Jump to content

No Guts No Galaxy Podcast #170 W/ Justcallme A S H (Vod)


157 replies to this topic

#1 Sean Lang

    Member

  • PipPipPipPipPipPipPip
  • The Wolf
  • The Wolf
  • 969 posts
  • Facebook: Link
  • Twitter: Link
  • Twitch: Link
  • LocationNorth Carolina

Posted 08 January 2019 - 10:14 AM



#2 Bombast

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 7,709 posts

Posted 08 January 2019 - 12:03 PM

Everytime I see someone use the term 'vod,' I think about those old 'Bod' deodorant commercials.

#3 So You Say

    Member

  • PipPipPip
  • The Angel
  • 75 posts

Posted 08 January 2019 - 01:58 PM

Regarding the side torso destruction issue. I think there should be some kind of penalty for loosing a side torso based on the number of heat sinks that torso contains and how much heat is currently in those sinks. Maybe change the effect from one that adds heat to the mech to one that instead damages CT structure. When running one you save the weight of running a standard and you put in more heat sinks which allows you to ride the heat scale for more damage. By having another downside to their use IS XL's and standard engines would have more of a place in the game.

#4 Funk1777

    Member

  • PipPipPipPipPip
  • The Undertaker
  • The Undertaker
  • 104 posts

Posted 08 January 2019 - 03:41 PM

I liked Ash's thing about trying to rebalance teams after the MM kicked in. 99% players are way better and carry so much harder than 90% players to where these guys should be a whole separate tier.

Isn't the MM issue an easy fix? How is it ok that T1 drop against t3 by default? t1 vs t3?! The best players in the world routinely having guys that average 150 match score?! How is it ok that performing poorly will get you into t1 anyway as you play more games. Maybe you guys addressed this more as I skipped to the mm part and didnt stay long but couldnt PGI fix it in two simple steps.

The fix:
1. Make it so there are only T1 , T3 and sometimes T5 games when enough newbies warrant it. Allow the +1 level by default by only open the floodgates for the +2 after 2mins average wait time. Never open the T1 floodgates more than accepting t2.
If it was weighted that t2 and t4 filled in last wouldnt most t1 matches just be that and t3 be a mix of t3 and t2 by default. This leaves the t4 guys to help fill in the newbie matches for the most part but when pop is low it'll be t3-t5 matches and t1-t2 matches running with only a few outliers.

Wait time is better than feeling so horribly outmatched or sticking the struggling players in with pros to get hammered all the time.


2. Make PSR decrease and increase like it does during those first 25 games always. A t2 guy should hit t1 on a day he is playing a good mech and doing well and should drop like mad for doing poorly in a few games in a row.

I still think T1 should only be like the 95% players and the bulk of players should be around t3.

I spend like 80% of my time in MWO buying new mechs and just running them to skill them up. When they get mastered I sell the bad ones and buy more. Surely there are more people around like me that dont want to have to run their best mechs just to be able to try and hang with the great guys i see in MM all the time now. I find myself having to switch to better mechs that I have skilled when the pop gets lower at off times or I just get smoked.

Also, screw ATM's... that is all.

#5 justcallme A S H

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • CS 2020 Referee
  • CS 2020 Referee
  • 8,987 posts
  • LocationMelbourne, AU

Posted 08 January 2019 - 04:37 PM

View PostFunk1777, on 08 January 2019 - 03:41 PM, said:

Isn't the MM issue an easy fix? How is it ok that T1 drop against t3 by default? t1 vs t3?! The best players in the world routinely having guys that average 150 match score?! How is it ok that performing poorly will get you into t1 anyway as you play more games. Maybe you guys addressed this more as I skipped to the mm part and didnt stay long but couldnt PGI fix it in two simple steps.


It was covered in great detail.

And T1 to T3 hard locked to 2 Tier differential at a max - as the default is outlined -> HERE <- you really need to go and have a read of that one.

The issue is that someone who should be in Tier 4, can climb up to Tier 1.

Fix that you have gone a fair way to increase the quality. The 2nd step is then doing a secondary sort once 24 are found, granted that might not always be possible or work out - but it could again offer some improvement and that is all you can look for here.

View PostFunk1777, on 08 January 2019 - 03:41 PM, said:

The fix:
1. Make it so there are only T1 , T3 and sometimes T5 games when enough newbies warrant it. Allow the +1 level by default by only open the floodgates for the +2 after 2mins average wait time. Never open the T1 floodgates more than accepting t2.
If it was weighted that t2 and t4 filled in last wouldnt most t1 matches just be that and t3 be a mix of t3 and t2 by default. This leaves the t4 guys to help fill in the newbie matches for the most part but when pop is low it'll be t3-t5 matches and t1-t2 matches running with only a few outliers.


That is, essentially, how it works already.

#6 Funk1777

    Member

  • PipPipPipPipPip
  • The Undertaker
  • The Undertaker
  • 104 posts

Posted 08 January 2019 - 04:57 PM

View Postjustcallme A S H, on 08 January 2019 - 04:37 PM, said:


That is, essentially, how it works already.


Maybe but T3 drops against t1 all the time after a wait time of about 3 seconds for the matchfinder. Seems like if that is how it is supposed to be working something is broken.

My point is that a t3 game should never have t1 and T1 games should never have t3 in it. Do that and add that people get dropped down when performing poorly and it balances out. Yeah wait times suffer but match quality should improve.

I watched a cyclops loaded with lrm 10's and 5's and nothing else stand in the middle of the tallest buildings in river city firing into them repeatedly. He was "trying". He was one of the last guys on our team to die in what was a close game and he did 9 damage. You should never see that in a t1 game.

Edited by Funk1777, 08 January 2019 - 05:01 PM.


#7 justcallme A S H

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • CS 2020 Referee
  • CS 2020 Referee
  • 8,987 posts
  • LocationMelbourne, AU

Posted 08 January 2019 - 05:44 PM

You clearly have not read the thread I linked.

If you think it is possible for 12 T1 vs 12 T1 to occurr, you are absolutely dreaming. That will never happen / be possible as the population cannot support it. Looking at the top 100 pilots on Jarls, all of them are spread across 3 different timezones. At least 15 of the players I have never actually seen in game because of this.

You are literally asking for the possible because at the same time:
  • They are all online at the very same time in a 24hr day / 7 day week / 30 day month.
  • None of them are currently in game / match
  • They are all searching the same game mode Solo QP - So no Group / 7 Solaris Queue / Faction Scout or Invasion / Private Lobby
  • They all fit into the 3/3/3/3 MM requirement
  • They all have the same servers ticked
Do you now see how unreasonable what you are saying is yet? What you are saying is simply impossible to achieve. Watch the podcast again and look at the bell-curve graph and the explaination.


T1 players will never find a game against 23 other T1 users, ever. It is impossible to ever occur. PGI understands this, I undertsand this - the bulk of the playerbase also understands this from what I've read.

It will always need to expand to T2 and then T3 before it matches T1s, so doing it from the outset is absolutely the smart way. In that way Paul/PGI have it absolutely spot on.

Edited by justcallme A S H, 08 January 2019 - 05:45 PM.


#8 Funk1777

    Member

  • PipPipPipPipPip
  • The Undertaker
  • The Undertaker
  • 104 posts

Posted 08 January 2019 - 06:17 PM

I read yours, but Im not sure you read mine. I didnt say anywhere for t1 vs t1. I'm not sure where you got that. I said t1 and t2 should be matched up. T3-T5 should be matched up

The 3/3/3/3 weight balance is opened up after a short while so it still isnt an issue and I was talking about solo only. The other 2 queues are mostly dead and will stay that way with a low pop but fixing what the majority of players are playing seems like a solid idea.

You keep saying everyone is in T1 that shouldnt be but then say there arent enough T1 players.

The bell graph in the video is based on match score and the Tier numbers at the top dont mean anything currently. They are literally made up and show that the average match score is just X amount. Those smack dab in the middle of that graph are what maybe should be t3 but they are every tier with most being t1 and t2. If they arent then they are people that dont play that much and won't affect the MM much anyway.

Edited by Funk1777, 08 January 2019 - 06:22 PM.


#9 justcallme A S H

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • CS 2020 Referee
  • CS 2020 Referee
  • 8,987 posts
  • LocationMelbourne, AU

Posted 08 January 2019 - 06:21 PM

View PostFunk1777, on 08 January 2019 - 06:17 PM, said:

I read yours, but Im not sure you read mine. I didnt say anywhere for t1 vs t1. I'm not sure where you got that. I said t1 and t2 should be matched up. T3-T5 should be matched up

The 3/3/3/3 wieght balance is opened up after a short while so it still isnt an issue and I was talking about solo only. The other 2 queues are mostly dead and will stay that way with a low pop but fixing what the majority of players are playing seems like a solid idea.

You keep saying everyone is in T1 that shouldnt be but then say there arent enough T1 players.


View PostFunk1777, on 08 January 2019 - 04:57 PM, said:

My point is that a t3 game should never have t1 and T1 games should never have t3 in it


They must have T3 in T1 games. Simply due to population spread. Even just T1 and T2 together, will never be enough to make a bell curve system work. It is impossible to avoid, impossible.

Right now - Some T3 pilots are better than T1 - because PSR is just a EXP/Grind bar. So excluding T3 from T1 games currently would not work either.

Under the proposed/idea - T1 population would be slashed to the top say, 5% which is all of 1,000 players out of 28,000 currently active. Not like 50% (rough estimate) that it is now. 5% of the playerbase - fits into what I said above - never gonna be on @ the same time to fill a match. It will always need to expand.

I'm really at a loss as to how this isn't making sense.

Edited by justcallme A S H, 08 January 2019 - 06:44 PM.


#10 John Bronco

    Member

  • PipPipPipPipPipPipPip
  • The Fighter
  • The Fighter
  • 966 posts

Posted 08 January 2019 - 06:24 PM

Yeah, the secondary adjustment that Ash mentioned is key. Really hate looking at the team before the match and seeing several players, of well, noted skill and seeing nobody of any note on my side. You just know its a loss. Doing some sort of sorting on average MS or W/L is a must.

#11 Funk1777

    Member

  • PipPipPipPipPip
  • The Undertaker
  • The Undertaker
  • 104 posts

Posted 08 January 2019 - 06:51 PM

View Postjustcallme A S H, on 08 January 2019 - 06:21 PM, said:



I'm really at a loss as to how this isn't making sense.


I see what you are saying but you cant assume that every player plays the same amount. Literally saying that a t3 player is better than a t1 but doesnt play as much explains the flaw. If the top 30% of players are in t1 and t2 and they face each other only for the most part, you can be pretty damn sure these are the dudes who play all the time.

Im just saying you can't base it on the assumption that everyone plays the same amount. Im pretty sure those top 10% play a hell of a lot more than the others with a few exceptions. They play way more than t3 bob who has a 97% rating but doesnt play much so he is t3 or t2.

You dont think match quality would be better if the top 30% faced each other instead of getting 50% and lower guys thrown in. That is literally what T3 is/should be 40%-60% players. Expanding that to be 30% to 70% causes its own problems but it keeps them out of the top matches more often than not.

I do know you are in the aussie time zone so the populations could be way different than what I see regularly in NA.

Dont get me wrong but I love the idea of a second balancing so one team isnt stacked with 99% guys and the other has 80% guys in t1 balancing it, but my point was if the MM puts 4 T1 guys on each side(the real top 5% guys) and then adds the t2 guys to each to balance it it will work better.

Edited by Funk1777, 08 January 2019 - 06:53 PM.


#12 Khobai

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPipPip
  • Elite Founder
  • Elite Founder
  • 23,969 posts

Posted 08 January 2019 - 07:14 PM

View Postjustcallme A S H, on 08 January 2019 - 04:37 PM, said:

The issue is that someone who should be in Tier 4, can climb up to Tier 1.


No thats not the issue at all.

The issue is that the Tier system is NOT a true ELO system and trying to turn it into a half-assed ELO system will never work. You're just dividing a constantly dwindling player base into more buckets that can't play against eachother. That won't end well.

What the game needs is a true ELO rating system.

With a true ELO rating system you can draw all players from one bucket (instead of having multiple buckets that cant play eachother) and then balance the teams using average ELO.

The entire flawed Tier system needs to be scrapped in favor of a true ELO system.

Edited by Khobai, 08 January 2019 - 07:15 PM.


#13 justcallme A S H

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • CS 2020 Referee
  • CS 2020 Referee
  • 8,987 posts
  • LocationMelbourne, AU

Posted 08 January 2019 - 07:23 PM

View PostFunk1777, on 08 January 2019 - 06:51 PM, said:

Im pretty sure those top 10% play a hell of a lot more than the others with a few exceptions. They play way more than t3 bob who has a 97% rating but doesnt play much so he is t3 or t2.

You dont think match quality would be better if the top 30% faced each other instead of getting 50% and lower guys thrown in.


Ok I've got an idea to paint the picture... From Jarls we know the overall active population. Bare with me as these numbers are rough and definately flawed but not flawed enough the point will be hugely inaccurate.

Top 30% = ~11,500 players out of 39,500 going by Jarls currently.

Top 30% means player 11,500 has 230 Average Match Score - Barely above the overall average of the entire population.

Now because Jarls includes more than just last season. Lets trim the numbers down... Last season there was 27,000 active. This is a 31.5% drop.

Top 30% = 7,900 players out of 27,000.

7,900 players is not enough to support T1 & T2 in a 24x7 game with over a dozen queues.

PGI understand this as Paul outlines in his MM post that it needs to be Tier 1 to Tier 3. If there was not the population to support T1/T2 only 12 months ago - There is not the population to support it now is there?

Furthermore you clean up the Tiers - having even less people in there wait time isn't going to improve, thats not possible. It will tank and tank massively.

That is why T1 --> T3 (a 2 factor) system will always be required in MWO to find matches in a timely manner.

View PostKhobai, on 08 January 2019 - 07:14 PM, said:

No thats not the issue at all.


With PSR - the topic of discussion and the podcast - that exactly what it is.

#14 Knuckles OTool

    Member

  • PipPipPip
  • 79 posts

Posted 08 January 2019 - 07:48 PM

View Postjustcallme A S H, on 08 January 2019 - 07:23 PM, said:



That is why T1 --> T3 (a 2 factor) system will always be required in MWO to find matches in a timely manner.




Are you saying the current system is fine because bottom 30% players play against you? Thats what a t3 player who plays a lot of games is. The question is how much longer of a wait time is needed to have better players facing each other. They play way more games so they are in more games so they only way to find out is try.

You are basing the population on everyone that has played in the last month but you have to take into account many of those only played a few matches. Is there a way to know for a fact that the top 30% players dont play enough to keep games going? If the solo queue was more competitive would group players feel better about joining it or synch dropping?

Or is there some stat that show how many games all the people played that is presorted? I was checking Jarl's and I cant figure out how to sort without sifting through tons of data page by page.

#15 justcallme A S H

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • CS 2020 Referee
  • CS 2020 Referee
  • 8,987 posts
  • LocationMelbourne, AU

Posted 08 January 2019 - 08:43 PM

View PostKnuckles OTool, on 08 January 2019 - 07:48 PM, said:


Are you saying the current system is fine because bottom 30% players play against you? Thats what a t3 player who plays a lot of games is. The question is how much longer of a wait time is needed to have better players facing each other. They play way more games so they are in more games so they only way to find out is try.

Yes but do you know that as long as you have 170-190 average match score, with a 0.9 WLR - you can max out a Tier 1 PSR bar as long as you play enough games?

There are many players, who play regularly, that fall into this category

View PostKnuckles OTool, on 08 January 2019 - 07:48 PM, said:

Or is there some stat that show how many games all the people played that is presorted? I was checking Jarl's and I cant figure out how to sort without sifting through tons of data page by page.


Ya - kinda. https://leaderboard....stats#gamechart

Now I know this chart isn't accurate in terms of since the dawn of time history. But it gives you an idea of the bulk / spread. I'm guessing it might be averages or something? It's definately not the last 3 months and absolutely not the overall history.


I'll see if Taro knows.

Either way the bulk is 100 match score to 300 match score and ~1k-2k matches.

Edited by justcallme A S H, 08 January 2019 - 08:47 PM.


#16 justcallme A S H

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • CS 2020 Referee
  • CS 2020 Referee
  • 8,987 posts
  • LocationMelbourne, AU

Posted 08 January 2019 - 09:11 PM

View Postjustcallme A S H, on 08 January 2019 - 08:43 PM, said:

I'll see if Taro knows.


Ask and ye shall receive... (dont view on a mobile lol)

Posted Image

This was data current to end of Oct 2018. Sorry there is nothing Dec. It's near enough IMO for the point here. I also asked him why some of the guys with 20,000 games were not on there and explaination below of chart.

[15:51] Tarogato: yeah there's dudes way way above 2,000 matches, but the chart is so sparse, so I cut it off
[15:51] Tarogato: the numbers in each cell are exactly how many people fit that X and Y coordinate for matchscore and gamesplayed

You can see quite clearly that the bulk of the player base and the bulk of the games played are around the median. Nothing near the high end.

If you tip the graph on it's side - it fits the bell curve presented during the VOD quite well although perhaps a little pointy like a wizard hard but it still almost shows exactly what I think in a different manner.

Funny how that's turned out actually, almost scary Posted Image

Edited by justcallme A S H, 08 January 2019 - 09:12 PM.


#17 justcallme A S H

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • CS 2020 Referee
  • CS 2020 Referee
  • 8,987 posts
  • LocationMelbourne, AU

Posted 08 January 2019 - 09:16 PM

Also some other stuff back from --> 2016 <-- that Taro linked me (I'd never seen, before my time). It might be interesting to some, somewhat unrelated though.

#18 Nightbird

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • The God of Death
  • The God of Death
  • 7,518 posts

Posted 08 January 2019 - 09:22 PM

As someone who does statistics and data analytics professionally, the whole concept of tiers, based on a progression bar or based on average MS and a bell curve, is insane. The variation within each tier from top to bottom is huge, and you're leaving yourself to luck as to whether one team will get the best from one tier and the other team the worst.

(For QP) The best way to make matches is to compute a number 1-100. or 1-1000 whatever, which A) represents the skill and expected contribution of a player, and B ) when two teams each have the same total summed across team members, results in a 50/50 chance for each team to win.

The way to compute this number is via a generalized linear model with all of the collected metrics as factors.

Match score by itself is useless because actions are assigned arbitrary values, not based on the actual impact it has on the match. A tier system based on MS would likewise be random and arbitrary, and will fail to produce even teams.

Edited by Nightbird, 08 January 2019 - 09:34 PM.


#19 HammerMaster

    Member

  • PipPipPipPipPipPipPipPipPip
  • The Benefactor
  • The Benefactor
  • 2,516 posts
  • LocationNew Hampshire, USA

Posted 08 January 2019 - 09:44 PM

TOO LONG DIDN'T WATCH.
HATES LRMS.
PSR needs work.
Stuff.

Anyone who continues to state lurm rather than L R M has issue with LRM.
My issue with what you stated is you said Nerf. NOT rework.
It truly needs a ground up redo.

Edited by HammerMaster, 08 January 2019 - 09:58 PM.


#20 Prototelis

    Member

  • PipPipPipPipPipPipPipPipPip
  • Shredder
  • Shredder
  • 4,789 posts

Posted 08 January 2019 - 10:10 PM

View PostHammerMaster, on 08 January 2019 - 09:44 PM, said:


Anyone who continues to state lurm rather than L R M has issue with LRM.



This might be the dumbest thing I've ever seen on this forum.





1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users