Jump to content

Worthy Adversaries - What Can Explain This?


112 replies to this topic

#101 Mystere

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 22,783 posts
  • LocationClassified

Posted 18 August 2014 - 03:31 PM

View PostVassago Rain, on 18 August 2014 - 02:42 PM, said:

This last one, for some reason, was a proper test ...


Seriously? Testing random players during odd hours without the proper controls was a proper test? Really?



Posted Image

And what happened to all those claims that balancing should start from the level of the so-called "elite" "top level" "high Elo" "competitive" players? Suddenly that is no longer valid? Using random PUGs is better?

Thank you for giving me the best laugh of the day.

And in return, I am leaving this:

View PostMystere, on 18 August 2014 - 08:53 AM, said:

Real science is hard. That is why politicians and other folk just settle for pseudoscience.

:)

Edited by Mystere, 18 August 2014 - 03:36 PM.


#102 Henree

    Member

  • PipPipPipPipPipPipPip
  • FP Veteran - Beta 1
  • 501 posts

Posted 18 August 2014 - 03:41 PM

worth of mech you have taken down minus damage received and done to said mech equals score.

#103 Lefty Lucy

    Member

  • PipPipPipPipPipPipPipPipPip
  • Bridesmaid
  • Bridesmaid
  • 3,924 posts
  • LocationFree Tikonov Republic

Posted 18 August 2014 - 04:39 PM

View PostMystere, on 18 August 2014 - 03:31 PM, said:

Seriously? Testing random players during odd hours without the proper controls was a proper test? Really?


If you knew anything about statistics, then yes, you want your sample to be as random as possible while still allowing the variable you're testing for to exist.

#104 Angel of Annihilation

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The Infernal
  • The Infernal
  • 8,881 posts

Posted 18 August 2014 - 04:47 PM

View PostAresye, on 16 August 2014 - 11:40 PM, said:


The Mad Cat is better than any IS heavy, so that plays a large role, but damage also plays a big factor.

An IS mech that gets 8 kills is going to do considerably less overall damage than a Clan mech that gets 8 kills. Now since part of the score is influenced by damage (damage - team damage / 15), Clan mechs will overall get more points for their damage done than their IS counterparts.


Yeah to be honest your right. While I don't think the Timber Wolf is all that and a bag of chip like most people seem to, I have to admit if we are just looking at heavies, then yes, the Timber Wolf is the best one.

Also your right about damage. IS mechs in general are better at killing the enemy due to alot more pin point damage with its ACs plus the shorter pulses on all its lasers. However, Clan mechs tend to do more damage for those same amount of kills. Any formula that use damage amounts is going to be skewed toward Clans.

Edited by Viktor Drake, 18 August 2014 - 04:47 PM.


#105 Mystere

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 22,783 posts
  • LocationClassified

Posted 18 August 2014 - 05:02 PM

View PostLefty Lucy, on 18 August 2014 - 04:39 PM, said:

If you knew anything about statistics, then yes, you want your sample to be as random as possible while still allowing the variable you're testing for to exist.


That might work for polls, consumer preferences, or probably even for testing the safety of guns to make sure they're as idiot proof as reasonably possible.

But, that method does not work for comparing a Russian SU-27 to a French Rafale.

#106 JudgeDeathCZ

    Member

  • PipPipPipPipPipPipPipPip
  • The Defiant
  • The Defiant
  • 1,929 posts

Posted 18 August 2014 - 05:02 PM

View PostMalineczka, on 17 August 2014 - 02:20 AM, said:

Posted Image

Spoiler

Nice carry :) .
I have record only about 1,2k in Timby ;) .

#107 Lefty Lucy

    Member

  • PipPipPipPipPipPipPipPipPip
  • Bridesmaid
  • Bridesmaid
  • 3,924 posts
  • LocationFree Tikonov Republic

Posted 18 August 2014 - 05:08 PM

View PostMystere, on 18 August 2014 - 05:02 PM, said:


That might work for polls, consumer preferences, or probably even for testing the safety of guns to make sure they're as idiot proof as reasonably possible.

But, that method does not work for comparing a Russian SU-27 to a French Rafale.


It works great for comparing a large number of SU-27s to a large number of Rafales.

#108 Mycrus

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Overlord
  • Overlord
  • 5,160 posts
  • Google+: Link
  • Twitter: Link
  • LocationFilipino @ Singapore

Posted 18 August 2014 - 05:33 PM

View PostMystere, on 18 August 2014 - 03:31 PM, said:


Seriously? Testing random players during odd hours without the proper controls was a proper test? Really?



Posted Image

And what happened to all those claims that balancing should start from the level of the so-called "elite" "top level" "high Elo" "competitive" players? Suddenly that is no longer valid? Using random PUGs is better?

Thank you for giving me the best laugh of the day.

And in return, I am leaving this:



Wait you mean pgi telemetry is pseudowatsit

#109 Joseph Mallan

    ForumWarrior

  • PipPipPipPipPipPipPipPipPipPipPipPipPipPipPip
  • FP Veteran - Beta 1
  • FP Veteran - Beta 1
  • 35,216 posts
  • Google+: Link
  • Facebook: Link
  • LocationMallanhold, Furillo

Posted 18 August 2014 - 05:40 PM

View PostMycrus, on 18 August 2014 - 05:33 PM, said:

Wait you mean pgi telemetry is pseudowatsit

Nope. It's Scutchi Fatchi. pseudowatsit is a psychological term for we're making this up. :lol:

#110 Tyrolus Lucien

    Member

  • PipPip
  • 27 posts

Posted 18 August 2014 - 06:20 PM

What can possibly explain the fact that Celyth has 2617pts on his IS heavy, and 2635pts on his Clan heavy? TW...that must be the explanation!

Err...wait...the difference is only 18pts? Inconceivable!

#111 RussianWolf

    Member

  • PipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 2,097 posts
  • LocationWV

Posted 18 August 2014 - 06:39 PM

View PostLefty Lucy, on 18 August 2014 - 05:08 PM, said:


It works great for comparing a large number of SU-27s to a large number of Rafales.

So 90% of all players will win when running a Timberwolf???

No, A bad player will lose regardless of mech.

In statistics if you are trying to find the correlation of one variable, you try to eliminate or minimize all other variables. Then you get reliable data. You might luck into a tendancy with random variable, but you won't get reliable data.

Say for example, you won't to find out which car lineup gets better fuel economy. You use the same drivers and the same track, driving the same way (heavy foot or soft foot). And you get reliable data.

If you take random drivers over different tracks with no concern for driving habits, you get Corvettes showing with better milage than a Prius (hard to believe but conceivable). You might get to the right conclusion, but you can't trust the data.

#112 Mystere

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 22,783 posts
  • LocationClassified

Posted 18 August 2014 - 06:44 PM

View PostRussianWolf, on 18 August 2014 - 06:39 PM, said:

So 90% of all players will win when running a Timberwolf???

No, A bad player will lose regardless of mech.

In statistics if you are trying to find the correlation of one variable, you try to eliminate or minimize all other variables. Then you get reliable data. You might luck into a tendancy with random variable, but you won't get reliable data.

Say for example, you won't to find out which car lineup gets better fuel economy. You use the same drivers and the same track, driving the same way (heavy foot or soft foot). And you get reliable data.

If you take random drivers over different tracks with no concern for driving habits, you get Corvettes showing with better milage than a Prius (hard to believe but conceivable). You might get to the right conclusion, but you can't trust the data.



Well, Lefty did seem to imply expertise in statistics, and is applying that same expertise in advanced weapons systems testing.











:lol: :P :lol:

Edited by Mystere, 18 August 2014 - 06:49 PM.


#113 Malcolm Vordermark

    Member

  • PipPipPipPipPipPipPipPipPip
  • 2,520 posts

Posted 18 August 2014 - 06:45 PM

View PostTyrolus Lucien, on 18 August 2014 - 06:20 PM, said:

What can possibly explain the fact that Celyth has 2617pts on his IS heavy, and 2635pts on his Clan heavy? TW...that must be the explanation!

Err...wait...the difference is only 18pts? Inconceivable!


It is an interesting data point and it would be interesting to see how all the clan players would perform in an IS mech of the same class. However, it is not practical for everyone to do so and by itself this is just a single point of data that is not sturdy enough to base an argument on.





1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users