135 replies to this topic

[Mister Blastman]

Davers, on 22 September 2015 - 08:54 AM, said:

Why couldn't you have just said this to Winterburn and avoided a whole page worth of argument?

I'm in one of those moods.

[Roadkill]

Ultimatum X, on 22 September 2015 - 07:07 AM, said:

What's your opinion of this?

Damage != Teamwork.

Ergo, fail.

[GRiPSViGiL]

Have to admit my playstyle is tailor made for this sort of ranking mechanic. Would like to see a little bigger bonus for some of the more game mode specific things but it seems fair to me. Kills don't need a bigger bonus considering there is a boost to most damage dealt on a kill. When proper focus fire is underway the guy doing the lion share of the damage should get a good score especially if he don't get the kill.

I really rather like this way of ranking if this is indeed accurate. If so, I will have warm feelings toward PGI for at least the rest of the day.

[Roadkill]

Mister Blastman, on 22 September 2015 - 07:33 AM, said:

folks who DO farm damage will get what they deserve. They'll be thrust into tier 1 where they'll get their assess handed to them. I'm pretty darn sure the players at the top don't care about farming damage.

Exactly. Which is why the actual best players won't be T1 - they're far too efficient at killing and so won't be getting those astronomically high match scores. Unless they're just out farming pugs for giggles.

The system is borked because the most lethal players won't get the highest PSR.

Take it to the extreme if you can't see it. What if a player was so good that every time they fired they headshot their target? Their match score would be terrible because kills are only worth ~4 pts. The n00b next to them chain firing LRM-15s would be racking up huge points while the headshot king would "only" be getting all the kills.

The system's broken.

[merz]

I like how blastman articulated trolling as 'being the devil's advocate'..

PSR is a misnomer as it does a poor job of calculating anyone's skill. Instead it makes the queue for the direwolf look like a disneyland ride, since its one of the few machines capable of consistently, reliably producing the damage output necessary to maintain a higher tier. When PGI publishes the metrics for this weekend's tournament and you look at what the mech usage was for 6000-9000 score totals, I am going to be right there with you. and how I shall laugh.

Edited by merz, 22 September 2015 - 09:25 AM.

[Kubernetes]

My problem with some of this thinking is that kills can largely be a matter of luck and timing. I'm sure people aren't unfamiliar with games where they've done top damage but managed one or zero kills. I don't farm damage; I prefer pinpoint weapons and I shoot to kill (CT, ST on obvious IS XL, or legs on lights), yet I have lots of matches where I do a bunch of effective damage but miss out on kills by a split second (I suppose I could hold fire until my teammates crit an enemy, thereby getting extremely efficient kills, but that's just lame). Heck, I just had a game right before the cutoff where I killed 4 mechs in a row with a single alpha to each-- others had done most of the work and I just happened to luck into a succession of red-CT enemies. That was just pure luck, not elite efficiency.

I will say that when I spectate obvious Tier 1 players, they tend to score lots of kills *and* damage.

[Ultimax]

Mister Blastman, on 22 September 2015 - 08:24 AM, said:

Wow. I don't know what to say. You should lay off the caffeine. I'm not attacking your position--I'm instead expressing as a devil's advocate that other folks might want to see kills count for more. Or did you miss those threads?

I have nothing against someone playing devil's advocate, but do it without resorting to logical fallacies.

Otherwise the value of a devil's advocate is lost.





Let me lay this out again.

• Match making is based on PSR.
  
• PSR is based on match score.
  
• Match score is heavily based on raw damage, with other factors playing very, very minor roles.
  
• Once players on a team have a roughly similar skill level, you tend to see very even damage spreads.
  
• Very even damage spreads, tends to see most players scoring 300 to 500 damage.

The reality is that there just isn't enough armor to go around for everyone to score huge damage numbers.


There needs to be more factors weighted high enough to determine match score, raw damage alone is insufficient.



Running around in in-efficient streak-boats inflating damage is something that was prevalent this weekend - so was watching players specifically farm components/damage with laser vomit, and as such would be a good way to "move up in tier" even though you clearly aren't moving up in skill.

I don't think the system should reward degenerate gameplay.

Edited by Ultimatum X, 22 September 2015 - 09:39 AM.

[Sarlic]

PSR kills Communities.

[GRiPSViGiL]

Damage is fine as score metric, it strips armor and at least confirms someone is actually hitting a target. You should get some credit for that. Kills are secondary in my eyes since when proper focus firing is taking place the person doing most the damage usually doesn't score the kill. I have seen 200 or 300 damage racking in 4 or more kills several times. Should that score more than someone who did 2 kills with 800 damage? I don't think so. Of course there will always be ways to game the scoring mechanic but if someone wants to tailor their playstyle to game it then let them. Doesn't mean they play cohesively at all.

However, There are limitless scenarios and playstyles in which scores happen so it is very convoluted to say the least. Is someone sandbagging their team using everyone as shields to get their damage in? Did someone lead the charge and overwhelm every target they came across dosing most of the damage? PUG play will see these scenarios but at least PUG play has no merit or bearing on anything. Group play will be a much more consistent measure of rank when you are playing with teammates and friends. But hey, maybe someone people use their teammates to boost themselves up....naw that doesn't happen around here .

[Kyynele]

Roadkill, on 22 September 2015 - 09:08 AM, said:

Exactly. Which is why the actual best players won't be T1 - they're far too efficient at killing and so won't be getting those astronomically high match scores. Unless they're just out farming pugs for giggles.

Except that the actual best players also have to be good at shielding and spreading the damage they receive. And when good players play against each other, the damage will spread. Especially in the current long-burn-laser-meta. And there will be high damages before one side is dead.

The very good players will of course kill the very bad players very swiftly and efficiently, but this means that the bad players dying fast will get lower match scores and eventually drop out of the high tier games.

I'm not claiming the system is flawless or even good. It's really hard to figure out who was a good player just by staring at the statistics of a match. I'm just saying that the guys that are considered top tier in the game don't typically have the lowest damage numbers in the match end screen. I'm pretty confident that at least most of the truly best players will be T1 like they should.

[Deathlike]

Ultimatum X, on 22 September 2015 - 09:37 AM, said:

I have nothing against someone playing devil's advocate, but do it without resorting to logical fallacies.

Otherwise the value of a devil's advocate is lost.





Let me lay this out again.

• Match making is based on PSR.
  
• PSR is based on match score.
  
• Match score is heavily based on raw damage, with other factors playing very, very minor roles.
  
• Once players on a team have a roughly similar skill level, you tend to see very even damage spreads.
  
• Very even damage spreads, tends to see most players scoring 300 to 500 damage.

The reality is that there just isn't enough armor to go around for everyone to score huge damage numbers.



There needs to be more factors weighted high enough to determine match score, raw damage alone is insufficient.



Running around in in-efficient streak-boats inflating damage is something that was prevalent this weekend - so was watching players specifically farm components/damage with laser vomit, and as such would be a good way to "move up in tier" even though you clearly aren't moving up in skill.

I don't think the system should reward degenerate gameplay.

This still begs the question, how do you properly attribute good or bad gameplay within a match?

Without actual contextual evidence of good or bad play, you can't actually expect some magical algorithm to determine a player's worth in a match.

It is however easier to point out what circumstances that could exploit a "bad scoring mechanic", but it is completely unrealistic to expect it to account for flaws.

Much as I hate what PGI did with Elo, at the very least... people's contribution to a win is impossible to measure other than acknowledging a win is a win.. no matter how it was accomplished. PSR only demonstrates with Match Score by extension is that farming damage raises your PSR indefinitely unless you're bad (and you'll get moved to Tier 5 in short order).

Vomiting damage and not actually killing things (killing things efficiently isn't even rewarded well through C-bills anyways) is going to win the day (it's not that hard to scale damage rewards by efficiency, but PGI hasn't done that, because math).

It is what it is.

[Ultimax]

Deathlike, on 22 September 2015 - 09:50 AM, said:

This still begs the question, how do you properly attribute good or bad gameplay within a match?

Without actual contextual evidence of good or bad play, you can't actually expect some magical algorithm to determine a player's worth in a match.

It is however easier to point out what circumstances that could exploit a "bad scoring mechanic", but it is completely unrealistic to expect it to account for flaws.

You can't properly attribute good or bad gameplay completely, I agree.


There are however a number of factors that can be weighted highly to at least get a better overall picture.


Damage dealt is actually one of them.

Assists and kills are to a degree.

I think at the least a decent damage per kill efficiency score could also be used - something to both temper damage farming and also curb just pure kill stealing.

So is the amount of average damage you are capable of soaking before you die (i.e. percentage of armor left on average, per death - as opposed to all the people who just get CT cored all the time).

I also think mech class needs to affect your match score. It's clearly much easier to do huge damage numbers in a DWF compared to say a Mist Lynx.

I think friendly fire should lower your damage score, so should getting killed extremely early repeatedly in matches (a chronic display of poor play as opposed to just being unlucky once in a while).



The point is there are a lot of options that could be used to at least get a more overall accurate guage of player ability, and that I think they need to be implemented.

Edited by Ultimatum X, 22 September 2015 - 10:04 AM.

[Chuck Jager]

Winterburn, on 22 September 2015 - 07:32 AM, said:

And what is wrong with that exactly? How do you "steal" a kill anyway?

Should people not attack someone, who is being attacked by their teammate?

I had 89 damage and 3 kills in my high alpha SCrow the other night. It was a match where we had to run and chase the enemy before they self exploded in a 0-12 match. My biggest show of skill was not shooting team mates in the back while they jumped in front of me.

Many builds are more efficient at eliminating the mech quickly. Getting the killing blow is luck or holding your fire until the right time. There are matches where lazy lrm boats can really rake in the damage especially compared to their damage received. I have a feeling these folks have a feast or famine that will get averaged in. StreakCrows and Cats do better but they will be moved up a level and have to play a better game. If a LRM boat is actively creating a win by their placement on the field and selection of targets that win also also gives the team the chance to get more damage, assists and KBs.

PSR is about putting all effective high damage dealers against each other and yes their will be flaws. PGI may have math to add modifiers for how super efficient groups with lower numbers are handled, but we all know if they released it folks would cry more and have more ability to game the system.

Hopefully using PSR as a bragging right over seeing it as a balancing mechanism tells me that an individual was probably to fat and slow to play real sports. In competitive sports and the arts, a large percentage of the T1 folks I have met are the first to help other participants with technical and supportive advice. They also realize that they have gifts and also work very hard and grade themselves on the quality of their effort not the numbers on the scoreboard or the paycheck. (yes you can find plenty of exceptions to this rule)

[Davers]

GRiPSViGiL, on 22 September 2015 - 09:41 AM, said:

Group play will be a much more consistent measure of rank when you are playing with teammates and friends. But hey, maybe someone people use their teammates to boost themselves up....naw that doesn't happen around here .

But group play has a much greater spread of PSR than in solo play. There is no restrictions of how good players have to be to join a premade.

[AssaultPig]

as this tournament wonderfully demonstrated, match score is not correlated particularly strongly with 'things that make your team win' (especially in CW, ******* el oh el.)

I had many more high scores throwing missiles at things from a HBK-4J than from lights (cheetah, FS9s, etc) or various heavies because for the most part those mechs kill stuff before all that much damage is racked up. A fresh 100 tonner can go down in <100 damage if I get a good backstab, compared to the 400 (at least) damage the 4J would take to generate the same result.

At the macro level (i.e. PSR) it might not be that much of an issue though; even if the system is weighted a bit in favor of farming damage, better players are generally gonna do more.

[AdamBaines]

Mister Blastman, on 22 September 2015 - 07:54 AM, said:

Not trolling, I'm genuinely trying to be helpful.

This made me laugh. Good one.

[AdamBaines]

Winterburn, on 22 September 2015 - 08:15 AM, said:

Also, Santa isn't real. Yeah, I am feeling extra cruel today.

Aw come on man! Don't ruin it for us all because of one guy (who is USUALLY reasonable) being a tool bag.

Don't you DARE tell me the Easter Bunny is not real and that my wish WON'T come true if I blow an eye lash off the tip of my finger

[Mister Blastman]

Ultimatum X, on 22 September 2015 - 09:37 AM, said:

I have nothing against someone playing devil's advocate, but do it without resorting to logical fallacies.

Otherwise the value of a devil's advocate is lost.





Let me lay this out again.

• Match making is based on PSR.
  
• PSR is based on match score.
  
• Match score is heavily based on raw damage, with other factors playing very, very minor roles.
  
• Once players on a team have a roughly similar skill level, you tend to see very even damage spreads.
  
• Very even damage spreads, tends to see most players scoring 300 to 500 damage.

The reality is that there just isn't enough armor to go around for everyone to score huge damage numbers.




There needs to be more factors weighted high enough to determine match score, raw damage alone is insufficient.



Running around in in-efficient streak-boats inflating damage is something that was prevalent this weekend - so was watching players specifically farm components/damage with laser vomit, and as such would be a good way to "move up in tier" even though you clearly aren't moving up in skill.

I don't think the system should reward degenerate gameplay.

I don't think I was using logical fallacies. I know it when I (rarely) do--usually it is in the form of sarcasm.

Anyways... yes, damage shouldn't be the dominant metric like it is now. However, like it is now, it does work itself out...

Player first dozen games: 1k damage avg.
second dozen: 800 damage avg.
third dozen: 650 damage avg.
fourth dozen: 500 damage avg.
fifth dozen: 375 damage avg.

From this we could plot it (if I weren't lazy) and we'd see by the fifth dozen player is in Tier 1 and only doing 375 damage due to the high speed pace and kills. A Tier 1 player isn't going to average 375 @ Tier 4 and ever hope to make it to Tier 1.

However, those running troll builds like streakboats will get to Tier 1 and see their damage plummet possibly--to say 200 damage avg. and fall out of the tier back to Tier 2.

This doesn't make it right--but this is what should happen. Tier 1 should be highly optimized builds in capable hands that are meta (until we have perfect balance).

But as you point out, damage needs to be weighted differently. Let me re-iterate with an illustration and arbitrary numbers...

Do 50 damage... -150 to PSR score
Do 150 damage... -75 to PSR score
Do 300 damage... -10 to PSR score
Do 500 damage... +60 to PSR score
Do 700 damage... +75 to PSR score
Do 1000 damage... +100 to PSR score

As you see the PSR score curve ramps up quickly at the low end of the positive threshold and getting it to double... well, isn't possible. You have a cap at +100 for damage and once you hit it, you're done--but it also isn't the majority component, either.

I think that would be far more productive because it does two things:

1. It punishes the bads. They get hurt and fall down in tier fast.
2. The damage farmers can only go so far.
3. The good players who are consistent will have a plodding +60 bonus to their PSR score game after game and will help them stay at the top.

The metric needs to reward consistency without letting the extreme abusers distort everything. The cap is arbitrary. We could tighten it up, we could make it lower--but it needs to be there for this to work and reward consistent playing rather than blowout abusers.

In the end, though, I think it will work itself out. But who knows--it'd be interesting to see what builds the Tier 1 players use all the time.

Edited by Mister Blastman, 22 September 2015 - 11:57 AM.

[digitalMe]

How many games does it take to considerably improve pilot skill rating? The statistics at the end of the game show whether the skill rating has improved or not. However I do not have a clue how long it might take to achieve a higher rank.

When all games from January 2015 until today are used for calculating the score it might be really really difficult to change it. And I mean especially new players might have a lot of bad games until they know how to do it...?

Furthermore the GUI is - once again - pretty difficult to understand. What does the bar mean? Progress within a rank?

Edited by digitalMe, 22 September 2015 - 12:11 PM.

[Deathlike]

digitalMe, on 22 September 2015 - 12:08 PM, said:

How many games does it take to considerably improve pilot skill rating? The statistics at the end of the game show whether the skill rating has improved or not. However I do not have a clue how long it might take to achieve a higher rank.

When all games from January 2015 until today are used for calculating the score it might be really really difficult to change it. And I mean especially new players might have a lot of bad games until they know how to do it...?

Furthermore the GUI is - once again - pretty difficult to understand. What does the bar mean? Progress within a rank?

The more you win and deal damage, you will progress faster.

It's simple as that.

The bar shows how far up the tier you're at... when it's a little, you are "barely" above the tier... when it's a lot, you are almost ready for the next tier.