The hidden power of agg+cha (or, why Boudica is actually the best leader in the game)

[Anysense]

In the early game, bad RNG can screw you over. Later on, if losing 5 80% battles in a row inconveniences you that only means you did not make enough units.

It also means that you are probably on a loosing streak. RNG in CIV tends to produce very long streaks of good or bad luck.

 

[Fish Man]

It also means that you are probably on a loosing streak. RNG in CIV tends to produce very long streaks of good or bad luck.

I've heard that be true directly of the algorithm, and several people have chimed in to say on the whole they don't see that happen or negligibly at best...so, I'm still not certain about it. But yes, perhaps confirmation bias, but I notice this a lot, to the point where it's one of the first things that started my reloading habit (one 60%-80% loss shouldn't doom you to 4). At the same time, though, enjoying 3 20% wins in a row feels really good... though I think honestly, if both types of streaks were evened out, I would consider that a worthy price to pay for the ridiculous loss-streaks and maybe consider not as many reloads.

 

[Anysense]

I found it when AI appeared to cheat winning a lot more fights than was probable. After some 40 recorded outcomes the proof of that cheating seemed to be in the bag, as RNG kept favouring AI in obscene way. But then I began to win almost every fight and soon the mean was close to expectation. People who say it is negligible probably have short memory for such things, because it is obviosly not so. I had even HoF games ruined by RNG after losing 15/20 Praetorians at 70-80% winning odds (and of course every catapult at up to 50%). It is even more of a problem in ordinary games where a much smaller string of losses is enough to ruin it, hence the chance of it happening is several times greater.

 

[civac]

These kind of things happen in random sequences. It's also a typical failing of human cognition to see patterns where there are none.

 

[Pangaea]

I know there has been some larger auto-battles or something where it comes out as expected, so there probably isn't something inherently wrong about the RNG mechanic itself. But I have also played a stupid amount of hours of this game, and know when I see things that feel unfair, and 5-6 85-ish percent battle losses in a row are definitely in there. It can really mess you up, even in the middle or late game. In the early game it can be an outright loss, where it hurts a lot more.

No matter how much this is discussed, and people throw out this or that term (I know them, I'm educated), this aspect still remains the one blotch on the game for me. It can drive me away from the whole darn game because it's so frustrating and unfair. You put tens of hours into a game, meticulous planning, bulbing strategies, whatever. Then it can all come crumbling down because the RNG has a really bad day. Then it's tempting to call it a Random Loss Generator instead.

I'm sure it's nice for testers or theorists that it comes out as expected after 10,000 battles, but we don't have 10,000 units in a stack.

 

[Fippy]

Inaccurate RNG systems (comes with how comps work) tend to streak, more than if you'd just roll a dice over and over.
So while things even out long-term, i think it's well known how they can be unfair in shorter sequences.

 

[sylvanllewelyn]

With or without HC? Sadly he's the best leader for almost anything
Best traits for Iso (and maybe also semi) are Fin, Phi, Cha and Ind.
Prolly Lizzy if no HC. She can have much more trouble with barbs, but even a slower start can be boosted with those optimal traits.

I think financial leaders do not have trouble with barbarians because they can afford a detour to Archery. But what about semi-isolation with an aggressive neighbour? With Huayna Capac, as long the AI does not settle on top of strategic resources, and that the AI does not have an early resource-less UU, I can pillage the strategic resources. But how about Elizabeth? The best I can think of is research metal casting, get a great engineer for machinery, and train crossbows. If I have iron. But I would probably die well before that point!

Inaccurate RNG systems (comes with how comps work) tend to streak, more than if you'd just roll a dice over and over.
So while things even out long-term, i think it's well known how they can be unfair in shorter sequences.

Pseudo-random sequences are OK as long as the users cannot predict the result. Streaks are a lot more common than most people think. I am an advocate of both a random and a deterministic component to resolving battles. Ancient civilisations are subject to the whims of fate such as natural disasters or surprise defeats in one battle. Animals need to be nerfed. Axemen on flatland have 1% chance of losing to barbarian archers - - or bears.

 

[Lexicus]

At the same time, though, enjoying 3 20% wins in a row feels really good...

Honestly, for me I don't even really care about low-odds wins because I never attack at low odds with anything I'm not prepared to lose. It can be nice to get 2 promotions from 1 combat though.

In the early game, bad RNG can screw you over. Later on, if losing 5 80% battles in a row inconveniences you that only means you did not make enough units.

Could be, but there's also situations where e.g. you have trebs against hill longbows and you lose your 3CR trebs to bad luck, forcing you to attack with only 2CR trebs leading to a much higher casualty rate overall (and you probably have to at least delay your offensive because those 3CR trebs are worth ~3 times their number of 2CR trebs and you now need to bring those 2CR trebs up to keep pushing).

@OP awesome story, very entertaining to read.

 

[Lexicus]

but Buddism is only a single religion. Was you goal to cripple only Ragnar? Why don't you founded Hinduism / Judaism too (following similar arguments)?

It is an interesting strategy to cause quarrels between AIs if you know for sure you are in the Astro isolation (when you win all early religions) but I cannot justify it as lack game experience.

I may be wrong and wouldn't presume to speak for @Fish Man, but I believe that Buddhism is crucial to deny because it comes slightly before Hinduism and significantly earlier than Judaism, and that time delay is crucial in allowing Buddhism specifically to spread to every AI before some other AI can found a religion and begin spreading that other religion significantly. The goal is less to cripple any particular AI than to prevent the AIs from all perpetually allying: allowing two different AIs to found religions close enough together in time in turn allows religious strife to spread across the continent. And war slows the AI's teching and development down, for obvious reasons.

 

[Fish Man]

I may be wrong and wouldn't presume to speak for @Fish Man, but I believe that Buddhism is crucial to deny because it comes slightly before Hinduism and significantly earlier than Judaism, and that time delay is crucial in allowing Buddhism specifically to spread to every AI before some other AI can found a religion and begin spreading that other religion significantly. The goal is less to cripple any particular AI than to prevent the AIs from all perpetually allying: allowing two different AIs to found religions close enough together in time in turn allows religious strife to spread across the continent. And war slows the AI's teching and development down, for obvious reasons.

Yes, this is exactly it. Saladin got Poly and Joao got Mono, so Ragnar's continent didn't get a religion until Islam, which he founded in Haithabu. For thousands of years, then, the three monster civs on that continent who got at least 10 cities apiece, traded with each other less, and most importantly, three wars broke out. Qin declared on Surry, and then Ragnar declared on Qin, and later Surry declared on Qin too, but not before I cleaned him up and got the capitulation. If they were all friendly, I would've been bribed on instantly when I landed on Qin, and everyone would have been an era ahead because of the constant tech gifting.

 

[kostja]

This is a very interesting discussion. The question that remains then is, can and should this be incorporated into a standard campaing, where the player does not have the knowledge on T0, if his civilazation is isolated.

At which point, should one beeline for Mediation if there is any indication of an isolated start?

Kind regards

 

[Fish Man]

This is a very interesting discussion. The question that remains then is, can and should this be incorporated into a standard campaing, where the player does not have the knowledge on T0, if his civilazation is isolated.

At which point, should one beeline for Mediation if there is any indication of an isolated start?

Kind regards

Extremely difficult to tell; probably not feasible to ascertain 100% whether you are iso by turn 5, which is when you have to select a tech.

The only other situation where you can go meditation without much worry is, say, if you're Pacal and you have a corn + flood plains + gold start, and you haven't met anyone by t5. Then, going meditation gives you a benefit no matter what the result may be, a veritable Batman gambit. If you're isolated - great, the Buddhism denial and benefits associated with early religion play out. If you're not, then you can still probably guarantee an easier diplo situation by spreading your religion to your neighbor in due time, or burying a useless religion if your neighbor founds, say, Hindu. And for all these instances, and even in the freak accident when you somehow don't get Buddhism, an Oracle play becomes a high possibility since you getting meditation means another AI switches tech to something else and will get priesthood later (and also, Oracle is far more plausible with a good start). BUT - in that case, even if you're iso, Pacal with FP wetcorn gold is already a won game from t0 in 99% of situations, so whether or not you go meditation is a moot point?

 

[SittinDown]

5-6 85-ish percent battle losses in a row are definitely in there.

I believe the theoretical math of that situation is pretty straightforward:

5 wins , 0 losses is ~ 44% = (.85 ^ 5)
0 wins, 5 losses is ~ 0.01% = (.15 ^ 5)

So, if you are seeing 5 losses in a row, that's about 1 in 10,000 event. (1 / (.15 ^ 5)

If that happens once or twice in your Civ 4 career, it's probably just bad luck, but if it's come up many times, it's probably bad RNG.

 

[Anysense]

The example I posted earlier is about 1 in a million event, unless RNG is rigged. A simple way to check how likely an event is is to write a program that will run that event a million times. Takes a few minutes to write and takes all doubt away. And nobody can argue that probability theory is a mess and its derivations are unreliable. After all, we are talking about RNG and not mathematical theory or truly random events.

Edit: One can still argue that odds displayed in Civ are a mess, though.

 

[Pangaea]

Cool. So all we need to not get an extremely sore arse is a stack of one million units at all times. Checks out tbh. Fed up again. Bye.

 

[6K Man]

Maybe the idea behind the occasional long string of long-odds losses or wins is to simulate one side winning a battle against impossible odds, as sometimes happens in the real world (e.g. Thermopylae)?

Not that I am advocating for that. This is a game, not a simulation.

 

[Anysense]

Cool. So all we need to not get an extremely sore arse is a stack of one million units at all times. Checks out tbh. Fed up again. Bye.

Not necessarily. Something like this will do

Spoiler :

 

[civac]

Maybe the idea behind the occasional long string of long-odds losses or wins is to simulate one side winning a battle against impossible odds, as sometimes happens in the real world (e.g. Thermopylae)?

Not that I am advocating for that. This is a game, not a simulation.

Streaks are completely natural occurrence in random sequences. This idea that there is something wrong with the pseudo random number generator is unfounded. The exact same things which would make you and Pangaea mad in current Civ4 would still irk you if it used real randomness (if there is such a thing).

 

[Anysense]

Streaks are completely natural occurrence in random sequences. This idea that there is something wrong with the pseudo random number generator is unfounded. The exact same things which would make you and Pangaea mad in current Civ4 would still irk you if it used real randomness (if there is such a thing).

Er... I just gave you the code which shows that you are quite wrong. Streaks very common in Civ4 are extremely rare with an RNG which had not been tampered with. Besides, fights in Civ are not decided with one roll of a dice, meaning that when you lose five 85% fights in a row you don get unlucky an outrages five times in a row, but lose about 40 times out of 50, while the most probable outcome is winning 40/50 and winning 10/50 or less has miniscule probability.

 

[civac]

That's not current Civ code, or is it?