impossible.

I sometimes did that in civ1 and civ2. Pop a goodie hut, and if I don't get a good result, reload and do it again. But that always made me feel a bit dirty inside. I'm glad the game doesn't tempt me to do that anymore

And yeah, loosing 3 full strength gunships against one damaged one is pretty extreme. But it's not impossible. Human beings have a pretty lousy intuition for statistical results. What are the odds of loosing a battle like that? Perhaps 1 in 1000? That's very slim, so it should never happen, right? Wrong. How many battles do you actually fight? If you play this game a lot, and play it militaristicly, then you'll fight many battles. Over the past few weeks you've probably done many thousands of combats. It's hardly surprising that such extremely unlikely results show up then. In fact, it would be weird if they didn't!

And of course, there are many, many, people playing this game. Maybe there are 999 people who are not as unlucky as you. That is still not weird. The odds of winning a lottery are extremely small - but you know that *someone* will win it.

[Kataphraktoi]

What are the odds of loosing a battle like that? Perhaps 1 in 1000? That's very slim, so it should never happen, right? Wrong.

There are some statistics that only can happan on paper. if you flip a deck of cards out of a plane, they will never reform and land on the ground-no matter how many times they are flipped. This is because order cannot arise from non-order without something external to force them into a deck.

Have you turned off the RNG and redone the examples EN? I wouldnt take that data at facevalue without some better tests.

[Diadem]

Originally posted by Kataphraktoi

There are some statistics that only can happan on paper. if you flip a deck of cards out of a plane, they will never reform and land on the ground-no matter how many times they are flipped. This is because order cannot arise from non-order without something external to force them into a deck.

Actually, they can. The odds are just extremely small. This is the essence of the second law of thermodynamics, which you are referring to. It's a statistical law: An increase in disorder is more likely than an increase in order. However, the odds we are talking about then, are really extreme. In the order of 1 against 10^10^10. So even if you fill the entire universe with testing apparatus and you'd experiment during the entire lifespan of the universe, you'd still need extreme luck to find just one violation of the 2nd law.

However, all this is completely offtopic. We're not talking about physics here, we're talking about simple game mechanics.

Combat odds of 1:1000 are not vanishingly small. If you wage combat 2000 times, you'll find them twice - on average.

[grutz]

Personally, I think this is a sign of a good RNG. After all, if it's truly random, the dice have no memory. So it's very possible to get a run of odd results. As such, it's just as possible for the situation to flip the other way. Not likely, perhaps, but possible.

[Blake]

What I would actually like is a "Disengage" feature for combat, if the odds are overwhelmingly slanted (say 95%+) then the high-odds unit cannot actually lose, if it were to lose, it instead disengages, which acts like retreat (except it can force an attacking unit to retreat). This would mean that very one-sided battles can only end in victory for the stronger unit, or a draw.

How I would do it is have the disengage chance = 10* odds-89. 90% odds gives a 10% disengage chance on loss, up to 100% for 99%-100% odds.

The reason I want this is because it's a *strategy* game, there isn't much strategy in extreme (un)luck! I can live without winning a 99% odds battle, but losing such a battle is a real upset. This is why a draw is a nice happy medium, the element of uncertainty remains without being a massive upset.

The disengage concept could also be used to protect capitals - it should be impossible to take out a capitals only defender in one combat turn with only one attacking unit. By this I obviously mean raiding someone with the starting warrior and attacking their capital on the offchance your warrior wins (I think it's around 25% chance if it's pre-first border expansion). Frankly I don't think that offchance should exist.

[Runner]

I'll have to keep an eye out for this sort of thing, I don't believe I've seen any extreme examples such as the ones told of in here yet though. Then again if I did I probably got really pissed off and loaded the game and changed the seed. One thing I did notice though, at least before the latest patch, was that my nuclear plants tended to pop off in a not so random or at least horribly stacked way against me. By the time I stopped playing that game, 2067, I had had 7 nuclear plants explode, having had them roughly 70 years. 2 of them went off one after another in 2 turns around 2020. I also only built 10 of them. Pre-patch I believe it was a 1/1000 chance that they would go off, now since I had 10, 10/1000 or 1/100. Wow, I must have been really unlucky... I learned my lesson though, Coal or Hydro for me now. Haven't built a nuclear plant since.

[Crossfire]

I had a little war last night and was pretty surprised by the results. I was attacking vanilla archers with elephants, some fortified, most not. I probably fought 20 battles. In all but about 3 battles, one elephant died in the first round, and all but once, the second elephant killed the other 2 archers. The results are not that unexpected, just the reliability of predicting the battle was too weird. Something fishy or just a run of similar luck?

I think I prefered the reliability of warfare in MTW better. You didn't always win but you had better control over your chances. Maybe it just came down to having 40-100 tries to average it out against versus just 3.

[xxFlukexx]

Ha, to anyone who wants examples, I have lots of examples(started cleaning some out recently, but I've got some great ones)

I had this happen to me my first game I played on 1.52:



99.1% chance to win and I don't even get a friggin' shot off, come on, COME ON. I dunno what the odds are for me not even winning a round(used to be able to calculate that easy pre 1.52) but it has to be less than 1%, much less. Oh and my score is so low because I'm playing on Emperor, not because I suck.

This is what pisses me off more than anything in Civ. Now if I'm in the Renaissance Era, I don't really care if I lose one troop to ungodly luck because I probably have 10-15 other troops around. But if I lose like this in the early game, it is catastrophic. Early attacks in the B.C.s consist of about 5-6 units, so If I get really unlucky, boom game over wtf that was lame. Ancient Era combat can come down to a few key 1on1 battles as well, once I was playing on Rome, set up everything brilliantly, had 3 Preatorians patrolling my homeland while my army was off attacking, 2 horse archers and a spearman(he's on a hill but it's still a spearman) enter, boom boom boom all 3 preatorians dead, both horse archers alive. Great, now they get to pillage my iron and towns and other resources.

And I DO see this all the time, I haven't had my initial scout/warrior last for more than 20 turns in a long time.