63 replies to this topic

[Paigan]

Yes, yes, I know: the hundredth thread about PSR.

But no flaming or whining.
I think the thing is a good idea in principle.
However details in its logic are currently not sane and need adjustment. Because otherwise (i.e. current state), the system is highly counter-productive and absurdly unfair to both good and bad performance.

Examples (seen in dozens of matches now):

1.)
Say I make 400 damage and 1 kill and we win. I'm getting uprated for sure.
Say I make 400 damage and 1 kill and we get stomped (which is a CONSIDERABLY better performance because of harder circumstances). I get downrated EVERY TIME.
This is NOT sane.

If you want to tie PSR uprating to the team win, fine. Then let people performing "halfway good" in a defeat outcome at least keep their score even, but don't downrate them for performing well under hard circumstances. This is NOT sane.


2.)
Say I played badly, ~150 score (and not due to bad score formula but actual bad performance), but we win.
I get uprated. Why? Makes no sense.
Okay, okay, maybe I helped a little, contributed a little, weren't all too bad. So keep the score where it is, but don't reward bad play for christ's sake.
Again: this is NOT sane.


My (rough) suggestion* would be:

Win:
Player score < 100: downrate
Player score in [100;200]: keep even
Player score > 200: uprate

Defeat:
Player score < 200: downrate
Player score in [200;400]: keep even
Player score > 400: uprate

(the values can/should maybe be adjusted for each tier, but the principle could stay the same)



(*) Of course that only makes sense if the match score calculation is halfway fair.
Currently, it is extremely half baked. Objectively determinable. Note the following examples:

1.)
Say a player makes 12 headshots (bear with me, for the sake of exemplification).
Afaik that would give him a score of 342 (for 198 score for 12*(15+18)=396 damage plus 12*(4+8)=144 score for solokilling 12 mechs).
Compare that to a guy who LRM-sprays 1000 damage over 12 mechs without destroying a single component.
Whatever "help" for "opening up" he might be, he definitely was LESS help than killing 12 mechs on his own.
His score: 500. Considerably HIGHER that the godlike headshot guy.
This is NOT sane.

2.)
A more realistic example:
I try to play efficiently. I have a Wubhawk, halfway okay aim, I aim for CT, kills are not too hard if I don't make big mistakes. Especially on medium mechs.
Meaning I don't require a lot of raw damage to do a kill (solo kill even), I do precise damage.
Very often, my matches look like this: 350 damage, 2 kills, 200-250 score. Damage- and scorewise, I did "okay", but actual contribution wise, I lifted twice my weight.
This is neither fair nor sane.

3.)
Killing an UAV requires 10 damage (AFAIK) but yields only 2 score.
Randomly distributing 10 damage on mechs yields ... 5 score.
Similar with capping, etc. All the actual tactical contributions bring JOKE score compared to the HUNDREDS of damage score.
NOT SANE.



Conclusion:
Damage is WAY overrated in determining match score.
Especially useless damage.
Damage that just scratches the armor of an arm or the leg, when the kill actually happened through torso or head is IRRELEVANT damage.

Why not simple measure actual relevant damage?
- If a component is destroyed, use the damage inflicted (both to armor and structure) to it to determine who gets what share of the component's total score worth. Mass Effect Multiplayer works that way (surely many other games out there) and it was ALWAYS PERFECTLY FAIR AND SIMPLE.
- Same for a destroyed Mech. Of course only the damage that led to the kill counts (e.g. all damage to left and/or right torso or to head or to CT or to bothlegs).
- Damage that takes out a piece of equipment via critical hit (HS, weapon, ammo, whatever)

Everything else is IRRELEVANT, wasted damage and should NOT be rewarded with score.




The current match score formulas can be described by two very descriptive sentences:

1.) "If you want to have low score, do the mission objective."
2.) "Match score punishes accuracy, efficiency and actual contribution but rewards useless damage spaying"

Not sane.

Edited by Paigan, 09 October 2015 - 06:54 AM.

[Ted Wayz]

How you go up and down should be dependent on your performance that match. Not some monolithic system that tries to equate your match score to teamwork.

[Sorbic]

"Everything else is IRRELEVANT, wasted damage and should NOT be rewarded with score."

Of course this is wrong. Damaging that mechs arm or torso might not have killed it but it could very likely have changed the aggressiveness of the pilot. Made him start poking on the left side to protect his heavy guns on the right or just hanging back in general. Shooting someone and not killing them is far from irrelevant as it can prompt them to stop pressing an attack/killing your mate. It makes them turn/torso twist away and causing more of their shots to miss. It can put them behind cover and allow you and your mates to advance without getting cored yourselves. So on and so forth...

Edited by Sorbic, 27 September 2015 - 01:48 PM.

[Osteo2001]

Paigan, on 27 September 2015 - 01:09 PM, said:

3.)
Killing an UAV requires 10 damage (AFAIK) but yields only 2 score.
Randomly distributing 10 damage on mechs yields ... 5 score.
Similar with capping, etc. All the actual tactical contributions bring JOKE score compared to the HUNDREDS of damage score.
NOT SANE.

While all you say is true, point 3 burns me more than anything....

I scan the skies looking for UAV's as if they're up, we're seen. I think that taking out a UAV should be at least 5k c-bills and the formula equivalent gxp as strategically, this is at least as important as taking out a mech (hence the 5k, it's the same as kill most damage dealt).

[Devayner]

Osteo2001, on 27 September 2015 - 01:56 PM, said:

While all you say is true, point 3 burns me more than anything....

I scan the skies looking for UAV's as if they're up, we're seen. I think that taking out a UAV should be at least 5k c-bills and the formula equivalent gxp as strategically, this is at least as important as taking out a mech (hence the 5k, it's the same as kill most damage dealt).

I think a lot of the UAV kill and the scouting /spotting rewards need a upgrade. As its to biased to damage.It feels like a light gets the shaft, if you want players to work as a team then make the rewards worth it and they will.There is no point in taking a light when i can roll a dashi and skate through on damage alone fluffing my match score.

[MechWarrior5152251]

What is not sane is how much people cry about PSR. If PGI had known how many tears people would have at knowing they would have kept it secret, which they could still do.

[LordNothing]

i dont think anyone gives a crap what their score is. whet everyone does seem to be bothered by is that the game is suddenly much less fun. pri's crappy rng system could do a better job at finding matched opponents.

way i see it personal combat performance and teamwork are two different things and each should be rated separately. combine that with the fact that psr depends on score, and scoring has a lot of issues. there are no rewards for marksmanship. rewards for scouting are dependent on the team using the intel. so whatever issues there are with scoring is being magnified. there should be a relative skill factor on things like cbills and xp (not so much score) so you can still get paid in games where you face stronger opponents.

[An Atlas]

Paigan, on 27 September 2015 - 01:09 PM, said:

The current match score formulas can be described by two very descriptive sentences:

1.) "If you want to have low score, do the mission objective."
2.) "Match score punishes accuracy, efficiency and actual contribution but rewards useless damage spaying"

Not sane.

PSR is not a measure of how good at killing you are.

It is a measure of how good you are at making it easier for other people to kill.... aka being a good team mate.

/thread

[bad arcade kitty]

you should be uprated for every win and downrated for every loss even if you did 3000 damage and 11 kills in that loss
and the amount of the rating which you get or lose shouldn't depend on your performance too, only on the rating of your teammates and enemies

that's how the proper player rating should work, otherwise we have the current crap where everybody eventually moves to the top rating if they play enough games i.e. the top rating means nothing

[An Atlas]

bad arcade kitty, on 27 September 2015 - 04:57 PM, said:

that's how the proper player rating should work, otherwise we have the current crap where everybody eventually moves to the top rating if they play enough games and have enough skill to keep winning i.e. the top rating means nothing

Fixed that for you, b/c I know you didn't leave it out intentionally

[bad arcade kitty]

MechWarrior1086091, on 27 September 2015 - 05:44 PM, said:

Fixed that for you, b/c I know you didn't leave it out intentionally

no, you don't need to keep winning more than losing

you can wallow on the same place and since you don't lose psr on losses when you did well (and it happens to everybody) it will eventually ensure that you go upwards

in the games with a proper rating you raise your rating only if your performance significantly improves

[MoonUnitBeta]

Your rough suggestion kinda look like this? (I revised pauls chart)

[Mister Blastman]

I think PSR is too generous as it is. There should be more downrating.

[Satan n stuff]

MechWarrior1086091, on 27 September 2015 - 04:01 PM, said:

PSR is not a measure of how good at killing you are.

It is a measure of how good you are at making it easier for other people to kill.... aka being a good team mate.

/thread

A 50+ alpha to a CT makes it a hell of a lot easier for someone to kill a mech than several such alphas sprayed all over the mech, your argument is invalid.

Edit: I'd say the damage mechanic can be salvaged, if it were to factor in it's effectiveness. It wouldn't need to be entirely accurate, as long as it rewards damage to the most important sections of a mech the most.

For example:
All sections are valued by (item tonnage/total hitpoints+shield value/item tonnage)*multiplier, each point of damage will be multiplied by that to add to the match score. Match score is awarded per destroyed item or component, to anyone who damaged that component.
The CT will be valued at the total remaining value of the mech at the time it's destroyed.
The first leg to be destroyed has the engine tonnage included in the value, minus whatever engine tonnage would be needed to move at legged speed. The remaining leg will be valued at the total remaining value of the mech, like the CT.
The sidetorsos will be valued like the CT if the engine is an IS XL, in the case of a clan XL the second sidetorso to be destroyed will be valued like the CT. The first sidetorso to be destroyed with a clan XL will have a shield value added, so will both sidetorsos on mechs with a standard engine.
The arms will also have a shield value added.
It's also possible to value remaining ammo and heat sinks ( collectively for each type of item ) based on the value of the weapons that depend on them, so for example 3 tons of AC/5 ammo might be worth a third of an AC/5 if it exists on a mech in that combination.

A system like this will obviously never be perfect, but it will give significantly better rewards for effective damage than for ineffective damage.

Edited by Satan n stuff, 28 September 2015 - 02:17 AM.

[Galenit]

Devayner, on 27 September 2015 - 02:04 PM, said:

I think a lot of the UAV kill and the scouting /spotting rewards need a upgrade. As its to biased to damage.It feels like a light gets the shaft, if you want players to work as a team then make the rewards worth it and they will.There is no point in taking a light when i can roll a dashi and skate through on damage alone fluffing my match score.

After some time (some month, i was out, because of a stroke i have still to cure, to some degrees, if its possible,....) i played this weekend.
A narc raven need a lot of work to get the 300 points, removing the narc and putting bigger lasers or missiles makes it a lot better, but for damage, there are better mechs.

My e-wars, utility, scouting raven has still no role in this game.
There is still only one role: damagedealer

Edit:
Fun thing about psr:
I played only a few games this year, so i got t4.
Played my stalker (first match after psr), i went to t3, played my raven, went back to t4.
My tier was switching every few games depending on the mech i used.

Edited by Galenit, 28 September 2015 - 02:44 AM.

[sycocys]

I agree, the cutoff to break even on a loss is no where near close to accurate for skill/effort inputted into the match considering the reliance on 11 other players and how quickly/easily 2-3 bad players can drop the rest of the team's scores by 50% with their rambo rushes and 50-70 damage.

[Reptilizer]

MechWarrior1086091, on 27 September 2015 - 04:01 PM, said:

PSR is not a measure of how good at killing you are.

It is a measure of how good you are at making it easier for other people to kill.... aka being a good team mate.

/thread

Join me at tier4 and you will realize that actually the opposite is true...
A high matchscore will never get you a drop in rank and winning the game is near random.
Making it possible for other people to kill might well result in a low match score for you, giving no rise even if you win and resulting in a big drop if you lose...
The thing you can reliably influence is your own match score -> opportunistic play for the win!

[pyrocomp]

Paigan, on 27 September 2015 - 01:09 PM, said:

Win:
Player score < 100: downrate
Player score in [100;200]: keep even
Player score > 200: uprate

Defeat:
Player score < 200: downrate
Player score in [200;400]: keep even
Player score > 400: uprate

Right now, the following is true:
Win:
Player score < 100: unchange (downgrade and see no scouting lights, not good, positioning info in the beginning is vital)
Player score in [100;250]: low rise
Player score > 250: rise
Player score > 400: signiificant rise (note 15 mins delay in data update)

Defeat:
Player score < 250: downrate
Player score in [250;400]: unchanged
Player score > 400: ??? (haven't noticed yet)

So current scoring is not much different from yours. The factoring of the team win/loss is quite a strong suppressor to the ramboing and solo-running. Also punishes NASCAR drivers, which is good.

By the way, LRMs are a good suppression weapon forcing mech under fire to back and seek cover. This is really relevant, plus to deal 1000 dmg in LRM boat is no easier than in Gauss-Laser Dire, try to do so, and tell us just how many tries it takes to get that score in a pure LRM boat in pub queue, and how many more matches to repeat that.

Reptilizer, on 28 September 2015 - 02:54 AM, said:

Join me at tier4 and you will realize that actually the opposite is true...
A high matchscore will never get you a drop in rank and winning the game is near random.
Making it possible for other people to kill might well result in a low match score for you, giving no rise even if you win and resulting in a big drop if you lose...
The thing you can reliably influence is your own match score -> opportunistic play for the win!

To get no rise to your PSR in a win you should get less that 100 match score. That is... well, not a high requirement, really.

[Reptilizer]

pyrocomp, on 28 September 2015 - 03:02 AM, said:

*snip*
To get no rise to your PSR in a win you should get less that 100 match score. That is... well, not a high requirement, really.

While it is not a high requirement, about a quarter of all players in the matches i participated fail to reach this threshold...

[pyrocomp]

Reptilizer, on 28 September 2015 - 03:06 AM, said:

While it is not a high requirement, about a quarter of all players in the matches i participated fail to reach this threshold...

That's sad... I can do that only one way - running alone headfirst into the enemy deathball (in any mech), but that rarely happens.

PS: or I can be left alone in a Dire by NASCAR-lovers, which is a same thing, come to think of.

Edited by pyrocomp, 28 September 2015 - 03:10 AM.