Jump to content

The Fallacy Of "jarl's List = Knowledge"


226 replies to this topic

#201 Brauer

    Member

  • PipPipPipPipPipPipPipPip
  • 1,066 posts

Posted 21 January 2020 - 05:53 PM

View PostKhobai, on 21 January 2020 - 05:51 PM, said:


GroupQ has not been so dead that no groupQ games at all have been played. Stop claiming that because its factually incorrect.

What evidence have I dodged? Your unfounded claim that groupQ is dead? Its not dead because ive played games in group queue within the last 6 months.


There have been so few GroupQ games that they are irrelevant. There's been all of one time that I recall the queue not being dead since like July and that was a player/streamer organized event intended to revive the dead queue.

#202 Brauer

    Member

  • PipPipPipPipPipPipPipPip
  • 1,066 posts

Posted 21 January 2020 - 06:02 PM

Soooo editing after a reply is interesting.

When you work with data you always get some data that is not particularly useful. Let's say someone just randomly fills in bubbles on a survey. The idea is that you can still identify trends in the data. I haven't seen anyone but you Khobai argue that GroupQ is effectively dead. From that we can assume it is having a minimal impact on the data and work with what we have. Sure in an ideal world it would be excluded, but the world never offers perfect datasets, and expecting it to is folly.

My proof is that, despite the aforementioned event, we still had to wait in queue for a while for GroupQ during it, and I have at other times sat in GroupQ and gotten no matches during prime time. I have heard the same and nobody else is claiming GroupQ is bumping, so the burden of proof to show enough GroupQ matches are happening to matter is firmly on you.

FYI you showed no proof of all these GroupQ matches too :)

EDIT: lol you are editing your post to try to make it sound like you had already addressed my point in your original post. Just post a reply.

Edited by Brauer, 21 January 2020 - 06:05 PM.


#203 Brauer

    Member

  • PipPipPipPipPipPipPipPip
  • 1,066 posts

Posted 21 January 2020 - 06:10 PM

I never said stats tell the whole story...but if someone can't perform there's good reason to doubt that they know what they're talking about, particularly if their opinion conflicts with high performing players.

Sure someone may once have had a high rating, but maybe the meta changed and their skills don't translate to the new meta, or they didn't keep up with those changes. In any case I want to get advice from people who have proven through consistent performance that they understand the game and can deliver the goods. That's why I've been thankful to learn from some top comp pilots, even if I still have plenty more to learn.

#204 Xiphias

    Member

  • PipPipPipPipPipPipPip
  • Littlest Helper
  • Littlest Helper
  • 862 posts

Posted 21 January 2020 - 07:15 PM

View PostKhobai, on 21 January 2020 - 05:25 PM, said:

Ash knows what I said was factually correct. Hes just incapable of admitting when hes wrong.

So what he does is resort to name calling, stat shaming, and strawman arguments to try and save face and deflect people from the original argument. He does this every single time he loses an argument to me.

My post wasn't directed at the argument that you and Ash were having, it was specifically referencing the issue with Dee's post.

Since you want to bring me into this though, let's take a closer look at your argument.

Quote

When you try to win an argument by arguing a completely different and unrelated point its called a strawman argument; they are fallacious arguments that have no relevance.

To be clear on this, "A straw man is a form of argument and an informal fallacy based on giving the impression of refuting an opponent's argument, while actually refuting an argument that was not presented by that opponent"

You original argument was

View PostKhobai, on 16 January 2020 - 09:35 AM, said:

jarls doesnt separate stats for solo and group queue. that fact alone taints a lot of the data.

if you play with good players in group queue of course your stats will be better than if you play with bad players in solo queue.

The implication being that you can't tell how good a player is based on their stats because they could have farmed them up using group queue.

In response, this is Ash's original position:

View Postjustcallme A S H, on 17 January 2020 - 04:02 PM, said:

Plus it is EASY to discern when GroupQ vs Solo stat players in Jarls. Even when people cross over between the two.

Arguing that stats are still useful because it's possible to distinguish which players are primarily playing group or solo queue, even when there is a crossover between the two.

This is a straw man of his position

Quote

Okay so why dont you definitively break down your stats for us and tell us how many games are solo games and how many games are group games. And what type of game each of your stats came from? Oh you cant? Because youre !@#$ing wrong as usual.

No one has argued that individual games can be extracted out of the stats, just that you can tell the difference between group and solo players and that stats are still useful for evaluating player skill.

You are literally taking a straw man of original position and then accusing Ash of using straw man arguments. Can you see the hypocrisy here?

Everyone agrees that stats would be more accurate and better if group and solo stats were separated. However, to the original argument that this makes stats useless in evaluating player skill because it's impossible to distinguish between the two, I disagree for the following reasons.

1) Group stats do tend to have distinct indicators, a good group inflates WLR, usually KDR, and deflates AMS. If the group is bad, it's not going to inflate any stats so it's mostly irrelevant for this discussion. Solo matches tend to inflate AMS at the cost of deflating WLR.
1b) Based on these indicators players that play a lot of group matches are usually pretty easy to distinguish from those that mostly play solo. Someone with a WLR

Examples


2) Group matches are a small percentage of all matches. I don't have time to dig it up right now, but even in their prime group matches were a small percentage of total matches. The number of those groups who were good enough to influence stats, even smaller.
2b) Even if group matches are able to significantly skew stats, they only affect a small number of players

Consequently, the combination of the above two means that while some small percentage of players may have been able to indistinguishably inflate their stats, for the vast majority of matches and players it's possible to get a rough estimate of that player's skill compared to another player. The bigger the gap, the more accurate this guess is likely to be.

A predictor doesn't need to be perfect to be useful. We can't perfectly predict the weather and yet the forecast is still a very useful tool for planning. Jarl's list is just this, a useful, generally accurate, predictor of player performance and often a decent predictor of player game knowledge and understanding.

View PostKhobai, on 19 January 2020 - 12:26 PM, said:

And yes good players will have better stats in the long run. I am not refuting that. My point was simply that jarl's list has limitations and as such should not be used as the sole means of determining if a player is good or not. And your stats certainly should not determine the validity of your opinion or justify behavior like statshaming.

In absence of other evidence, stats are the best and most empirical tool for evaluating how good a player is. I don't have time to personally vet every player who thinks they know how to play the game to determine if they are actually good or not. If they're offering solid evidence, I'll evaluate that on its merit, but if they are claiming to be a knowledgeable source I'm going to check their stats and if they are bad I am going to assume that player doesn't know what they are talking about unless they can provide evidence to the contrary.


View PostKhobai, on 19 January 2020 - 04:55 AM, said:

Also the concept of players having multiple accounts still seems to elude you; which only further proves how ignorant you are. I havent played on this account in 3 years that doesnt mean I dont play the game on a different account.

Quote

it amazes me too. especially when I tell people i havent played on that account for 3 years and use an alt account.

But it doesnt suite people like Ash to believe I play on an alt account. Because then they couldnt claim I dont play the game anymore. Because that is one of the cornerstones of his fallacious arguments.

There's an easy solution to this problem. Tell us what your alt is. You could easily discredit this argument if you wanted to. Refusing to disclose your alt just makes it look like you have something to hide.

If you're unwilling to do this (for whatever reason) that's fine, but stop complaining that people don't believe you about something you could easily prove

#205 Vxheous

    Member

  • PipPipPipPipPipPipPipPipPip
  • CS 2019 Gold Champ
  • CS 2019 Gold Champ
  • 3,822 posts
  • Location2 Time MWO World Champion

Posted 21 January 2020 - 08:12 PM

View PostVonBruinwald, on 21 January 2020 - 03:45 PM, said:


I've seen this claim a lot but is there any proof?


Please go look at my current month's stats. This month, for the first time in many months, we actually had group queue! (last Friday night). My average match score for the month is 338 (much lower than usual for me). Why is that you ask? It is because we were killing mechs too fast in group queue, and there was no damage/match score to be had. Here's 2 examples:

Posted Image

Posted Image

Here's my stats for the previous 3 months (zero group queue games) I can guarantee that my Jarl's ranking will tank this month because of all those group queue games where we all tanked our match scores (I don't actually care what my Jarl's ranking is BTW):
Posted Image

Edited by Vxheous, 21 January 2020 - 08:20 PM.


#206 justcallme A S H

    Member

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

Posted 21 January 2020 - 08:25 PM

Yep, my average from those 10 or so GroupQ matches was ~320. KDR of ~7.0 and WLR of ~2.5.

Only reason the WLR was lower than what GroupQ should show was because I had already done ~12 QP matches. Lost 8-9 of them. Not hard to work that out though.

If I had not played any QP, it would be a 10.0 WLR+

#207 MischiefSC

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPip
  • The Benefactor
  • The Benefactor
  • 16,697 posts

Posted 22 January 2020 - 12:01 AM

View PostVxheous, on 21 January 2020 - 08:12 PM, said:


Please go look at my current month's stats. This month, for the first time in many months, we actually had group queue! (last Friday night). My average match score for the month is 338 (much lower than usual for me). Why is that you ask? It is because we were killing mechs too fast in group queue, and there was no damage/match score to be had. Here's 2 examples:



Please do keep in mind that the sample size is really small but in a larger scale, the math works out to 4-6 being ideal. Larger than that and you're spreading the damage and kills out too much with a diminishing return on higher w/l. Also the lower tonnage roles get passed around to more people in your group so your odds of being the guy 30 tons under optimal is there to skew your average down.

4-6, you're all in good performers and able to consistently secure 2+ kills a match, win almost every game and die only every few games (again, you've got more tonnage to field).

Which is why it's such a statistical anomaly. The number of players who can consistently run 4-6 players who can carry almost every match and farm that well is well under 100. Even with the MWO population shrinking by 10k players to 15k that makes you guys about 1/2 of 1% of the games population who can consistently inflate your stats from group queue.

While you do create an artificial downward pressure on everyone elses stats you're too few to make a real impact as most players will play against you maybe 1 game in 20 in a 'per 100 matches' sense so your tiny population representation carries through to your overall impact on stats.

Or, in simple terms, too few players can play that well as to have a significant impact on anyone elses stats and while they (in this case, YOU and the people you play with) do have their stats skewed, it's just inflation of an already big gap. 6 months ago you were about 1/8th of 1% of the population. With population dropping you're still only 1/2 of 1%, and that's generously assuming that about 80+ of the top performing players still play regular in group/pug queue, no counting FW matches.

So. To reiterate. Group queue wins are a statistically insignificant part of Jarls List stats once you get past the first 2 pages or so.

#208 Sjorpha

    Member

  • PipPipPipPipPipPipPipPipPip
  • Philanthropist
  • Philanthropist
  • 4,472 posts
  • LocationSweden

Posted 22 January 2020 - 04:09 AM

Khobai:

I don't really care if you play on an alt or not, all I can say is that you talk on the forums like someone who isn't up to date with the game.

#209 K O Z A K

    Member

  • PipPipPipPipPipPipPipPipPip
  • 2,322 posts
  • LocationTrue North Strong and Free

Posted 22 January 2020 - 10:15 AM

"it's a game who cares" ..... proceeds to write a wall of text about the game

I've never heard anyone except a few bad founders claim that being a founder means anything in terms of performance, where did you get that silly idea?

#210 Marquis De Lafayette

    Member

  • PipPipPipPipPipPipPipPip
  • CS 2023 Top 12 Qualifier
  • CS 2023 Top 12 Qualifier
  • 1,396 posts
  • LocationIn Valley Forge with General Washington

Posted 22 January 2020 - 10:27 AM

View PostHazeclaw, on 22 January 2020 - 10:15 AM, said:

"it's a game who cares" ..... proceeds to write a wall of text about the game

I've never heard anyone except a few bad founders claim that being a founder means anything in terms of performance, where did you get that silly idea?


^

And his jarl’s looks like it’s probably an alt. Rolling up the stats on some poor tier 4-5’s and coming back to earth as he/she hits tier 1.

#211 MischiefSC

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPip
  • The Benefactor
  • The Benefactor
  • 16,697 posts

Posted 22 January 2020 - 10:28 AM

The bigger concern to people should be what I'm seeing in the stats. You take that top two pages of players, that 1/2 of 1%. Almost all of them spent 2018 with 220-330 matches/month as an average. Suddenly half way through 2019, BOOM. Their total games played drops almost by a factor of 10. Suddenly they're 25-40 matches a month.

No such thing as casual comp tier play. Oxymoron. GIT GUD and STAY GUD takes a lot of practice and it involves an investment in time and energy, the motivation to figure out and practice what wins, new challenges, etc. So when that population segment cuts their game time by 90% pretty much across the board it's a bad thing.

All but a couple of the top 10 teams have 7 or less members that are not 'retired' per Jarls.

Of that top 60 players who are not retired I could only find NINE with an average of more than 50 matches a month since the middle of 2019. Conversely all of them had an average of HUNDREDS (generally hovering around 300) a month prior to that.

So you go to the Global Stats and dig there and find we're equal to the lowest average matches played per player that we've ever been. There were two prior months that were this low - one was a big FW release, the other was our last big crash, August 2018, which was the result of horrible balance changes and some of the big FW units voiding out of MWO after the realities of One Bukkit settled in.

Also, NKVA came back for a few games. I blame this post.

Given the 'maintenance mode' label on the game I don't foresee any amazing new update or big promises coming to turn that around. Only question is where the slide bottoms out. Ironically MW5 has brought a small bump of new players.

#212 _Magno_

    Member

  • PipPipPipPipPip
  • Bad Company
  • Bad Company
  • 118 posts

Posted 22 January 2020 - 10:32 AM

Update of contributions that are hard to capture in the stats.

I took out a Stealth Locust with a unit friend who brought a LRM boat.
What luck that we both got on the same team, and we went to Polar Highlands.

The idea was to spot for him and his LRM build, attempt to squirrel the enemy to get them to turn their back on our front line.

For about 6 minutes of game time, I was in their back line maintaining LoS on targets and staying within ECM range to jam and allow my side to target fast and shoot. I would pop off a few small laser shots, comm-ing with team to get ready to focus on marked target, because they're about to show their back.

It was a very tight game, we were down 8-10. At this point, I'm the only fresh mech and the enemy is pretty soft.

I ended up getting 3 kills, almost got the final kill but I collided with the exposed CT Black Knight and my leg blew out and he was able to shoot me down.

The game ended in 12-11 loss.
I had 3 kills, 10 assists, 305 damage, and a match score of ~425.


I have these type of games maybe 1 out of 3.
1 of 3 are stinkers
1 of 3 are middle of the road. maybe get a kill, maybe survive, maybe win. These games usually feel like I was along for the ride.

#213 A Rabid Raccoon With A Shotgun

    Member

  • PipPip
  • 24 posts
  • LocationSomewhere in Alaska foaming at the mouth

Posted 22 January 2020 - 10:33 AM

View PostMarquis De Lafayette, on 22 January 2020 - 10:27 AM, said:

^

And his jarl’s looks like it’s probably an alt. Rolling up the stats on some poor tier 4-5’s and coming back to earth as he/she hits tier 1.

No not an alt, just someone that likes stompy robots, blinky lights and boom boom guns. I used to play Hawken before it died tho.

#214 MischiefSC

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPip
  • The Benefactor
  • The Benefactor
  • 16,697 posts

Posted 22 January 2020 - 10:38 AM

View Post_Magno_, on 22 January 2020 - 10:32 AM, said:

Update of contributions that are hard to capture in the stats.

I took out a Stealth Locust with a unit friend who brought a LRM boat.
What luck that we both got on the same team, and we went to Polar Highlands.

The idea was to spot for him and his LRM build, attempt to squirrel the enemy to get them to turn their back on our front line.

For about 6 minutes of game time, I was in their back line maintaining LoS on targets and staying within ECM range to jam and allow my side to target fast and shoot. I would pop off a few small laser shots, comm-ing with team to get ready to focus on marked target, because they're about to show their back.

It was a very tight game, we were down 8-10. At this point, I'm the only fresh mech and the enemy is pretty soft.

I ended up getting 3 kills, almost got the final kill but I collided with the exposed CT Black Knight and my leg blew out and he was able to shoot me down.

The game ended in 12-11 loss.
I had 3 kills, 10 assists, 305 damage, and a match score of ~425.


I have these type of games maybe 1 out of 3.
1 of 3 are stinkers
1 of 3 are middle of the road. maybe get a kill, maybe survive, maybe win. These games usually feel like I was along for the ride.


W/L captures 100% of all your behaviors with a sample size over about 80 matches.

Everything you do well or poorly that impacts your teams ability to win is, without question, reflected in your W/L. It's absolutely true that high KDR and AMS contribute to a higher W/L but not always. For example if you drive nothing but LRMs, hang in the back, use your team like meatshields and farm damage you'll end up with a high KDR and AMS while a low to mediocre W/L. However if you pilot high alpha, high precision mechs and consistently kill mechs with minimal wasted damage you'll have a relatively low AMS but a strong KDR and a high W/L. There's a lot of examples of these variables.

However law of large numbers and the nature of averages in a QP environment means that you're being passed around in the same random pool of players as everyone else. Over a good sample size (about 80 matches) the impact of your behavior on the odds of your team winning (or losing) plays out. Your W/L is a clear reflection of how well what you do helps your team win matches.

#215 Marquis De Lafayette

    Member

  • PipPipPipPipPipPipPipPip
  • CS 2023 Top 12 Qualifier
  • CS 2023 Top 12 Qualifier
  • 1,396 posts
  • LocationIn Valley Forge with General Washington

Posted 22 January 2020 - 11:03 AM

View PostA Rabid Raccoon With A Shotgun, on 22 January 2020 - 10:33 AM, said:

No not an alt, just someone that likes stompy robots, blinky lights and boom boom guns. I used to play Hawken before it died tho.


Well....if it’s not an alt, then you are off to a pretty fast start in this game, congrats. Play a few months in tier 1, put up those kind of nice stats and everyone will be convinced of that. You will probably also learn that more than a few founders are perma-bads. Tenure doesn’t mean anything really ....talent and knowledge of the game always shows through though.

#216 Marquis De Lafayette

    Member

  • PipPipPipPipPipPipPipPip
  • CS 2023 Top 12 Qualifier
  • CS 2023 Top 12 Qualifier
  • 1,396 posts
  • LocationIn Valley Forge with General Washington

Posted 22 January 2020 - 11:08 AM

View Post_Magno_, on 22 January 2020 - 10:32 AM, said:

Update of contributions that are hard to capture in the stats.

I took out a Stealth Locust with a unit friend who brought a LRM boat.
What luck that we both got on the same team, and we went to Polar Highlands.

The idea was to spot for him and his LRM build, attempt to squirrel the enemy to get them to turn their back on our front line.

For about 6 minutes of game time, I was in their back line maintaining LoS on targets and staying within ECM range to jam and allow my side to target fast and shoot. I would pop off a few small laser shots, comm-ing with team to get ready to focus on marked target, because they're about to show their back.

It was a very tight game, we were down 8-10. At this point, I'm the only fresh mech and the enemy is pretty soft.

I ended up getting 3 kills, almost got the final kill but I collided with the exposed CT Black Knight and my leg blew out and he was able to shoot me down.

The game ended in 12-11 loss.
I had 3 kills, 10 assists, 305 damage, and a match score of ~425.


I have these type of games maybe 1 out of 3.
1 of 3 are stinkers
1 of 3 are middle of the road. maybe get a kill, maybe survive, maybe win. These games usually feel like I was along for the ride.


Not trolling here . Looks like your contributions were captured pretty decently. Good match score, great kills and not terrible damage for a locust. So, what the issue? Looks like it was captured in the stats.

Edit: one thing off about your stats is you can’t have 3 kills and 10 assists....that would be 13 mechs. So, you missed something there.

Edited by Marquis De Lafayette, 22 January 2020 - 11:10 AM.


#217 Teenage Mutant Ninja Urbie

    Member

  • PipPipPipPipPipPipPipPip
  • The Tip of the Spear
  • The Tip of the Spear
  • 1,678 posts

Posted 22 January 2020 - 11:30 AM

View PostMarquis De Lafayette, on 22 January 2020 - 11:08 AM, said:

Edit: one thing off about your stats is you can’t have 3 kills and 10 assists....that would be 13 mechs. So, you missed something there.


Entirely possible - You just have to count the 'good kills', too .. *

Posted Image


*yeah, I know.. but still ;)

#218 Dee Eight

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Big Daddy
  • Big Daddy
  • 6,271 posts

Posted 22 January 2020 - 12:40 PM

View PostDr Cara Carcass, on 21 January 2020 - 03:20 PM, said:

He last played in 2017 and still is on the forums and acts as if he actually experienced the current state of the game.


You really shouldn't use that argument when your own play time has dropped off significantly the past few months.

#219 Prototelis

    Member

  • PipPipPipPipPipPipPipPipPip
  • Shredder
  • Shredder
  • 4,789 posts

Posted 22 January 2020 - 12:41 PM

Still more games than khobai, still better games than yours lol.

#220 justcallme A S H

    Member

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

Posted 22 January 2020 - 12:44 PM

View PostBud Crue, on 22 January 2020 - 11:33 AM, said:


He implied ASH, of all people, is both white knighting and acting as a social justice warrior in the above exchanges.

I get the feeling that perhaps he may not be clear on the meaning of these labels.


Yeah...

I'm a SJW... About 90s of watching my stream would pretty much denounce that... Let alone the last 3 years of it.





1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users