Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[orange]

Frankly, Tarunik's ISK efficiency is the most meaningless killboard stat of all (it's a ridiculous 98-odd% due to the LE supercap kills he's been on).  I value much more the :goodfights: that someone can get (their war stories, in other words) than their raw stats.  Sadly, stats-worship has led to people who simply won't give you :goodfights: any longer, and just blue-ball and/or uber-blob (anything more than say 3:1 to 5:1 numbers, given fairly evenly matched fleet comps) you.

I think the main answer to that would be to rework kill statistics (ISK efficiency in particular) so that blobbing wouldn't be so productive in that regard.  Sadly, the last time this was tried (by Agony Unleashed, no less), they were forced to make their killboard private due to unceasing howls of "URDOINITWRONG".

As a side note: what would be a good place to discuss a revamped ISK efficiency calculation?

So, this is an awesome question!

Isk Kill/Isk Lost lacks a lot of nuance to it.  Are the majority of your kills solo or do you only ever get kills in larger blobs?

Perhaps an initial starting point is to calculate your Isk Killed by multiplying each kill by the percent (%) damage you caused.  This does not capture the whole picture, since EW inflicts no damage, but it is a starting point.

This would immediately lead to people considering the kinds of numbers they are bringing along.   When you have 100 people on a KM, most pilots contribution is likely minimimal.

"Good job, 50 of us alphaed the Battleship! 100m isk for everyone's KB stats!"  vs "What the ...  I helped kill that Battleship and my KB stats only went up by 2m!  I am not flying with you again."

It would be interesting to try and capture the impact of EW & Logistics, but lets start somewhere easy.

[tarunik]

As I see it, there are 4 choices for "weighting" ISK efficiency by participant count:

• Uniform weight (1/number of participants)
• Weight by raw damage (damage done/total damage)
• Weight by adjusted damage (damage done/total damage + some EWar adjustment)
• Point weighting (assign points based on hull class, weapon of record, and probably damage done as well, then use points assigned/total points as a proxy weight)

[Misan]

To some extent the points system that KBs (eve-dev anyway) use currently already weighs against blobbing. The problem with it is that the points are only used to measure weekly and monthly top scorer awards. I don't know the actual formula, but it seems to use a simple uniform weight based upon the ship hull divided by participants involved. Looking at some of my recent kills there are the more blobby kind vs a more balanced fight yields very different scores for similar hull types. I think this type of point system is preferable as it doesn't make assumptions about which pilots contributed the "most" to any given kill. There simply are way too many things that influence that in a fight for a killboard to be able to express it accurately. Otherwise obviously the FC and scouts should always get the most points per kill. 

I think relying on damage measurements is badly flawed, both because KMs have (and remain) inconsistent on damage reported and because it marginalizes important roles. Any kind of tackle (scout or heavy tackle), EWAR, logistics, and neut ships lose out when using damage metrics for efficiency.

[tarunik]

To some extent the points system that KBs (eve-dev anyway) use currently already weighs against blobbing. The problem with it is that the points are only used to measure weekly and monthly top scorer awards. I don't know the actual formula, but it seems to use a simple uniform weight based upon the ship hull divided by participants involved. Looking at some of my recent kills there are the more blobby kind vs a more balanced fight yields very different scores for similar hull types. I think this type of point system is preferable as it doesn't make assumptions about which pilots contributed the "most" to any given kill. There simply are way too many things that influence that in a fight for a killboard to be able to express it accurately. Otherwise obviously the FC and scouts should always get the most points per kill. 

I think relying on damage measurements is badly flawed, both because KMs have (and remain) inconsistent on damage reported and because it marginalizes important roles. Any kind of tackle (scout or heavy tackle), EWAR, logistics, and neut ships lose out when using damage metrics for efficiency.

Sadly, the points system is indeed under-utilized.  Perhaps a ratio of "points earned to points lost" might be a better overall statistic to look at than one's raw ISK efficiency? (Also, not all EDK setups provide point values; my alliance's killboard doesn't, for one.)

Agreed that damage measurements aren't terribly useful here.

[Esna Pitoojee]

One of the biggest troubles when talking about 'ISK efficiency' or 'ISK value' of losses is the inherent instability and manipulation of the EVE market.

For starters, how is an ISK value for an item stablished? Value at closest market hub? Value at all 4 major market hubs, averaged? Build cost? How is that build cost found? What about market manipulation that could suddenly drive the average for an item far, far lower or far, far higher?

How often, then, would killboards update their internal item values? Would the updates be retroactive?


This would requie a LOT of planning and coding to try and produce a reasonable solution to.

[tarunik]

One of the biggest troubles when talking about 'ISK efficiency' or 'ISK value' of losses is the inherent instability and manipulation of the EVE market.

For starters, how is an ISK value for an item stablished? Value at closest market hub? Value at all 4 major market hubs, averaged? Build cost? How is that build cost found? What about market manipulation that could suddenly drive the average for an item far, far lower or far, far higher?

How often, then, would killboards update their internal item values? Would the updates be retroactive?


This would requie a LOT of planning and coding to try and produce a reasonable solution to.

Eve-Kill uses Eve-Central for market prices (I suspect it uses the median price Eve-Central provides, and median = mean if you assume a symmetric distribution...) and some other site for faction/... modules and ships.  I think EDK does apply price changes retroactively as it has a single itemID -> price table though.

[tarunik]

Speaking of broken killmails:

4 people shoot at a raven, but only 2 get on the KM? Wha?

[Esna Pitoojee]

Speaking of broken killmails:

4 people shoot at a raven, but only 2 get on the KM? Wha?

I've seen this recently as well - it seems to have appeared with Crucible. Did the raven jump through a WH during the course of the fight?

[tarunik]

Speaking of broken killmails:

4 people shoot at a raven, but only 2 get on the KM? Wha?

I've seen this recently as well - it seems to have appeared with Crucible. Did the raven jump through a WH during the course of the fight?

It sure did.  The next KM that involved a ship jumping through a WH and back came out OK though, so maybe it has something to do with the WH collapsing during the fight?

[Esna Pitoojee]

No, it's a session change thing - if you begin to fire on a ship and it does something that causes a session change timer - docks and undocks, jumps a stargate, jumps a WH, cynos somewhere, etc... - anyone shooting at it before the session change timer doesn't appear. I don't know if this was even partially intentional, as it smells of a change that ended up overreaching to me.

[tarunik]

No, it's a session change thing - if you begin to fire on a ship and it does something that causes a session change timer - docks and undocks, jumps a stargate, jumps a WH, cynos somewhere, etc... - anyone shooting at it before the session change timer doesn't appear. I don't know if this was even partially intentional, as it smells of a change that ended up overreaching to me.

Not in all cases.  We had a more recent Drake KM that came out correctly formed despite several participants jumping through and back.

(P.S. #1: points are broken on EVSCO's boards.  :karbocode:)
(P.S. #2: EVSCO is down. :karbocode:  )

(P.P.S.: Apparently Eve-Kill goes down nightly due to Karb's database backup scripts hogging the database   :karbocode:

[Bacchanalian]

Is this where I wave my dick?

http://www.rotekapelle.com/killboard/?a=alliance_detail&all_ext_id=927292903&view=pilot_scores

But seriously, I have no idea how this shit gets calculated.  I'm quite far from 1st in my alliance in raw killmails.

http://www.rotekapelle.com/killboard/?a=alliance_detail&all_ext_id=927292903&view=pilot_kills

Kinda odd.