Watching New Eden Open’s broadcast today where three Malices got killed actually reminded me that I wanted to write about that.

There’s one interesting conceptual problem that comes up often, both in games like Eve and in a proper shooting war, and which, so far, I have yet to see solved conclusively: The killmail score problem. Assuming we’re all firmly in the context of treating Eve as a sport, or close by — what does a killboard actually tell us about people’s skill, and what exactly is ‘efficiency’?

The question is over a hundred years old.

The actual idea of keeping score in that fashion goes all the way back to French fighter pilots of the first World War, for whom the term "flying ace" was first coined.1 While there is some anecdotal evidence that snipers started doing something of the sort at least back in the Second Boer war,2 only the use of single-seater aircraft made it a thing, when simple statistics showed that the number of enemy aircraft shot down actually follows a very steep Pareto’s law curve.3 Aircraft didn’t come in large fleets yet, usually produced an easy to find wreck, it was relatively simple to count the kills as opposed to, say, calculating the work of a foot soldier, and writing up fighter aces made for good propaganda. That involved strict numeric qualifications for medals, so methodologies for keeping an objective count have been established. Getting called an "ace" involved having a specific number of recorded victories, and the number went up as the war progressed.

The numbers have often been blurred in the records, as every air force evolved their own rules. Shared victories in particular have been treated very differently. German units required very strict confirmation of a kill4 and usually credited singular pilots for group kills, while the British never actually published individual statistics because they thought it’s unfair for the bombers, who don’t get to shoot anyone down quite as often — their unpublished statistics were a mess, as each unit kept count according to their own rules. Some of these created the concept of a fractional victory, where each participant in a kill would get a fraction of a unit "kill" towards their score.

The Second World War saw even more of a mess on that front, and every few months a typical Russian military history forum has a flamewar regarding kill statistics — "The German aces have much higher kill counts, therefore they were better pilots." Once the flames die down, the thread usually results in rediscovery of a few important points:

  1. Their combat tactics were different.
  2. Their selection of targets was different.
  3. They kept score differently too.

Luftwaffe tactics were based on the idea of winning the air war with small numbers of skilled pilots flying the best aircraft possible, thus achieving and keeping air superiority. As a result, they would often credit group kills to the most skilled pilot of the group, who would typically be at the front and in the air for as long as he possibly could. At the same time, Soviet air doctrine for much of the war saw the role of fighters primarily in keeping the bombers and ground attack aircraft flying, and a successful bombing run was seen as more of a victory than an enemy fighter shot down. Shared kills were either credited to the unit as a whole, or counted separately, but rarely were credited to any specific pilot. Successful pilots were often promoted into non-combat positions and shuffled back home to train more pilots, while Soviet aircraft were typically designed with mind-boggling mass production in mind.5 That by itself created a target rich environment for Luftwaffe pilots, a term you are obviously familiar with. And it shouldn’t be discounted that more often than not, German pilots fought over Soviet territory and thus simply had no opportunity to count the wrecks afterwards, so partial credit would sometimes turn into full credit.

As the war rolled on, while Luftwaffe could still field their small numbers of skilled pilots in top of the line aircraft, it became apparent that in the grand scheme of things it matters relatively little. British, US and Soviet bombers eventually got numerous enough to blot out the sun, and Luftwaffe could no longer maintain air superiority. The question of "actual skill" remains a point of much contention and simply isn’t answered properly by numeric totals.

In Eve we have it easier than WWII fighter pilots, if only because we get killmails, which are about as objective proof of a kill as it can possibly get. But I still think the most commonly seen counting methodologies are flawed.6 It’s probably impossible to make them perfect, since while we get more or less perfect kill information, we have no clue regarding whether the mission has been accomplished or not — too often, the mission is just to shoot something, anyway. But there are some things I’d love to improve.

For one, just about every killboard out there that I can see credits a group kill to every participant as one kill each. The killmail API provides no method of telling who started killing the poor victim first, and whether they were all together when they started or by the time they were done, so it’s at least justified in that no single player gets the kill credit to the exclusion of everyone else. It would obviously be grossly unfair for support pilots without which the kill might not have been possible. But simply crediting everyone with a single point creates numbers so inflated that they eventually become totally meaningless, especially if someone often flies in large fleets and engages in killmail whoring practices, made possible by exactly that quirk. As time goes by, kill numbers are being progressively seen as less and less useful indicator of skill,7 and ideas like "the killboard must be green" or "ISK efficiency" become preferred indicators.

But it’s fairly trivial to keep the killboard green, just don’t undock until your neutral scout locates a target you know you can trample. ISK efficiency is just as flawed an indicator, since every participating pilot is typically credited with the full cost of destroyed ship. A gang of frigs chewing up a T3 cruiser is enough to keep each member of the gang in the green for months, even if there were ten of them or twenty, and even if they keep going through faction frigs like they’re paper towels, collectively losing far more than the cost of that one T3 that props their statistics up at 90%.

Dunno about you, but I think this is kind of silly.

If I ever get enough time to devote to that, I will probably try to siphon the eve-kill.net API output and recalculate statistics using the idea of fractional kill credit, both for the ISK destroyed and the actual unit ‘kills’. Let’s see how much it changes things and which way the global statistics turns after that.

  1. Interestingly, while English uses a direct translation ("Ace") for the concept, Russian loans the French word. ("As", sometimes "Ass") Please don’t laugh.

  2. I only have Boussenard’s word for it really. Since that war essentially pioneered modern sniping tactics, I expect that’s the earliest occurrence of the practice.

  3. That is, 5% of all pilots were actually responsible for over 50% of all kills on record.

  4. Most air combat went on over their territory, so they got access to more wrecks and witnesses.

  5. Plywood was a favoured construction material for Soviet aircraft in part because every furniture factory could then make aircraft as long as motors were available. There are quite a few stories about the fierce competition for the limited supply of motors between aircraft designers.

  6. Even as I look at my own score, which at least amply demonstrates I rarely, if ever, lose in small gangs, but lose just about as often as I win when solo.

  7. At least in my neck of Faction Warfare woods, but I hear similar notions from nullsec players from time to time as well.