Player Features and Ideas Discussion

Why is it every time this is brought up people immediately bring up scenarios that anyone with an IQ over 10 would account for before putting in this system?

I could make sure that what you suggested doesn't happen with 5 lines of code.

Why don't you consider the current consequences that make fleet combat so horrible, people would rather drop their drones and go AFK until it ends.

Except for MasterNerdGuy, you are all seriously underestimating the number of calculations needed to make this happen.

Considering that you would have to determine the model and orientation of each ship, and determine how much of the ship needs to be eclipsed in order to be considered "blocked" this calculation would be extremely complex, but lets set that aside. We can assume each ship is just going to take up a sphere the size of its signature radius.

Each time you fire a shot, the server has to check the relative radius of each ship on grid against the relative radius of your target as it appears to you because an atron right in front of you is going to be more in the way than a megathron at 50km. Lets say for the sake of the argument that you can do this for each ship in 6 calculations. Thats 6 calculations per object on grid per ship thats firing!

You think there's lag in eve now? try fighting in an asteroid field with LOS on. In fact, I bet with LOS on you couldnt get 100 people on each side shooting at each other before the server crashes.

even 200 ships on grid and nothing else would cause massive amounts of lag.

After reading up a bit, i do have to admit, it's not as simple as i thought it would be. Too bad... I still think the mechanic itself would be a pretty good idea. Though, i still think there should be a much more efficient solution somewhere... AFAIK Sins of a Solar Empire has something similar going on between a comparable number of ships? Oh well.

That said, i do still think this would be a good idea for doomsday devices.

LoS has interested me since the first day I started playing.

Performing the calculation is not the problem. The problem is when the server has to ask every single object on grid whether its position in space fits this straight line equation every time anyone fires their guns. I don't even want to think about how this would be implemented for missiles.

I am pretty sure Sins never performed LoS calculations. Based on my experience with it, I think it applies the damage to the other ship then draws a projectile that travels from ship A to ship B. I've seen damage applied even when the projectile clearly missed its target, and I've seen missiles pass through capital ships to hit a frigate on the other side.

Actually, that's how EVE does it too isn't it. It applies damage then draws some effect to make it look like something was fired.

If you want an example of an RTS that does a full simulation including LoS calculations, Supreme Commander (the first one) does this. That was one of the big selling points back in 2007 because it meant that formations, maneuvers, and terrain played a role in the outcome of fights.

EDIT: You may also remember how slow Supreme Commander gets, even for an RTS.

The problems with fleet formations have been discussed in other threads. It's not a good idea and is open to significant abuse. Having arbitrary fleet formations that provide static buffs and debuffs is bad gameplay, and only doing LoS calculations on ships in formation is going to be exploited. Not to mention that the biggest gains of doing those calculations, the ability to hide behind asteroids or structures, are lost. So what would the point be?

Also, this isn't an exponential increase, it's a polynomial time problem. If it were exponential, it would be something more like 2^N which grows far faster than N^2. In fact, exponential (non polynomial time) algorithms are so hilariously inefficient that we have a special name for them: intractable.

It's intractable because the growth rate is so rapid that even adding more hardware isn't going to help you. Exponential algorithms are not good solutions because they slow down so quickly as the number of objects it needs to be performed on increases.

Fortunately, our problem is a nice polynomial time algorithm so we don't have to worry about that, but even adding new hardware isn't a great solution. Yes, adding new hardware will make it possible, but it could be better spent on just allowing more people to participate in the fight, or by staving off TiDi so fights are more fun. So if we had some epic supercomputer from the future that gave the game server a massive boost in computing power, yes it would be possible. But that added raw computing power could be better used elsewhere.

I don't think you are going to make this problem much more efficient. Most of the papers on this stuff were written decades ago and few advancements have been made (this is the case in disciplines where math dictates what is possible). If you do figure it out, please publish immediately. The problem is that all the more efficient shortcuts make assumptions we can't make in EVE PVP such as a static scene or some entities never being able to interact with each other.

The thing is, EVE makes lots of compromises to be as big as it is. There's always someone suggesting being able to fly your ship with a joystick, or having more realistic collision detection, or line of sight calculations for weapons. The problem is that these ideas drastically increase the complexity of the physics simulation and that just can't scale to the number of players involved here. We'd have to do what FPS servers do and limit you to 32 or 64 players in a match, which is not something I am willing to accept no matter what because having thousands of players involved is one of the things that makes EVE unique.

Yeah this thread has kind of turned into a computer science lecture.