Jump to content


Higher Skilled Players Like Artillery Less - Confirmed, But Why?

artillery

  • Please log in to reply
1650 replies to this topic

XJCL #41 Posted May 03 2014 - 23:00

    Major

  • Players
  • 17480 battles
  • 7,197
  • Member since:
    08-26-2012
Could you post the WR graph (or recent WR) in addition to the WN8 graph? I believe WN8 was purposefully constructed to stretch out stats over blue for greater differentiation, which could have skewed your P value. I think Rommel pretty much said everything regarding possible reasons for the trend though. 

Edited by XJCL, May 03 2014 - 23:00.


_Intrepid #42 Posted May 03 2014 - 23:00

    First lieutenant

  • Players
  • 9878 battles
  • 927
  • Member since:
    06-26-2012

View PostCinphul, on May 03 2014 - 23:51, said:

You are dishonest.  If you're going to present data, you present all of the data you said you would provide at the same time, not one piece of it plus your already determined conclusion.  That's idiotic.

 

I do not care what stupid, self serving excuse you have for cherry picking the one you want, then immediately drawing your conclusion.  That is pure dishonest.  Deny all you want - you presented a plan and ignored it because you wanted to post your dishonest conclusion, which everyone knew was exactly the purpose of your original poll.

 

But go ahead and continue to blame me for not accepting your dishonesty, which has been pervasive throughout all of your whiny anti-arty hissy fit posts - clearly that's the only tactic you have left because you can't defend the garbage you post here.

 

I'm blaming you for your angst and anger over something that didn't happen. If my posting the WN8 conclusion first so offends you, well, you'll just have to get over it. There is no lie there. If it makes you feel any better, I have the Winrate data already, I just didn't have time to make the chart presentable. The p-value for Recent Winrate and Vote is .00000000145, in other words, undeniably related. You're just frothing at the mouth any time I post, that's not my problem and not something I can fix for you.

 

View PostCinphul, on May 03 2014 - 23:51, said:

Why should anyone else be responsible for your data?  You are the definition of slimy.

 

They're not, but when I provide the data in a free format, one anyone can grab an app to view, I've done the required first step. Putting it up in everyone's preferred format is not something I'm obligated to do the very second the post goes up.

 

You can save the slurs/insults/ad homs, they affect me none at all and really only risk your being moderated. 



Cinphul #43 Posted May 03 2014 - 23:00

    Major

  • Players
  • 36488 battles
  • 5,395
  • [Y0L0] Y0L0
  • Member since:
    07-07-2012

View Postaqollo, on May 03 2014 - 17:56, said:

 

You have to realize that not everything is linear. This can be a quadratic relationship and crash down at the end but that is not plotted because of lack of sample size. 

 

He doesn't want to realize - he wants to draw a nice straight line through an otherwise very volatile, small sample of data.  Because like Bernie Madoff knows all to well, straight lines are deceptive. 



Aethon #44 Posted May 03 2014 - 23:01

    Staff sergeant

  • Beta Testers
  • 15198 battles
  • 400
  • [-CHH-] -CHH-
  • Member since:
    07-09-2010

View Postaqollo, on May 03 2014 - 17:59, said:

 

Arty tracers are still ingame...

 

I said 'old arty tracers'.  I was referring to the ones in beta and early live gameplay, where counter-battery was common.  Most of the time, arty just ignore each other these days.



Cinphul #45 Posted May 03 2014 - 23:01

    Major

  • Players
  • 36488 battles
  • 5,395
  • [Y0L0] Y0L0
  • Member since:
    07-07-2012

View PostXJCL, on May 03 2014 - 18:00, said:

Could you post the WR graph (or recent WR) in addition to the WN8 graph? I believe WN8 was purposefully constructed to stretch out stats over blue for greater differentiation, which could have skewed your P value. I think Rommel pretty much said everything regarding possible reasons for the trend though. 

 

He posted only the data that most emphasises his point.  He does not want to post a flat line.



_Intrepid #46 Posted May 03 2014 - 23:05

    First lieutenant

  • Players
  • 9878 battles
  • 927
  • Member since:
    06-26-2012

View PostDonward, on May 03 2014 - 23:53, said:

A sample size of 193 votes for a player population size in the millions has a shockingly large error rate approaching the double digits. Adding to the issue is that these forums are a poor way of measuring the playerbase's attitude to arty in general. It would be like polling the town of Hayden Lake, Idaho and using those numbers to extrapolate America's views on race relations. So you have a poor sample size and poor sample subject.

 

All your results show is that the anti-arty war has slowly come to an end and your side has lost. Even in the WoT forums - once a fever swamp for arty hate - you were unable to garner more than 1/3 of the votes to remove arty.

 

But do go on with your delusions that good players hate arty and baddies love it because - herpa durr - they can't play "real" tanks.

 

Actually, no. If we're generous and say that peak concurrents in NA is around 50,000, and give them a 20x multiple for active users, our "population" is 1,000,000. We'd need 384 votes for a 5% margin of error with 95% confidence. I have 375. Also, no matter who voted for what, it didn't matter. That is what was so amusing. Some people kept thinking that by voting for "I like artillery" they were destroying the poll and ruining the outcome, when in fact they were just helping gather data. The results could have been 90% pro artillery and still this regression analysis would have been meaningful because it doesn't matter what % voted for what. This measures the relationship between your stats and your vote, it doesn't care at all what % voted for what and that does not impact the results.



Cinphul #47 Posted May 03 2014 - 23:08

    Major

  • Players
  • 36488 battles
  • 5,395
  • [Y0L0] Y0L0
  • Member since:
    07-07-2012

View Post_Intrepid, on May 03 2014 - 18:00, said:

I'm blaming you for your angst and anger over something that didn't happen. If my posting the WN8 conclusion first so offends you, well, you'll just have to get over it. There is no lie there. If it makes you feel any better, I have the Winrate data already, I just didn't have time to make the chart presentable. The p-value for Recent Winrate and Vote is .00000000145, in other words, undeniably related. You're just frothing at the mouth any time I post, that's not my problem and not something I can fix for you.

 

There you go with yet more dishonesty - trying to demonize me because you can't argue against what I've said.  Ad hominem attacks do not work, and calling me angry is a childish attempt at distraction from your dishonest statements.

 

Quote

They're not, but when I provide the data in a free format, one anyone can grab an app to view, I've done the required first step. Putting it up in everyone's preferred format is not something I'm obligated to do the very second the post goes up.

 

There are many free file share sites that do not require the person install EXE garbage onto their machine.  Learn to google, the internet isn't complicated.

 

Quote

You can save the slurs/insults/ad homs, they affect me none at all and really only risk your being moderated. 

 

Stop being dishonest and I'll stop calling you such.  As for ad homs, did you already forget that you are in the one that does that - you are the one that calls out people's stats, then turns around and accuses them of being first to sling mud.  You are the one that leaves out your entire stats analysis from your original post, then turns around and starts accusing others of not reading it until woops, you figure out once again you are wrong.  How many times do you have to apologize to me for your mistakes before you get your head out of your butt?  We're at 3 already.



wh1skybarrel #48 Posted May 03 2014 - 23:09

    Major

  • Players
  • 21454 battles
  • 7,644
  • [RELIC] RELIC
  • Member since:
    08-22-2011

View Post_Intrepid, on May 03 2014 - 18:05, said:

 

Actually, no. If we're generous and say that peak concurrents in NA is around 50,000, and give them a 20x multiple for active users, our "population" is 1,000,000. We'd need 384 votes for a 5% margin of error with 95% confidence. I have 375. Also, no matter who voted for what, it didn't matter. That is what was so amusing. Some people kept thinking that by voting for "I like artillery" they were destroying the poll and ruining the outcome, when in fact they were just helping gather data. The results could have been 90% pro artillery and still this regression analysis would have been meaningful because it doesn't matter what % voted for what. This measures the relationship between your stats and your vote, it doesn't care at all what % voted for what and that does not impact the results.

 

Are you one of those people that argues every point in a conversation?  Devils advocate kind of thing?  I have friends like that.  So tiresome!  I can relate to your nature.  :smile:



_Intrepid #49 Posted May 03 2014 - 23:10

    First lieutenant

  • Players
  • 9878 battles
  • 927
  • Member since:
    06-26-2012

View Postaqollo, on May 03 2014 - 23:56, said:

You have to realize that not everything is linear. This can be a quadratic relationship and crash down at the end but that is not plotted because of lack of sample size. 

 

Rather than debating back and forth, I just ran one with 2 3500 players voting "I love artillery.". P-value changed to .0000000547. The slope nearly unchanged. The effect you're imagining is not there, not with even 195 votes, the confidence and reliability is so beyond the risk of randomness affecting it (several decimal places removed) that a few values won't shift it.



Cinphul #50 Posted May 03 2014 - 23:11

    Major

  • Players
  • 36488 battles
  • 5,395
  • [Y0L0] Y0L0
  • Member since:
    07-07-2012

View Post_Intrepid, on May 03 2014 - 18:05, said:

 

Actually, no. If we're generous and say that peak concurrents in NA is around 50,000, and give them a 20x multiple for active users, our "population" is 1,000,000. We'd need 384 votes for a 5% margin of error with 95% confidence. I have 375. Also, no matter who voted for what, it didn't matter. That is what was so amusing. Some people kept thinking that by voting for "I like artillery" they were destroying the poll and ruining the outcome, when in fact they were just helping gather data. The results could have been 90% pro artillery and still this regression analysis would have been meaningful because it doesn't matter what % voted for what. This measures the relationship between your stats and your vote, it doesn't care at all what % voted for what and that does not impact the results.

 

Nobody thought they were destroying the data by voting - that is simply a willfully ignorant statement.  Why do you keep posting such dishonest garbage?  They were making the point that it doesn't matter what garbage you were trying to prove, you still had 2/3s of the people responding that they were fine with arty, so it doesn't matter who likes it and who doesn't.  Are you really so clueless that you thought anyone thought they were destroying the data?  Or are you just so arrogant that you think that nobody else can look at the data in any other way but that which you determine?  Get over yourself - we see the for and against results just fine, and can draw our own conclusion just fine from it.  



Quaffer #51 Posted May 03 2014 - 23:12

    Staff sergeant

  • Players
  • 5048 battles
  • 372
  • Member since:
    04-22-2012

View PostQuaffer, on May 02 2014 - 20:58, said:

 

I choose to differ.  There are very good players who have an intense hatred for SPG s.  There are very good players that have a great affection for SPG s as well.  There are poor players that have never played it and hate it because they want to fit in and Arty haters are very vocal so they stand out.  Run with that pack boys.  Truth is, where ever you look in the skill spectrum you will find folks on both sides of the discussion. You will also find a lot of people who don't express an opinion because they don't care.  It's a complex issue and I have yet to read the post that can sum it all up with a pretty bow.

 

A word about this particular thread.  _Intrepid is trying to build up a database by comparing the skill levels of the respondents with how they voted in the poll.  I expect he is hoping to say "See the people who are good don't like it and the people who are bad do.  Make Arty go away for the sake of the skilled players."  Well that's my take. 

 

I find it unreasonable to measure an SPG driver by the same metrics as you measure a heavy tank, etc. I've been in matches where I have done significantly more damage than people who earned more experience than I did. There are a lot of ways to get experience that most SPG s don't do in the game,  spotting, capping, what have you. Arty marches to a different drummer.  As long as they are judged the same way as the rest of the tanks they are going to be out of step.

 

Keep your linkage clean.

 

Hey, can I call them or not?

 

Something to consider.  The game is for those who play.  It is not just for those that rate high in the statistics tracked by XVM.

 

Keep your optics clean.



Donward #52 Posted May 03 2014 - 23:13

    Major

  • Players
  • 46957 battles
  • 7,083
  • [C-BOO] C-BOO
  • Member since:
    08-29-2011

View Post_Intrepid, on May 03 2014 - 15:05, said:

 

Actually, no. If we're generous and say that peak concurrents in NA is around 50,000, and give them a 20x multiple for active users, our "population" is 1,000,000. We'd need 384 votes for a 5% margin of error with 95% confidence. I have 375. Also, no matter who voted for what, it didn't matter. That is what was so amusing. Some people kept thinking that by voting for "I like artillery" they were destroying the poll and ruining the outcome, when in fact they were just helping gather data. The results could have been 90% pro artillery and still this regression analysis would have been meaningful because it doesn't matter what % voted for what. This measures the relationship between your stats and your vote, it doesn't care at all what % voted for what and that does not impact the results.

Wow. You really don't understand polling or analogies. Any poll size under the 375 arbitrary number makes the error rate go up exponentially. You just can't jigger the numbers and pretend the outcome works out right. And once again you error in your sample subject - the forums - and conflating them with the general playerbase. 

 

And finally, you were only able to dredge up - what - 60 or 70 remove arty votes total. In the old days that would be the "bacon" choice for any arty poll.

 

Give it up. Your side has lost. 



_Intrepid #53 Posted May 03 2014 - 23:14

    First lieutenant

  • Players
  • 9878 battles
  • 927
  • Member since:
    06-26-2012

View PostCinphul, on May 04 2014 - 00:01, said:

He posted only the data that most emphasises his point.  He does not want to post a flat line.

 

As usual, wrong. I just haven't had time to do the color overlay. (I had to use whole numbers as the recent numbers in your signature from WotLabs are not displayed as a number with decimals in the sig or anywhere actually (60D is close, but not the same)).

 



aqollo #54 Posted May 03 2014 - 23:14

    Major

  • Players
  • 21725 battles
  • 3,327
  • Member since:
    02-21-2011

View PostCinphul, on May 03 2014 - 16:11, said:

 

Nobody thought they were destroying the data by voting - that is simply wrong.  Why do you keep posting such dishonest garbage?  They were making the point that it doesn't matter what garbage you were trying to prove, you still had 2/3s of the people responding that they were fine with arty, so it doesn't matter who likes it and who doesn't.  Are you really so clueless that you thought anyone thought they were destroying the data?  Or are you just so arrogant that you think that nobody else can look at the data in any other way but that which you determine?  Get over yourself - we see the for and against results just fine, and can draw our own conclusion just fine from it.  

 

I sent you a msg on the forum regarding this.

Also intreped. do other regressions other than linear and show me the graph pls.



_Intrepid #55 Posted May 03 2014 - 23:16

    First lieutenant

  • Players
  • 9878 battles
  • 927
  • Member since:
    06-26-2012

View PostCinphul, on May 04 2014 - 00:08, said:

Rawrrrr stuff

 

Not going to do this with you. Post like an adult, we can discuss it. Post things like this and I'm going to ignore it to avoid a derail.



Evocoot #56 Posted May 03 2014 - 23:16

    Captain

  • Players
  • 7974 battles
  • 1,388
  • Member since:
    12-03-2013
Since its an open poll I assume anyone questioning his presentation of results could go ahead and backcheck them. It would seem a fairly simple conclusion that those who rely least on randomness would be those who least like the most random of classes, I'd have probably agreed to the op's hypothesis before seeing data to back it up. Questioning his research methodology without actually presenting anything to prove he's shady is just as shady as some are accusing him of being.

DarkStar80 #57 Posted May 03 2014 - 23:17

    First lieutenant

  • Players
  • 13278 battles
  • 983
  • [CBC] CBC
  • Member since:
    01-15-2013

View Postaqollo, on May 03 2014 - 14:25, said:

The possible flaw with your graph is that good players are rarer than bad players and uneven distribution are expected.

 

That's true in the general wot population, but on these forums?  I would imagine there are proportionally more "good" players on the forums than there are in the general game population



_Intrepid #58 Posted May 03 2014 - 23:18

    First lieutenant

  • Players
  • 9878 battles
  • 927
  • Member since:
    06-26-2012

wh1skybarrel, on May 04 2014 - 00:09, said:

Are you one of those people that argues every point in a conversation?  Devils advocate kind of thing?  I have friends like that.  So tiresome!  I can relate to your nature.  :smile:

 

Haha... not quite. I discuss something until I'm bored, and answer anything that seems either like they put in effort or is worth answering. I took the time to make the opening post, why not take a little more to reply? :)



XJCL #59 Posted May 03 2014 - 23:18

    Major

  • Players
  • 17480 battles
  • 7,197
  • Member since:
    08-26-2012

IDK why ppl are getting so worked up about this, while interpretations of the data may differ, the data itself is somewhere between "mmkay" and "well yeah." 

 

 

 



_Intrepid #60 Posted May 03 2014 - 23:19

    First lieutenant

  • Players
  • 9878 battles
  • 927
  • Member since:
    06-26-2012

View PostEvocoot, on May 04 2014 - 00:16, said:

Since its an open poll I assume anyone questioning his presentation of results could go ahead and backcheck them. It would seem a fairly simple conclusion that those who rely least on randomness would be those who least like the most random of classes, I'd have probably agreed to the op's hypothesis before seeing data to back it up. Questioning his research methodology without actually presenting anything to prove he's shady is just as shady as some are accusing him of being.

 

Just in case someone does this, and a bunch of people got together and changed their votes to muck with me (hard to believe anyone would bother... but I have a small "fan club" that has a hard on for me here), I took screenshots of the people that voted for each option and saved it. I can put those up in a .zip somewhere too.

 

100% transparency.







Also tagged with artillery

1 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users