Jump to content

Equal Map Vote Defaults To Alphabetical Selection


3 replies to this topic

#1 Bonger Bob

    Member

  • PipPipPipPipPipPip
  • Ace Of Spades
  • 373 posts
  • LocationPerth, WA

Posted 21 January 2016 - 09:25 PM

noticed with the map votes of late, a lot of equal vote % for maps are occurring, great in my opinion, but when it does happen it sorts the choice by alphabetically.

this often results in either canyon or frozen city as the default, as they are usually one of the two that end up with equally split votes on the select screen.

wouldn't be an issue if canyon wasn't so anti-assault mech, and frozen city has enough of of a selection bias from the lazor vomit mechs.

so yeah, if your holding on to the hope the equal split vote is going to random in your favor, it ain't.

its at least a factor for the mini-game, lol

pretty sloppy of PGI, you'd think a 50/50 "coin toss" would occur to balance it.

Edited by Bonger Bob, 21 January 2016 - 09:27 PM.


#2 Chuck Jager

    Member

  • PipPipPipPipPipPipPipPipPip
  • The 1 Percent
  • The 1 Percent
  • 2,031 posts

Posted 21 January 2016 - 11:25 PM

How do you know that what you saw on the client side is reflected by what the the servers said when you see the winner.

I am pretty sure that waiting until the last second to cast the vote means that some peoples vote may look like they got counted but because of "lag" the server did not get them in time or not everybody had an accurate view of the votes at the last second. Of course, I can not prove this either, but be aware that if one person is waiting till the absolute last second to cast their vote to game the system so are a few more. Do onto others etc etc.

I do know client is not always equal to server (ppc srm hit reg and warhorns). I am pretty sure this is the reason we do not see the finally tally on the next screen (or I am blind and stupid - been so in past), and why we do not have an in game combat log. I do not think the seas would ever be calm if the player base had an accurate measurement of what really gets lost or what actions/mechs really are the most OP for gaming the mechanics of the game.

#3 Bonger Bob

    Member

  • PipPipPipPipPipPip
  • Ace Of Spades
  • 373 posts
  • LocationPerth, WA

Posted 21 January 2016 - 11:46 PM

View PostChuck Jager, on 21 January 2016 - 11:25 PM, said:

How do you know that what you saw on the client side is reflected by what the the servers said when you see the winner.

I am pretty sure that waiting until the last second to cast the vote means that some peoples vote may look like they got counted but because of "lag" the server did not get them in time or not everybody had an accurate view of the votes at the last second. Of course, I can not prove this either, but be aware that if one person is waiting till the absolute last second to cast their vote to game the system so are a few more. Do onto others etc etc.

I do know client is not always equal to server (ppc srm hit reg and warhorns). I am pretty sure this is the reason we do not see the finally tally on the next screen (or I am blind and stupid - been so in past), and why we do not have an in game combat log. I do not think the seas would ever be calm if the player base had an accurate measurement of what really gets lost or what actions/mechs really are the most OP for gaming the mechanics of the game.


yup, PGI do like hiding the metrics from us...

but not sure thats the case, even if you factor people last minute switching into it, there is still going to be some occasions when the map vote is split equally between the highest two, and EVERY time its seems to pick based on alphabet.

watch when it happens, i seriously do hope i'm wrong, but it does seem slanted not randomized.

#4 Chuck Jager

    Member

  • PipPipPipPipPipPipPipPipPip
  • The 1 Percent
  • The 1 Percent
  • 2,031 posts

Posted 22 January 2016 - 12:56 AM

View PostBonger Bob, on 21 January 2016 - 11:46 PM, said:


yup, PGI do like hiding the metrics from us...

but not sure thats the case, even if you factor people last minute switching into it, there is still going to be some occasions when the map vote is split equally between the highest two, and EVERY time its seems to pick based on alphabet.

watch when it happens, i seriously do hope i'm wrong, but it does seem slanted not randomized.

Sometimes what appears to be a common sense coding error just got overlooked or the second part of the equation may not be triggering because somebody else renamed a variable or the trigger was replaced by another function somewhere else. This really is not PGIs fault it is just the way internet programming works, it took us how long to learn to make interchangeable parts in mechanical devices.

IF you have 24 players with probably 20-40% having more than 1 vote and another 10-20% not fighting over 2 maps and 30-50% trying to cast their vote at the last minute so they may not be seen on either the client or server side, I do not think there really is a way to know if there was a tie. Canyon network, Caustic Valley and HPG manifold are maps choose to vote against Terra therma and Viridian and Alpine comes up so rarely it is kinda an outsider. I do not think that anymore than 1 of 10 matches could be a tie, and then you would need to look at a huge sample of individual choices versus what got chosen.

I do applaud folks for saying "hey is this an issue" and I do not think we need to belittle folks for not have a "burning bush" for proof, but I just do not think it is possible to observe the results. I take medicines that really mess with my sleep so I play at many different times. I have noticed that the map choices are grouped very differently based on this and day of week as well as the tier of my main and alt accts. You may be encountering this and the groupings may also happen to be alpha ordered.

It really is game, with the only identifiable given being human selfishness or individual free will.

OFF SUBJECT - Most of my background is in web design and the front end design/programming, so I am used to tripping over anything and everything. I do understand why PGI hides their metrics, it is the same reason I do not go nude in public. It really does not help the situation. Many of the issues with online "whatever" are not always easily identifiable or fixable and sometimes just not worth the potential hassle. Remember "What you say can and will be held against you" and "no good deed goes unpunished".

It took me a long time to realize that even though many companies are greedy and would sell their young for a buck, they do have a financial stake in how they treat their customers. Many customers are the same greedy jerks who act just like their selfish corporate brethren, but they are no longer hindered by a huge loss if they act poorly.





1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users