The random generator HAS to be broken

[Venger]

I've modified all the hit points in the game double - and adjusted the ROF for artillery to try and keep pace.

Anyways, I attack a french veteran pikeman on open terrain with my elite cavalry (9/10 hit points).

So this is 6 vs. 3.

The cavalry lost every round, and retreated at his last point. That's EIGHT consecutive rounds of combat lost. At a 33% chance to lose, that's .33^8. Or .014%.

Ignoring the cavalry versus pikemen debate, the numbers just don't match. The unit had just landed from a ship, so no defensive modifier for fortification and he was on your average turf...

There is something in the combat generator that results in ridiculously long strings of numbers, or that incorrectly applies numbers during combat. That's a 1 in 7000 chance. Something is rotten in Denmark indeed.

I know others have seen similary "streak" results that bely logic at times. Anyone have any concrete examples or thoughts on it?

Venger

[Kenjura]

here's a concrete example.

In my current game (monarch as egypt), Rome (the big bad-ass to the east) asked me to declare war on the iroquois. All I could really do, lacking naval units, was attack a single size 1 city on my peninsula. I sent 5 archers at the city, which was defended by two archers.

I saved the game, fearing the worst, then attacked. It was the worst. All five archers were slain, with no damage to the enemy archers.

I re-loaded the game and tried again. Wouldn't you know it, same thing.

I re-loaded a third time and moved the archers to a different square. I easily captured the city.


It's almost as if the randomizer has written out all random results beforehand. Or perhaps the computer just cheats. ^_^

[korn469]

Kenjura

when you reload a game, the outcomes are "locked" so that you don't gain an advanatge by reloading again and again until you get the results you want...but i do agree with Venger that funny things seem to be going on with the random number generator

EDIT:

Venger all terrain confers at least a 10% defender bonus iirc so the percentage of that happening should be something more like .0252%

maybe you are just one of those 3 lucky people of 10,000 who got to see this occur

[Sevorak]

Yes, I'd have to say my observations with double-hit-points are looking quite similar, it's either that there's a dependent pRNG or there's an auto win percentage.

BTW, with your double hit point units, are you getting odd graphic bleed involving the tops of the hit point bars?

-Sev

[Fistleaf]

Maybe on the day of the battle, there is an alignment of the planets, eclipse of the sun, or major sunspot activity, resulting in the freak results.

[Monoriu]

I once had a veteran ironclad down to 3 hp. I made the mistake of not covering it with a full health unit and the next turn 5 AI frigates attacked it.

Result:

Ironclad became elite and with 1 hp left
all 5 AI frigates sank

Another one:

Bombarded enemy ironclad to 2 hp left. Attacked with 4hp veteran battleship.

Result:

Battleship lost and ironclad didn't suffer a scratch.



Draw your own conclusions. Mine is:

1. The AI doesn't cheat in combat. The random generator isn't favouring the AI.

2. I do have a suspicion that the random generator isn't really random.

[cassembler]

Has anybody considered the fact that perhaps units have certain characteristics (strength vs. units, etc.) that are hard-coded?

[Shiva]

Originally posted by cassembler
Has anybody considered the fact that perhaps units have certain characteristics (strength vs. units, etc.) that are hard-coded?

If they did that then we would end up screwed for modding.

[LaRusso]

can we award venger a 'whiner of the year' award, please? he produces 2,37 whining threads a day

[Altuar]

About random generator:

This I've been suspecting for quite a while, check my earlier posts about combat. Anyway, my last two games I've noticed something funny: in the same turn that a resource would run out, another one would appear. Now, since the odds for that happening is quite low, my guess is this...

(now this happens at the end of turns, the {}s denote functions)

{check for resource exhaustion}
{die roll}: 2
CHECK - eliminate resource
end function

{check for resource appearance}
{die roll}: 2 (the same one!!!)
CHECK - resource appears
end function.

Same thing happens in combat.. So the {die roll} function is broken... my guess is this: since most randomizers rely on the computer clock (since it changes constantly) as seed, perhaps the Civ randomizer looks at seconds digit - so two consecutive calls to the function, if made in the same second, returns the same result.

[Mannamagnus]

I re-loaded a third time and moved the archers to a different square. I easily captured the city.

Maybe, by moving your archers to a different square, you crossed a river?

[Cian McGuire]

Same thing happens in combat.. So the {die roll} function is broken... my guess is this: since most randomizers rely on the computer clock (since it changes constantly) as seed, perhaps the Civ randomizer looks at seconds digit - so two consecutive calls to the function, if made in the same second, returns the same result.

Nope- I've tried waiting for a little bit, such as going about the rest of my turn or walking away in frustration. What I have noticed is that it's all dependant on the turn- this goes for goodie huts as well- get something you don't like, reload, and wait until the next turn to plunder

[Monoriu]

Originally posted by Altuar
About random generator:

This I've been suspecting for quite a while, check my earlier posts about combat. Anyway, my last two games I've noticed something funny: in the same turn that a resource would run out, another one would appear. Now, since the odds for that happening is quite low, my guess is this...

(now this happens at the end of turns, the {}s denote functions)

{check for resource exhaustion}
{die roll}: 2
CHECK - eliminate resource
end function

{check for resource appearance}
{die roll}: 2 (the same one!!!)
CHECK - resource appears
end function.

Same thing happens in combat.. So the {die roll} function is broken... my guess is this: since most randomizers rely on the computer clock (since it changes constantly) as seed, perhaps the Civ randomizer looks at seconds digit - so two consecutive calls to the function, if made in the same second, returns the same result.

nah, strategc resource is a completely different question, and there is a good justification for replacing a disappeared resource immediately with another one.

Basically, I think there is a fixed set number of each strategic resource in the game. Firaxis has openly stated that there should be enough strategic resource for everybody in the game, but the trick is they are not evenly distributed. The equation is likely to be # of civs in game plus a few. If a strategic resource disappear and there is no replacement for it, then eventually the # of that strategic resource will be lower than the # of civs in game.

Hence, its deliberate that when a strategic resource disappear, another appears immediately. You can say that strategic resources never disappear. They just move around.

[Wrong_shui]

Venger, give in, people wont accept its broken even if your had soren come on here and say so.

[Boney]

I too have had my probs with strange results. I have even seen strange results between the AI. I once had 2 Man o War attacking and being defeated by a caravel. No Prob. The Spanish Armada is testament to that possibility.

What I don't like is that a warrior can enter a city with 20 battleships and 15 bombers and destroy them all. any comments?

[Grumbold]

I, personally, in real life managed to roll a 1 in 46,656 chance with a balanced die in front of 4 witnesses. Guess that means I'm broken, huh?

If you fight 2,000 combats in the course of a game (I tend to fight rather more...) then the odds of a 0.1% chance arising at least once approaches certainty. Its like the UK National Lottery. Only a 1 in 14,000,000 chance of winning and somebody achieves it practically every week. Go figure.

[Boney]

Talking about national lottery. I can remember in the early 1990s when someone bet with ladbrokes that Elvis would have a concert with the Loch Ness monster in Sun city before the year 2000. The odds they gave him were 10,000,000 to 1. The joke at the time was winning the lottery was less likely to happen than the concert in sun city.

[Bobstah]

Originally posted by Grumbold
If you fight 2,000 combats in the course of a game (I tend to fight rather more...) then the odds of a 0.1% chance arising at least once approaches certainty.

The problem is not with 0.1% chances once per game but with 0.000001% chances fairly frequently. There is clearly a 'random' chance that a unit will win regardless of the odds.

Its like the UK National Lottery. Only a 1 in 14,000,000 chance of winning and somebody achieves it practically every week. Go figure.

1 in 14million per ticket, but if there's 14million tickets...

Maybe we're playing on the Discworld, where 1 in a million chances crop up nine times out of ten.

[zapperio]

Originally posted by Wrong_shui
Venger, give in, people wont accept its broken even if your had soren come on here and say so.

Down, dog! Bad puppy! Don't slobber all over the nice people. Sit!

Zap

[Grumbold]

Originally posted by Bobstah
The problem is not with 0.1% chances once per game but with 0.000001% chances fairly frequently.

Oh really? Like what? The closest I can find to that is a veteran modern armor attacking an unfortified spearman in plains and not inflicting any damage before dying. I've never lost any of those fights, let alone without inflicting damage. It must be your computers internal random seed generator that is broken. Better take it back to the shop and complain.

[zapperio]

Well, hmmm.

I've been doing some testing.
I wish I had the numbers in front of me but I'm at the office right now but I can share my limited conclusions.

Like most of you I had seen streaks that have made me blanch also so I wanted to know how exactly the RN generator was working.

I placed 3 artillery, 6 horsemen, 3 sword men, 2 pike men and a warrior in three stacks, within striking range of four stacks that consisted of unfortified 2 horsemen, a bowman, a pike man and a warrior. All are on desert tiles with no modifiers.

I re-loaded around 30 attacks with 5 different combinations of units and tried hard to spot a pattern.

Example of a combination would be
1st attack hm(1st stack) attack, swm(1st stack)
2nd swm(1st stack) attack, pm(1st stack)
3d hm (1st stack) attack, hm(2nd stack)
4th pkm(2nd stack) attack, w(3d stack)
5th hm (3d stack) attack, hm(4th stack)

If I used the same combination twice I’d invariably get the same result, pointing towards a fixed queue of seeds. However, if I changed only one of the attacks, ie replaced

2nd attack swm(1st stack) attack, pm(1st stack)
with hm(1st stack) attack, pm(1st stack)

then the result for 3d, 4th and 5th attacks would change also.

Anyone have any theories as to how that may be working?

Zap

[Grumbold]

There is certainly a queue of numbers lined up to be used so you can't easily cheat your way out of an unlucky streak of numbers. I assume this is an early part of attempting to make play-by-email games etc more playable because you can't reload your turn repeatedly to win all your fights or find a settler in every goody hut. The best you can do is have a few meaningless attacks prepared so if you are desperate you can insert some shelling of a coastal tile in among your assault attempts to 'spend' that bad random number. This is not perfect but it does at least mean you waste some potential part of your combat power.

[zapperio]

Yes, there is a queue, but all you need to do to break the queue or reset the queue is to use a different unit in 1st, second or third attack the second time around. Maybe I wasn't too clear about my experimentation above but the conclusion was that

If I didn't like the result from my multi-battle combat I just have to reload and change the order of attack.

[zapperio]

To make it even clearer:

You don't reset the queue because no matter what combination you use you'll get the same result for that combination, but you can reload untill you find a combination that is in your favour.

Zap

[Excelsior]

Originally posted by Venger
I've modified all the hit points in the game double - and adjusted the ROF for artillery to try and keep pace.
Venger

I would recommend quadrupling the hitpoints. It takes four times as long to fight a battle, but it is becomes much more likely that any streakiness will reverse itself before you lose (or win) the battle. Try this and see if you agree.

[Excelsior]

Originally posted by Sevorak
BTW, with your double hit point units, are you getting odd graphic bleed involving the tops of the hit point bars?
-Sev

I am! The colored ball at the top of the unit sort of streaks across the screen as the unit moves. It is not that bad a thing, though, and it only seems to happen when the units are next to a border. Increased HP helps combat enough that I'm more than willing to live with that.

[J10]

On another threads people were accusing me of making sh]t up.

I attacked an enemy calvary with 1 hp left and lost 2 elite calvaries and win with 1 veteran with 1hp left, and they calculated the chance is 0.002% or something. I also lost 3 fortified riflemen to 3 calvaries in a size 12 city in a row. Look, I don't care if you believe me or not, but that was totally a fun killer, and that's what I care about, cuz overall I still think Civ3 is a good game, but that toally tempted me to reload, and I don't think reload autosave is fun at all.

[Excelsior]

Originally posted by J10
I attacked an enemy calvary with 1 hp left and lost 2 elite calvaries and win with 1 veteran with 1hp left, and they calculated the chance is 0.002% or something.

I calculated the odds at 0.00014%.

Originally posted by J10
I also lost 3 fortified riflemen to 3 calvaries in a size 12 city in a row.

The odds of that was a more reasonable 0.2% (assuming 4 HP each).

[SuiteSisterMary]

I will point out that spear/pike units have SOME sort of advantage against horse-based units (I think it alludes to it in the civopedia) but I've been saying since the second day I had this game that the PRNG is broken. Hideously and viciously.

Has somebody tried taking Civ2, modifying IT'S units to have Civ3-like HP and A/D/M and running some combat?

[Excelsior]

Originally posted by SuiteSisterMary
Has somebody tried taking Civ2, modifying IT'S units to have Civ3-like HP and A/D/M and running some combat?

I posted this in another thread a couple of days ago.

Originally posted by Excelsior
I just tested this in Civ2 with legions vs. warriors and I found that not once was the attacking legion defeated. Never did a legion lose more than 50% of its hitpoints. I then edited the rules such that both units had firepower of 3, essentially reducing the hitpoints to 3.33. Thereafter, the legions had a much more difficult go of it. Out of ten subsequent combats, 3 were unharmed, 3 lost 3 hitpoints, 2 lost 6 hitpoints, 1 lost 9 hitpoints, and 1 was destroyed.

I am not convinced that the random number generator is broken, but I admit it does seem suspicious. More often than not, it seems, units get on streaks. One unit will get 6 hits, then the other will get 4, alternating back and forth. The streaks, though, seem to last longer the greater the unit's combat advantage. Thus, greater hitpoints reduce the effects of the streaks, making combat more balanced.