Player Features and Ideas Discussion

On the test server there has been the introduction of the concept of a Weapon Accuracy Score for turrets. (Thanks Ja'e.) This new number utilizes both tracking and signature resolution to create a score that's directly comparable across different turrets. Previously, a gun with 0.12 rad/s tracking could have extremely different accuracy compared to a different gun with 0.12 rad/s tracking if the two guns had extremely different signature resolutions. With this new score, a gun with 40 WAS is known to be exactly as accurate as a different gun with a 40 WAS.

However, it's now more problematic when comparing weapon accuracy to what's happening on the grid. Currently, if a pilot knows that their gun has a similar signature as their target, they know comparing the gun's tracking to the target's angular velocity is a very good indicator of the to-hit chance. With the Weapon Accuracy Score however, there's nothing about the ship that makes an easy comparison. A cruiser with 120 meter signature orbiting 1300 meters away at 200 m/s (0.15 rad/s) doesn't translate to an obvious to-hit chance when using a 40 WAS gun. Without doing some math it's not obvious whether it's 10% chance or a 90% chance. All we know is that a 90 WAS gun is more accurate than a 40 WAS gun. (It's a 32% chance to hit in this example, btw.)

My proposal is to add an "Evasion Score" as an overview column. This Evasion Score would be calculated exactly like how the Weapon Accuracy Score is calculated. That is, 40000 * angular velocity / signature. In the above cruiser example, the ship has a 51 Evasion Score, which is slightly better than the gun's 40 Weapon Accuracy Score, so it's immediately obvious the gun has less than a 50% chance to hit. (When the two scores are equal, there's a 50/50 chance to hit.) This isn't a static value that can be added to the stats since it changes depending on the ship's speed and direction, which is why it needs to be an overview column.

This will also help out with target painters, since as they dynamically effect a ship's signature radius they would dynamically effect a ship's Evasion Score. Similarly, MWD penalties would also show up on the overview. This is a very synergistic way to represent the turret chance-to-hit equation. If the above cruiser were to turn on its MWD and bloom its sig radius from 120m to 720m, it would have an Evasion Score of only 8.5, barely anything, until it was able to get up to full speed.

The problem with evasion score is that a good portion of that revolves around how you fly your ship. If you're flying straight at the person shooting at you, you're easy to track. If you're orbiting at 50km sniping, you're easier to hit than if you're at the same speed orbiting at 5km.

And yes, Robbert, while adding the transversal column to your overview is rather silly, it is still important to know how your transverse velocity impacts your opponent's ability to hit you.

And when you change even one of those variables, the evasion score changes. You're looking at recalculating the evasion score every second for every ship on grid with every other ship in space on the entire server. That's a lot of load on the server there.

Edit: Also, the formula used to determine tracking hit chance is a lot more complex than it appears you think it is.
Edit 2: Thanks for being patient if you read the edit before this one.


This is incorrect. According to the wiki on turret damage, specifically the part about the to hit equation, radial speed is how the ship is moving along the radius of the circle around your ship. AKA, it is the movement towards or away from your ship, not it's movement as it transverses the circle.

This is all for the edits for this post. Again, I apologize for the disjointed quality of this post.

It's not really hard to figure out. Let's use a moderate sized fleet engagement of two fleets, one with 100 pilots and one with 101 (for ease of calculation in the example).

For one ship, the server would have to recalculate the evasion score every second as it relates to each of the 200 other ships. This has to be done using the variables from each ship, as each one can potentially be different. This then has to be repeated 200 more times, for a total of 40,200 additional operations every second on top of the work the server already does. This is just for 201 ships, this gets magnified greatly when you account for the fact that this calculation would have to be done for every single ship in space as it relates to every other ship, NPC, structure, etc. that share the same grid.

With tens of thousands of players spread over seven thousand systems, you'd be looking at adding a few million more calculations per second to servers that already have to use tidi to manage the load at times.

Edit: Oh, drones. Can't forget drones. Add 5 drones to each of the ships in the example fleets. Can you do the math to figure how many calculations it'd take to work this when you add a thousand drones into the mix?

Someone could probably even make this as an overlay, that works in real time, that is if it wouldn't get them banned.

Further, with the information the overview currently provides in addition to stats from the guns, ammo and ship bonuses combined with an estimate as to the targets sig radius, (you don't know if they are using shield extenders, halo implants, so this will only be an estimate) and display an estimated chance to hit.

TIL: calculation don't require CPU cycles.

It's a behemoth when you realize each entity in space has to be calculated in reference to each other entity in space on the grid since the game has no way to know who is an enemy. Those entity include everything from ship to drones and cans because they can all get shot. Hundred of entity on grid create thousands upon thousands of 1v1 relationship that has to be calculated at all time on top of what is already chocking the server.

You client side idea is also only viable if CCP accept to let the client do any calculation instead of leaving everything server side like it is right now.

It's not like the server has to feed us all the sig radius data so our CPU can do the math right? Sending hundred if not thousand of value to each client every ticks surely won't add any load...

Where do you think rad/s come from...?
Especially considering that rad/s is relative to the the person's ship. i.e. it's calculating that for every person with an overview, for every ship on grid.

I though we only had small, medium or large and not the actual numerical value.