I'm guessing that it's just a coincidence that they show the exact same thing every time?
Do good kings die quick?
- Thread starter highsis
- Start date
-
We have updated our Community Code of Conduct. Please read through the new rules for the forum that are an integral part of Paradox Interactive’s User Agreement.
Firstly, this is an argument from personal incredulity.
Secondly, I've only seen two or three so-called "experiments" investigating this, and neither of them used a sample size above twenty. (The OP hasn't stated their sample size.)
Apart from a much too small sample size, there are a number of other problems with these investigations such as:
- They don't take into account the total lifespan of the monarch, only the number of years on the throne
- They provide no evidence that they checked for and removed outliers
- They don't use the same number of "good" rulers and "bad" rulers in their sampling
- Instead of setting out to investigate the question of whether or not good rulers die early, they set out with the preconception that good rulers die early, leaving them highly prone to confirmation bias
I could list many more, but the point I am making is that I have yet to see someone do anything even close to any sort of meaningful data analysis on the subject.
Also, if it were true, why would Paradox keep it a secret? They're completely transparent about mechanics such as the AI cheating. There's just no incentive for them not to just come out and tell us that it's true if it is.
Its worth pointing out the Incas are almost set up to get wrecked. Their starting king is a God but he's also a general so he gets two death checks rather than 1. The heir dies young after that and then king number 3 gets killed by the civil war DHE they get. On the plus side you get a normal civil war after every reform so if the king sucks too much you can always lose. Although given the tight timescales to westernise or die thats a hell of a gambit. Spain will DOW by 1530 no matter what.
Actually, aside from statements issued alongside untrustworthy statements we don't have any evidence one way or the other. It's not like you're attacking something in Civ with displayed odds you can test against.
Simply discerning that making the ruler a general raises the odds of killing him was like pulling teeth for many months before someone definitively tested it, and only long after that did the game's UI tell you that making your ruler a general puts him at risk.
Distrust on issues like this has been emphatically earned, especially when coupled with stealth changes to rules patch to patch.
Know what's even better? There is precedence for the displayed odds to be inaccurate in EU IV, multiple times. We saw it when The_King tested 80 battles on terrain (back when it was % based) and got 80 of the same outcome in a row, despite the displayed percentage of the top terrain in the province being < 50%. You also saw grossly inaccurate fabricate claim discovery chance for numerous patches that got massacred when tested against.
Is it really so hard to believe that people don't trust the game? They have been given incentive to distrust it repeatedly, when the UI bothers to tell them anything at all. The policy whereby EU IV presents its UI and rules changes *actively encourages* distrust.
Even so, in this case I've not seen any evidence of stats impacting ruler death (only claim strength, which you can tell simply by looking at the events that kill rulers). But to just blithely assume nothing is amiss in the face of us repeatedly being lied to by the game UI and presented inaccurate information both there and in patch notes? That's reaching a bit far in the opposite direction, to put it mildly.
^ well said. Until I read this post I was beginning to think the disrelation of ruler's demise and their stat was *proven*.
When I uploaded a thread regarding 0.06% discovery chance being not accurate months ago, most people disregarded my experience saying I'm making a mistake of confirmation bias until someone statistically proved them wrong days later.
I'm not sure as I was back then about discovery change at 0.06%, but it's a fact that in my last 2 playthroughs with dozens of kings, the average reign of rulers with less than 9 stat was twice as long as the rulers with more than 9 stat. I need more samples to test it, but it's still a possibility, albeit unlikely.
On my Najd game i finished few days ago, i had a 6-5-5 ruler that was king for 50 years.... And his heir had 4-5-5 and also ruled around 25 years.. Its just a matter of luck to be honest...
In one of my first games I got the Lux Stella event for a 6/6/6 Caesar. He was only my second ruler, lived into his late seventies, and ruled for over fifty years. It was amazing. Obviously anecdotal evidence but I'm certain that stats and life expectancy are completely unrelated.
that guy is chained in paradox basement right now.
But i agree, it happens a lot more for good stat king/heirs. I watched *countless* heirs/kings die using PU mechanic, and I can safely say good kings/heirs vanished a lot. But take into acccount 1. lucky nations get better stat kings/heirs (and i watch lucky nations a lot) and 2. that AI makes good king/heirs generals *aswell*, which might lead to confirmation bias danger.
As for MY heirs/kings, the event fires a lot. But the dynasty events don't show as stats being a trigger.
People just remember good kings dying, but not bad ones.
I haven't noticed good kings die young more often then normal, but it does always seem like I always have a king that lives into his 70s anytime I have an amazing heir waiting behind him :laugh:
It's in all probability quite random, if anything my current game points to good rulers living long lives while shoddy ones die young. As Poland/Lithuania my first non-Jagiellon ruler was a godlike 5-5-6, and lived into his 70's. His heir was a quit good 5-4-6, and lived into his 60's. The replacement was a horrendous 0-1-2, but he sadly died young leading the troops against the dastardly Ottomans. His heir was a mediocre ruler 4-3-3, and lived a long life bringing much joy to the populace.
The point is, our sample size and confirmation bias tend to make us think good rulers die young or before they even inherit. I've frequently raged and tried to find the button banning hunting expeditions for my heirs, but it does affect the poor heirs as well as the good ones. Poor ones naturally get the honour of leading troops from a young age, so they tend to die even sooner.
The point is, our sample size and confirmation bias tend to make us think good rulers die young or before they even inherit. I've frequently raged and tried to find the button banning hunting expeditions for my heirs, but it does affect the poor heirs as well as the good ones. Poor ones naturally get the honour of leading troops from a young age, so they tend to die even sooner.
Bug in the 1.9.x versions which caused attrition to be calculated per country, per province, not total number of troops in the province. Though I think this affected players as well.
Bug in the 1.10.0 version which caused the AI attrition to be halved. Was fixed in 1.10.1.
I had 3 good heirs die; only for my king to die afterwards [1/1/2 at age 75]; and me to be lumped with an absolute max-length 1/0/0 Regency.
FUN.
FUN.
Wait. I'm confused.
He said "cheat" but you're saying "bug".
I thought the definition of an AI cheat was "an exception or rule deliberately given to the AI which gives it an advantage against the player".
These seem to be legitimate bugs.
How does a bug constitute "cheating"?
Every other nation is envy of a good king, so he get's assasinated faster. On the other hand bad kings should die fast too, because they are stupid and kill themselfs.
my 2 cent: i have studied statistic (several exams), i just want point out here nobody can say no or yes, because nobody has made a study about it. Of course if you can believe Paradox, but this isn't a proof (anyway i'm not english speaker so i hope this is understable)
If you want make a proof you should take a large enough sample (how much is "enough" it depends) and you should start with some hyothesis, which should be proved of rejected. And you can't just see a chart and say "Hey there is a pattern!", you should use at least some of the statistic distribution to analyze if there is or not a correlation (i.e. the Chi squared http://en.wikipedia.org/wiki/Chi-squared_distribution ).
When it comes to choice a sample, the idea is you should choice a sample which well-represent the experiment which you want make it (i.e. if you want study a correlation between election and poverty, you should choose poor people who can vote, if you choose kid or rich your data will be wrong). Normally you can define the sample about your hypothesis and reality, but in a game?
It really depends on your hypothesis only, so let say you think good king dies early (or better good king rule less than bad king) and you start take data from every religion, culture, tech and gov form. But you don't know if those things influence experiment (maybe Sunni iqta king die faster than French Catholic King, which can really screw your analysis). And you will never know if a general king died for the military code or for life_expectancy code.
So you can make a sample and analyze it, but it's probably your test will reach any conclusion, at least if you don't make many different test (and by the way this is why there scientist not agreeing about some result, because for them the sample or/and the hypothesis were wrong).
and please stop saying "my good king rule 50 years so no" because this prove nothing (as much as say "my good king rule 1 year so yes")
edit: if we have the code we could actually make an analysis to study the correlation between death and stat
If you want make a proof you should take a large enough sample (how much is "enough" it depends) and you should start with some hyothesis, which should be proved of rejected. And you can't just see a chart and say "Hey there is a pattern!", you should use at least some of the statistic distribution to analyze if there is or not a correlation (i.e. the Chi squared http://en.wikipedia.org/wiki/Chi-squared_distribution ).
When it comes to choice a sample, the idea is you should choice a sample which well-represent the experiment which you want make it (i.e. if you want study a correlation between election and poverty, you should choose poor people who can vote, if you choose kid or rich your data will be wrong). Normally you can define the sample about your hypothesis and reality, but in a game?
It really depends on your hypothesis only, so let say you think good king dies early (or better good king rule less than bad king) and you start take data from every religion, culture, tech and gov form. But you don't know if those things influence experiment (maybe Sunni iqta king die faster than French Catholic King, which can really screw your analysis). And you will never know if a general king died for the military code or for life_expectancy code.
So you can make a sample and analyze it, but it's probably your test will reach any conclusion, at least if you don't make many different test (and by the way this is why there scientist not agreeing about some result, because for them the sample or/and the hypothesis were wrong).
and please stop saying "my good king rule 50 years so no" because this prove nothing (as much as say "my good king rule 1 year so yes")
edit: if we have the code we could actually make an analysis to study the correlation between death and stat
In my Cusco game Pachacuti lived to be 58 and his son until 60 or something like that.
