New Kill Scoring logic

[Guest deleted@83466]

Apologies if this has been discussed elsewhere,

 

It's my understanding that 3.0006 introduced a new kill scoring logic.  The thing is, it's seems to be often times allocating credit where it really isn't due.  For example, earlier tonight in MP, I bounced a guy, and only got a few minor hits.  The enemy was still very very much alive and continued to dogfight vigorously.  Another guy came in and blasted the heck out of him, tore his wing off, and did far more damage than I had done.  I got the credit for the kill.  I didn't deserve it; the other guy did, pretty unambigously, even though I happened to be the one that got the initial hits.  This kind of situation seems to be happening quite a lot lately since the update.

 

Now admitedly, I suppose it all balances out in the end:  You get credit for some you don't deserve, and sometimes somebody else gets credit and they don't deserve it.  So, I don't consider any of this a big deal, but still, something doesn't seem quite right about it.

 

What is the new logic behind kill allocation now in 3.0006?  Is it still a Work In Progress?  It seems like the kill logic was working just fine before, so what is the reason for the changes?

Edited October 30, 2018 by SeaSerpent

[VBF-12_Stick-95]

Update 3.006 occurred September 26

Quote

56. Statistics system accuracy increased (this should fix the situations with unaccounted kills)

 

The new kill logic was announced in Dev Diary 207 on October 12

Quote

Speaking of more imminent changes, we plan to improve the in-game statistics and 'who gets the kill' logic in the next update.

 

So no, the new kill logic is not out yet.

Edited October 30, 2018 by VBF-12_Stick-95

[Guest deleted@83466]

29 minutes ago, VBF-12_Stick-95 said:

Update 3.006 occurred September 26

The new kill logic was announced in Dev Diary 207 on October 12

 

So no, the new kill logic is not out yet.

 

Yeah, I thought so too, and therefore was hesitant to make this topic, but as you point out, 3.0006 does contain item #56 " Statistics system accuracy increased (this should fix the situations with unaccounted kills)".  So, I guess my question should more accurately be, what exactly did #56 do?

 

Honestly, I've only noticed this happening in the last couple of days, not all the way back to Sep 26.  I won't rule it out, but I don't **think** it's all in my imagination, because I've been on comms with several different people who've all commented on it just this last week.   It could be though, if nobody else notices it.  (maybe I'm having a Raaaid moment? ?)

Edited October 30, 2018 by SeaSerpent

[EAF19_Marsh]

I thought #56 was “Troll all Jedi” ahead of the infamous #66?

[RedKestrel]

8 hours ago, SeaSerpent said:

 

Yeah, I thought so too, and therefore was hesitant to make this topic, but as you point out, 3.0006 does contain item #56 " Statistics system accuracy increased (this should fix the situations with unaccounted kills)".  So, I guess my question should more accurately be, what exactly did #56 do?

 

Honestly, I've only noticed this happening in the last couple of days, not all the way back to Sep 26.  I won't rule it out, but I don't **think** it's all in my imagination, because I've been on comms with several different people who've all commented on it just this last week.   It could be though, if nobody else notices it.  (maybe I'm having a Raaaid moment? ?)

It says it solves the issue with 'unaccounted kills', I read that as kills that weren't awarded to anyone. In this case it was awarded to the wrong person.

In general the game sees a lot of things as 'fatal damage' where people have still managed to get their crate back to the airfield. in this case I would imagine that the game considered the little damage you did to be fatal in some way (maybe you got hits on the engine or radiator that would have eventually killed the craft). The guy who came in and tore his wing off is considered to be hastening the aircraft's demise.

We haven't had any updates in the last few days, so anything that's changed would have had to have been in place a lot longer than that. It could be just coincidence that its being noticed more now. 

[-TBC-AeroAce]

@SeaSerpent my experience is the opposite of yours in most cases. You can inflict major damage and somebody gets a little mg hit and then they get the kill. 

[PatrickAWlson]

From a SP point of view the logs often contain unidentified victories.  Sometimes this is the result of AI lawn darting, other times delayed death, and still other times ... who knows.  I have had to write a lot of code to do fuzzy assignment.  Whatever they do, I hope they don't muck up the logs as I am completely dependent on them.

[Guest deleted@83466]

Happened again tonight, as it has every night, for the past half week, on WoL.  In the parser stats, you can see that when an aircraft is hit by multiple people, the "catastrophic" damage (Like when you blow a wing off and the damage is 92.5% as opposed to 0.2% or something) is no longer attributed to a player, so whoever got the most fractional damage up to that point (even if it is extraordinarily minor) is the one that gets the kill.  Here is an example:  http://il2stat.aviaskins.com:8008/en/sortie/log/3666109/?tour=40  Since it only started happening recently, I can only assume that they made a recent change in the way the master server reports kills.  I doubt that the guys at WoL changed their parser script, and I'm almost positive this was not happening at the release of the Sep 26 update, but much more recently, so that's the only thing I can figure.  In any case, I'm 100% sure there has been a change.  Whatever it is going on, the game no longer is doing a very good job of assigning the kill to the right person.

Edited October 31, 2018 by SeaSerpent

[1PL-Husar-1Esk]

This might prevent shoulder shooting. Other hand I don't care much about been not scored but if someone already damaged enemy plane, in 80% this is more important than that final blow I think.

Edited October 31, 2018 by 307_Tomcat

[Guest deleted@83466]

1 hour ago, 307_Tomcat said:

This might prevent shoulder shooting. Other hand I don't care much about been not scored but if someone already damaged enemy plane, in 80% this is more important than that final blow I think.

 

I care about scores.  I see where you're coming from Tomcat, but we aren't talking about some kill stealer putting the last bullet into a plane that is already almost guaranteed dead;  we are talking a plane that is still very much alive, perhaps even with minor damage, not being credited to the person that really comes in and blows him to smithereens.   Trust me, whatever has changed, this new scoring method is not discouraging anyone from shoulder shooting or anything like that, it's just simply not awarding kills to the right people, and that can't be a good thing.   So I cause 5% damage to a guy (like a fuel leak), and then the next guy comes in and causes 4% damage + 91% damage from the 20mms he placed in the guys wing root, and tore him to shreds, and I get the kill credit?  That can't be right.

Edited October 31, 2018 by SeaSerpent

[1PL-Husar-1Esk]

There is no perfect solution ,but if my attack (first to the enemy) is successful in average is more that 5%  Damaged plane is easy to finish most times. We can have only  shared kills if more that one do damage but nobody will get that kill and ppl would not like it, credit kill by percentage this is most unrealistic to RL but fair , computation and net performance cost unknown. 

Edited October 31, 2018 by 307_Tomcat

[InProgress]

I would like to see things like damaged, probably destroyed For multi, maybe just add thing like coop kill. So both of you get a kill but in diffrent category.

Edited October 31, 2018 by InProgress

[PatrickAWlson]

The right answer was posted above.  This game cannot use a simple percentage based system to determine who should be credited.  The enemy may be fighting on and may do so capably for awhile, but if he is leaking coolant or fuel, or has suffered some engine damage that degrades performance,  that could be viewed as a critical hit.  It is possible that the initial hits weakened the structure and the guy who came along and finished the job really just added the straw that broke the camels back.  There are a lot of scenarios where appearance does not correspond to facts.

 

Or maybe the algorithm is just hosed - who knows  

[Guest deleted@83466]

4 hours ago, 307_Tomcat said:

There is no perfect solution ,but if my attack (first to the enemy) is successful in average is more that 5%  Damaged plane is easy to finish most times. We can have only  shared kills if more that one do damage but nobody will get that kill and ppl would not like it, credit kill by percentage this is most unrealistic to RL but fair , computation and net performance cost unknown. 

 

Well, I hope this change is only an interim change, or a work-in-progress, until it is completed in the fix announced on the 27th of October, because how it is now seems neither fair nor realistic.

 

Since it only became an issue in the last week, not with the Sep 26 3.006 update, I'm correct in deducing that this was a change in something on the Master server within the last week, Tomcat?

[Aap]

14 minutes ago, SeaSerpent said:

Since it only became an issue in the last week, not with the Sep 26 3.006 update

 

Online statistics logic, created by Vaal, does not necessarily count things the same way as the game does internally. So does not have to be related to a new game patch.

Edited October 31, 2018 by II./JG77_Kemp

[Guest deleted@83466]

28 minutes ago, II./JG77_Kemp said:

 

Online statistics logic, created by Vaal, does not necessarily count things the same way as the game does internally. 

 

The parser stat on the game server is the only score I care about, not the in-game score, but in this case, the logic of in-game kill credit and what shows up on the parser is consistent with the eyebrow-raising results being seen as of late.  I don't think the script changed, and I'm not sure it was something in the 3.0006 update, so my only guess is something on the master server was changed, and very recently.  As in the stat I posted above, the chief change appears to be that the "kill shots", the ones that cause something like structural failure and is accompanied by high damage percentages seems to no longer be assigned to the shooter.  Thus someone who comes in with a well aimed, effective volley of fire that tears a still-viable target to shreds is being counted as less important than somebody who might have inneffectively plinked away at the target beforehand.

Edited October 31, 2018 by SeaSerpent

[BraveSirRobin]

Just now, II./JG77_Kemp said:

 The last line "damage - 81.9%" does not have a "responsible" after it.

 

Prior to this recent change it did.  That’s the whole point.

[Aap]

13 minutes ago, BraveSirRobin said:

Prior to this recent change it did.  That’s the whole point.

 

The point of my deleted post (but you were quick to respond) was that if SeaSerpent wants to check the cause, he could check if the parser itself has changed, or if Vaal's logic just  does not count the last one that hit the enemy as an "active" one doing the final damage. So, from the link posted by him previously, the sum of the damage caused by his bullets is higher than the sum of the damage caused by your bullets, even if your bullet was the last one to hit the enemy after which came the 81,9% damage. So, if he does not have the log files to check from there directly, could do some detective work and see if similar change happened in TAW stats at the same time.

Edited October 31, 2018 by II./JG77_Kemp

[BraveSirRobin]

I’m pretty sure Vaal has never made changes to his program without announcing them.  And this wasn’t changed by the last update.  There really is only one possible source.  The only question now is whether this is the final version.  If it is, they didn’t make things better.