I think the bottom line to your argument is that you have a vision of the game, and anyone who sees it differently is wrong, which is ironic given your last statement.
Surely that’s your argument since a survey given to the players, CCP and an article from INN support my argument.
Meanwhile who or what agrees with you?
The problem is that correlation is not a causation and you both act like correlation had any intrinsic meaning.
A correlation is just an observation that variable are varying together (from -1 for “those variables are in total opposition” to 1 for “those variables are always moving together”, and 0 for “no correlation at all”). This observation is only valid for a given dataset, so another dataset may as well give the opposed correlation . So trying to make general rules from correlations in graph is just a nonsense : correlations help you to try to find causations, but NEVER are a sign/proof of causation.
The second issue is the use of un-rigorous terms : you use terms that are not rigorously defined after looking at the data, so you can use the terms that fit the data with an error margin acceptable. Then your confirmation bias (that everybody has) will stretch the data and the terms used. But in reality, this is just making arbitrary interpretations without any rigorous sense.
So things like “saturated”, “stagnated”, “healthy economy” too are just made-up terms that have no reality besides the will for their user to create a representation that match their hidden agenda.
You need to define those terms BEFORE you start looking at the data, otherwise your interpretation is biased and meaningless.
1 Like
So your argument is based on opinion. Inherently subjective and devoid of any actual defined criteria. Yes, I know.
I also think it’s funny that you side with CCP as doing the right thing when you are against them making hi-sec safer. So they are right when they agree with what you want, but wrong when they disagree with what you want. Do you want to keep digging your hole or?
All causations are correlated. You can disprove causation by showing an absence of correlation. I was telling him you can’t determine no correlation exists from the graphs, which implies you can’t rule out a causation. That was my message.
No they are not.
Since you can have a correlation in a dataset, and no correlation in another one, this is also the case even if there is a causation : some causations can be “hidden” by eg data sampling, probabilistic noise, etc.
Therefore even in the absence of correlation there can still be a causation.
You can observe there is no/low correlation in the graph, but you are correct that you can’t rule out causation just because you don’t find a correlation in a specific dataset.
1 Like
That’s lack of data. That’s a practical issue, not a theoretical contradiction. You would find the correlation eventually if you collected enough data.
Well in this case we had ‘causation’ and were looking for correlation.
All of this is about opinions. I’m trying to explain to the guy that his opinions are not fact and actually most people, including the general playerbase, CCP and some dude on INN have different opinions to him.
Hence ‘speak for yourself’.
So is yours.
That’s what I’m trying to tell you. The difference being there is nothing to support your opinion.
Well yes.
I can agree with CCP sometimes and disagree with them at others. I’m not a sycophant.
But as the arbiters of the game it’s they who decide where a healthy ecosystem should be. And you can of course disagree, but try not to do it with dishonest statements like ‘pvp players are limited by isk’ or ‘cheap ships means more pvp’.
Cause it’s demonstrably not the case.
So you’re saying correlations (or lack there of) should be ignored?
I hypothesized causation with a logical argument. That is I said lower ISK generation and mineral collection would results is a lower number of ships destroyed. We were looking for evidence to support that argument, and I showed you what I consider to be evidence to back up my hypothesis.
Sure, but you can’t explain to me why my opinion is wrong, so…
No it’s no if you actually listen to what I’m saying.
No, what he is saying is you need to explain any correlations you find, or understand why the correlation you thought would be there isn’t. It’s quite funny that you think the correct thing to do is ignore data.
No, you may not. Correlation is ONLY linked to a dataset. You can create an infinite amount of dataset based on the same model that contains a causation, and never find a correlation.
They are not proof, just hint what you should work on. That’s the meaning of “a correlation is not a causation”.
That’s also what reminded CCP rise when he said “we need actionable event” : correlation bear no systematic meaning, and actually you can’t find causation by observation only. That’s why BlackOut was a good things, from a scientist point of view, and so are CCP with their scarcicity &co. Shoot the ant house and look how it recreates itself.
Note that you can make model based on the correlation observed, and they may actually have a predictive power good enough for a short range of event, but it does not mean they ARE correct, as when applied to events outside of the range of observations already made they lose any predictive power they had. Many times we don’t need to be 100% accurate to make predictions.
Causality is the area of statistics that is commonly misunderstood and misused by people in the mistaken belief that because the data shows a correlation that there is necessarily an underlying causal relationship
also
Therefore, every time someone’s argument uses correlations or the absence of, you know he is wrong.
Correlations, or their absence of, can never be used as a proof.
Only people with very bad statistical knowledge affirm the opposite . They make a special kind of fool when they then affirm to have a PhD, since that’s one of the most important thing you are warned against when doing a PhD. Like this one :
But you don’t understand how pointing out one year of correlation vs the other 16 years of not correlation might actually not be a correlation.
You’re trying to say there is a relationship between ship price and pvp despite those graphs showing no such thing, except for maybe one year?
You understand thats the argument you’re making? You’re sticking with that?
I quoted what you said earlier in this thread.
What are you saying if it isn’t ‘cheap ships means more pvp’?
Emphasis mine!
Not proof. But supporting evidence.
Can you say here and now that correlation is useless and should always be ignored or sometimes useful for indicating relationships (or lack there of) between two variables?
Correlation is the manifestation of two causally linked variables. If your dataset is both infinite and the observed variables are relevant, then the correlation must appear.
Nope, neither. It does not support anything at all. Actually, there is no correlation between correlations and causations.
They ARE meaningless, by their nature of depending on the observer, NOT on the reality. They however can help you search things.
You are completely wrong when you use correlations as proof or “supportive evidence”.
Nope.
That’s literally written in the linked I gave above :
Causality is the area of statistics that is commonly misunderstood and misused by people in the mistaken belief that because the data shows a correlation that there is necessarily an underlying causal relationship
This is just wrong. Even with homogeneous distributed observed variable (I believe that what you think for relevant), nothing prevents you from having an infinity of datasets which do not show a correlation for an existing causation.
You both should take lessons in statistics. Your discussion is sterile for the sole reason that you have no idea what you are talking about.
It’s actually the total period since rorquals were in the prenerfed state which I’m interested and the correlation is there to see, as I explained. If you think I’m talking about a 1 year period then you don’t even understand the argument, therefore, how can you claim I’m wrong.
No I’m not. I’m saying there is a relationship between ISK generation and production, and the amount of ships destroyed. I’m hypothesizing about the ship price relative to player purchasing power as the cause of why the relationship exists. You have yet to explain why my hypothesis is wrong.
That’s not how it works.
Just because your interpretation is not inconsistent, does not mean it’s true.
You are inverting the burden of proof.
You can make hypothesis. But then you need to evaluate when they are correct, and when they are wrong, with the same importance.
Of course not, but he is saying I am wrong, so obviously that means he can prove it, which is what I’m asking him to do. I’m under no illusion that I am 100% correct. Hence, why it’s a hypothesis.
What is asserted without a proof, is dismissed without a proof.
If you have no proof for your hypothesis, then your hypothesis is wrong without any further proof.
Your hypothesis may be consistent with the observed dataset, but being consistent is not enough to be considered a proof.
So when CCP nerfed wardecs based on correlation they were completely wrong?
And most of the scientific community is wrong about global warming?
Even then, you’re still saying the correlation only applies some of the time.
It is the equivalent of saying your hypothesis is correct some of the time.
Do you understand?
Even if this is what you were saying. They still don’t follow eachother.
Production and isk generation is in decline the past year, yet destruction has been steady or increased.
It’s not even consistent…
For some reason this is not widely accepted on the forum.
Inconclusive, not wrong.
Agreed. But the profession of statistical analysis is mostly about trying to explain your data and building up a case for your hypothesis. Then the strength of your hypothesis increases or decreases depending on new data and analyses. I’m not doing anything unreasonable by performing an analysis and discussing it.