Hi folks. I'm CCP Masterplan from Team Five-0, and I'm here to talk about a significant change to the aggression system in EVE coming along in this winter's expansion.

I'm going to cover a bit of history, a discussion of why we're making these changes, and then get in to the details of what the changes will be. The current system is notorious for its lack of documentation, both publically and also internally within development. This dev blog is the first step towards improving the public documentation of the game's aggression mechanics. Therefore it is going to be quite long, as there are a lot of details to explain. If you read nothing else, I suggest you at least look at the “Summary of significant changes” paragraph further down, as some previous assumptions are getting changed.

Previously, from Five-0...

Back at Fanfest in March, we first talked about what Crimewatch is, some of the problems it has, and some of the changes we wanted to make. During the summer releases of Escalation and Inferno, we undertook significant behind-the-scenes work to get Crimewatch in to a better state, but with the constraints of keeping the design changes to a minimum. This put us in a position where we could then embark on more significant changes at the design level. With the Retribution expansion, we are going to be shaking up some of the rules, making improvements to user visibility and understandability of the system, and taking the opportunity to close a number of ‘creative abuses’. In the old system (which I'll refer to as CW1 for brevity), much of the behavior was built upon Aggression Flags. These flags were between two parties (characters, corps, factions and others). These flags could be triggered in a wide variety of ways, from direct activation of your guns, through stealing items from a container, to simply being ‘observed’ by an invisible NPC controller. Sometimes the aggression flags were visible in the client, sometimes they weren't. Attempting to document the complete behavior of the CW1 aggression flagging system was a daunting task; attempting to make design and code alterations even more so. The system also scaled unevenly in terms of server performance: Low-sec space often came off the worse, because it sometimes saw major fleet fights, and also had to handle the legality issues of Global Criminal Countdowns.  

When we started on this project, we identified a number of outcomes that we wished to satisfy:

• The new criminal flagging system should be easy for players to understand
• The new criminal flagging system should be easy to developers to work with and unlikely to break
• The new criminal flagging system should not impose a significant performance load

Note that we're not attempting to dumb down the system, or restrict what you can or cannot do. In fact, these changes should empower you to find even more creative solutions to the problems that other players might present you, and punish them accordingly. Our overriding design goal for this feature is to “tame the beast”: the root of all evil in the current system is that it's just too complex, so our number one goal is to simplify. In terms of balance, we have a secondary goal to try and maintain the current “balance of power” as much as is possible within the scope of the main goal. We're attempting neither to make space safer, nor make it less safe, but rather to come up with a system with approximately the same consequences as the current system but in a much simpler package. This process is necessarily imperfect, and we're inevitably going to find that some things are slightly easier and other things are slightly harder. This is regrettable, but we're absolutely not going back down the rabbit hole and re-complicating the new system to solve comparatively minor imbalances, because the overall cost is just not worth it.

If we were making a theme park game where your actions are tightly circumscribed this could be problematic, but thankfully we're making a pretty open-ended sandbox, which gives us developers a lot of confidence that you players will find new ways to do old things pretty quickly.

Introducing some new Flags

A core concept of CW2 is Flags. We aim to decouple Actions from Consequences via a flagging system. For each character, we track a number of flag types. Performing certain actions will cause yourself (and possibly others) to pick up one or more types of flag. Having a flag will cause consequences for you and others around you. Flags are always per character. There is no character-to-character flagging any more. (But see Limited Engagements further down) We want to make sure that you know your own flag states at all times. A lot of work has gone in to making sure we can keep your client up-to-date at all times, whilst minimizing the overhead on the servers. The following image shows a character with three flags. From left-to-right: PVP flag, Suspect flag, Weapons flag. Each flag also has an associated timer, shown by the clock around it. The specifics of what each flag means will be explained later.

Some actions that will give you a flag are long-lived. For example, activating a turret module against a player will give you a PVP flag. As long as you have one or more relevant modules active, the timer will be held at the start, and the countdown will not commence. Once the last module has been deactivated, the timer will begin counting down. Restarting the module will obviously move and hold the timer back at the start. Other actions are instantaneous ‘one-shot’ events, such as stealing from a container or getting hit by a smartbomb. On such an event, the related flag will be triggered and immediately begin counting down. Triggering the event again will restart the countdown. In most cases, using assistance modules on another player will cause you to inherit all of his flags, including whatever their current countdown state is, if the flag has a higher time remaining than your own. Assisting someone with 10 seconds left on his Weapons Flag will give you a 10 second Weapons Flag (as long you your current flag doesn't have more than 10 seconds on the clock). The system will work in such a way that a pair of mutually-assisting logistics ships won't lock each other in to being perma-flagged (as long as they themselves aren't also aggressing), but will have their timers count down in sync with any combat ships they are assisting. In general, logistics ships should be able to jump through gates at the same time as their fellow combat ships.

Summary of some significant changes:

Note: By ‘offensive modules’, I mean turrets, missile launchers, electronic warfare modules (such as webs, neuts, ECM), as well as drones with similar effects. ‘Assistance modules’ refers to logistics modules such as shield transfers, remote armor reps, sensor boosters etc., as well as drones with similar effects. Here's a brief summary of what the different flags mean:

Weapons Flag: This flag is activated by using offensive modules against another player (or simply by activating certain non-targeted weapons such as smartbombs). Having this flag will prevent a character from performing actions such as jumping, docking and switching ships in space. This flag functions in all areas of space.

PVP Flag: This flag is activated when one player uses offensive modules against another. The initiator of the action will get a PVP flag. If the recipient is a piloted ship, then the owner of this ship will also get a PVP flag. Having this flag will prevent a ship from being removed from space if the pilot logs off. This flag functions in all areas of space.

NPC Flag: This flag is activated when a player uses offensive modules against an NPC (or vice-versa). Having this flag will prevent a ship from being removed from space if the pilot logs off. This flag functions in all areas of space.

Legality Flag: This flag exists to penalize those who commit criminal/suspicious actions in Empire space (high and low sec). A character with a Legal flag is always a legal target for offense from all other players. This flag has two severity levels: SUSPECT and CRIMINAL. The severity depends on what action is performed and where it happens. The severity also controls what consequences can occur (CONCORD will always attack CRIMINALS but not SUSPECTS, for example). This flag functions only in Empire areas of space. (But you can't shed it by jumping in-and-out of null). The CRIMINAL severity will always override SUSPECT.  

Here's some of the important changes compared to the old system that players should be aware of:

• Illegal attacks on ships (not capsules) in low-sec only incur a Suspect flag. No CONCORD response if the attacker subsequently jumps in to high-sec.
• Stealing from a container will expose you to potential attacks from all players (but not from sentry guns). The existing rules for what constitutes 'legal access' to a container are the same (I am the owner of the container, I am in the corp registered to the container, I am in the fleet registered to the container, The container is Abandoned), but we are adding one additional rule: If I can legally attack the owner of a container, then I can legally take from the container.
• Using assistance modules will pass on all flags to the assistor, possibly preventing them from docking/jumping for the same interval as their assistee
• It is possible to be prevented from switching ships or ejecting (whilst in space) by your actions
• After losing a ship and entering a capsule, players will still be restricted from docking/jumping for up to a minute (if they have an active Weapons flag).
• There is no A-to-B aggression flagging anymore, just global flags. (Well, sort-of, see Limited Engagements)
• Assisting an outlaw in low-sec (outside of a combat situation) will not be penalized
• Security-status penalties are now ‘front-loaded’, so a criminal/suspect will incur the full penalty when an illegal attack starts, not when (if) the target is destroyed.

Flags in detail

The following two charts show which actions cause which flags (and where), and then what the consequences are for having a flag.

click image to enlarge

click image to enlarge

We're still tweaking the contents of these charts, so not all decisions are finalized. If there's a case that you think isn't covered, then please let us know about in the comments section. Hopefully the fact that we can actually explain much of the new system in this way shows how improved things are compared to the old system with all of its special-cases and exceptions.

Limited Engagements

The personal-flags system tidies up a lot of problems with the old system, but still leaves us with a couple of cases that aren't covered. The main one is that a suspect can be freely attacked, but he has no way to defend himself from attack without committing further crimes. We want to ensure that a player always has a right to self-defense, even if he is A Bad Guy. To solve this, we still require a form of A-B flagging. However this will be heavily limited in application, and won't be propagated via assistance chains like the existing aggression flags are. This is where we introduce the concept of a Limited Engagement. An LE is between a pair of characters. (Always characters, not corps, alliances, factions or anything else). An LE gives each party a legal right to attack the other, without triggering any Legal flag. An LE is ACTIVE as long as offensive actions are on-going. Once offensive acts have stopped, it will begin to count down. Resuming hostilities will reset the timer. If the timer expires (probably 15 minutes but still TBC) then the LE is ended. An LE is created when character A attacks character B, and where B is globally-attackable due to being a Suspect, Criminal or Outlaw. This then allows B to defend himself against A. Like Criminal and Suspect flags, An LE is only effective in empire space. Assisting someone who is engaged in an LE will cause the assistor to receive a Suspect flag. This is to prevent neutral logistics interfering in ongoing combat without risk to themselves.

Kill rights

Performing an action against another player that gets you a Criminal flag will also award a kill-right to that person. This will happen regardless of whether or not the target ship was destroyed. This will feed in to the revamped bounty system that Team Super Friends will be talking about very shortly, so look for a dev blog coming from them soon.

Next time, from Five-0...

Team Five-0 have a few more features lined up that we'll be talking about in the run up to Retribution, such a Safety System to prevent Accidental Condordokken that is better than modal dialogs. We're also working on a replacement for the usage of loot-theft as a way to initiate consensual 1v1s without incurring criminal penalties that we hope to release for Winter. Keep an eye out for an announcement about when these changes will become available for public testing. In the mean-time, please let us know what you think about these changes in the comments thread. We'll be reading it all, especially posts with useful feedback in them. (Here's a guide to making a Good Post by my good friend CCP Fozzie)