KSP 2 Multiplayer Discussion Thread

[Vl3d]

6 hours ago, Pthigrivi said:

Because there’s no common metric of time it seems like you could only really race in real time? As in you could agree “who can get to the moon in 1 hour”, or “who can get to Debdeb by this coming June” but not “Who can reach a new star system by year 20 on the in-game calendar”. 

You're still "in love" with thinking about time as a reference metric. My advice is to just think about "who can get there first". In game universal time is irrelevant - only the planted flags matter (thus relative mission time).

Edited November 28, 2022 by Vl3d

[Pthigrivi]

2 hours ago, Vl3d said:

You're still "in love" with thinking about time as a reference metric. My advice is to just think about "who can get there first". In game universal time is irrelevant - only the planted flags matter (thus relative mission time).

Which is fine if its 1v1v1v1 in single launch missions to the Mun, but if you have several players all with multiple vessels and stations all existing on different internal clocks and timelines (and hopping on and off those timelines) how do you compare who won? You don’t share a game clock even with your teammates, so you can really only measure based on IRL time. This would go for any MP scheme without a shared timeline. 

Edited November 28, 2022 by Pthigrivi

[Vl3d]

1 hour ago, Pthigrivi said:

if you have several players all with multiple vessels and stations all existing on different internal clocks and timelines (and hopping on and off those timelines) how do you compare who won?

In my mind the are no different timelines. Internal clocks (actually the solar system configuration) are (is) synced if you are in multiplayer mode on the same celestial body. If you are in deep space you're single player (maybe with map view showing you the other agency craft on rails) or co-op on the same ship.

Agency A challenges agency B who gets to Duna first starting on Kerbin. Mission starts at T0 for both agencies. Mission can have multiple launches. Everyone plays whenever they want or simultaneously, doesn't matter. First one to get there wins.

What's so complicated about it?

PS 1: Nate said in the PC Gamer interview that warping under acceleration is allowed because they have a new non-eliptical trajectories solver. Meaning that "on-rails" is no longer just "on-a-cyclic-orbit". So there are new ways to estimate trajectories. But actually it's simpler than that - you can just show the opponents position in map view without the orbit when he is performing a maneuver.

PS 2: @Pthigrividon't confuse yourself trying to solve time travel.. multiplayer is simple - you're either playing together synced or you're playing alone with "images" of other players "on-rails". There is no time travel, there's only the local PC game state and the server game state. You switch between the two when playing. But remember: warp drives are not allowed. So the Luna/Dark implementation where players just pop in and out of existence in front of you and break the simulation by teleporting is not acceptable. Also for me having players "work together" on deciding when to sync or warp is not acceptable. The system should do it automatically for them.

Edited November 28, 2022 by Vl3d

[Pthigrivi]

34 minutes ago, Vl3d said:

What's so complicated about it?

There's nothing complicated about it. It just doesn't allow teams to measure progress in an apples to apples way. Agency A has 4 players and different amounts of time has elapsed in-game for each of them. The only clock all players from both teams share is the IRL clock. 

When you say I'm "in love" with time as a metric thats because thats how physics operates. If when a player time-warps all of their vessels, all planets and moons, and all other players vessels don't move naturally along their orbits then you're not working within a consistent physical system. That makes it very difficult to do any of the things that a game like KSP requires--rendezvousing, docking, planning for transfers and organizing supply routes. T_v's cyclic/local bubble idea has its pros and cons but it does at least function. They accomplish this by breaking the game into a series of different realities. When a player enters a new SOI the objects in it become visible and their positions are transposed onto that players timeline. When they time-warp all of their vessels, all of the planets, and all of the vessels in that SOI move relative to their progress through time. If some of those objects are owned by other players and those other players time-warp it doesn't affect what you see. There is no fixed SOI reality for each world, each world has a separate reality for each player. There's another bubble-reality thats created when you enter the physics range of another players vessel as you are teleported to its position as viewed by its owner. Now... is this slightly confusing? Maybe, but so would be managing a 4d reality shared by multiple players. It does at least allow players to time-warp at will when they need to without anything wacky going on. When Player 1 enters a rendezvous maneuver with Player 2's station both objects move naturally along their orbits even if both players time-warp. They don't have to sit and wait for 30 minutes while the approach happens in real time, or worry that the target might suddenly vanish and reappear somewhere else because its owner timewarped. 

Edited November 28, 2022 by Pthigrivi

[Vl3d]

10 minutes ago, Pthigrivi said:

When they time-warp all of their vessels, all of the planets, and all of the vessels in that SOI move relative to their progress through time. If some of those objects are owned by other players and those other players time-warp it doesn't affect what you see.

And what if they leave by warping and quickly come back? Do you see doubles?

Who said anything about waiting 30 minutes?

10 minutes ago, Pthigrivi said:

T_v's cyclic/local bubble idea has its pros and cons but it does at least function.

Can you summarize the idea please?

Edited November 28, 2022 by Vl3d

[Pthigrivi]

Just now, Vl3d said:

And what if they leave and quickly come back? Do you see doubles?

No, it presumably it just resets. 
 

1 minute ago, Vl3d said:

Who said anything about waiting 30 minutes?

It takes about 30 minutes to do a full orbit around LKO, so I used that as a benchmark. Its much worse if your orbit is higher up or you're around another body. If the objects in each body's SOI are locked to that body's universal real-time then when I time warp other objects in that system don't progress naturally around it. They appear nearly fixed in place because time is progressing much faster for me than for objects fixed to the local space. That means I can't time-warp while rendezvousing because the target isn't moving naturally through space at the same pace as my vessel. That means I have to sit and wait for all rendevous to happen in real time, which could be 15 minutes or several hours. 

[SkyFall2489]

wait wha?

step back a little. These messes each seem more elaborate the the previous one....

The old subspace, vote controlled, and encounter systems seem way better.

 

What is our goal, anyways?

"Allow players to compete in teams in small scale multiplayer, racing to get to objectives first while also being able to see or interact with eachother's missions, all keeping the feel of the original KSP"?

[Vl3d]

3 minutes ago, Pthigrivi said:

That means I can't time-warp while rendezvousing because the target isn't moving naturally through space at the same pace as my vessel.

You can set as target and warp to rendezvous as usual. No waiting. When you warp you exit Kerbin multiplayer, do maneuvers in single player map mode and enter the target bubble when you arrive. Both of you then sync to share a solar system configuration. Let's assume the sync is hidden.

[magnemoe]

On 11/27/2022 at 12:04 AM, t_v said:

I agree that this system results in a lot of waiting in between time warps, but your explanation for how competitive play exacerbates this doesn't make sense for me. Even within one agency, players will be able to fly multiple craft, so this problem exists whether play is competitive or not. The reason I was asking was because I was unsure of whether some facet of this system was particularly bad for competitive play. Since it seems that the problem is one that has already been discussed and accepted or rejected, I'll just keep to my own opinion of it. 

Also, to explain how other systems have problems with competition, the Local Bubble/leapfrog system can have a player go into the future and lock down the entire game, forcing people to abandon everything and warp into oblivion, and systems with automatic synchronization have the same problem, except you can see the griefing in real time. All of these systems allow people to easily abuse time warp to hinder others, if someone is determined enough. That is why we need moderation tools, to deal with these problems, because ultimately they won't be fixed by the systems themselves. 

This is an existing issue in KSP 1 single player if you do a lot of stuff, interplanetary missions takes forever in real time as I keep doing stuff in the Kerbin system as in doing contracts, training kerbals and expanding my bases. 
With multiplayer this will be worse as you will have to wait for everybody else doing stuff. 
Time / warp bubbles solves much of this by decoupling time. Yes you need to catch up at the end but you don't have to wait for everybody launching and assembling ships. 
I also assume KSP 2 will be kind of cooperative even in competitive play so you agree on goals so if the other want to have an race to Jool, some don't spend 10 hour driving rovers on the Mun. 

[Vl3d]

We're all just arguing about how big the multiplayer bubbles are. I support a hybrid system - big bubbles for celestial bodies and smaller ones for stations / player encounters. Traveling between them would be single player with map view also showing some other craft "on rails". Why do we have to make it so complicated?

Edited November 28, 2022 by Vl3d

[SkyFall2489]

Just now, Vl3d said:

We're all just arguing about how big is the multiplayer bubble. I support a hybrid system - big bubbles for celestial bodies and smaller ones for stations / player encounters. Traveling between them would be single player with map view also showing some other craft "on rails".

in Encounters, the bubble is single SOI.

But otherwise, one server at one time is one bubble, right?

if subspace, there's a bubble per time which covers all locations in space.

if vote, lock, or master controlled warp, then there's one bubble.

 

[Vl3d]

6 minutes ago, SkyFall2489 said:

if subspace, there's a bubble per time which covers all locations in space

You think it's a good technical solution to have N players generating potentially N game states? Think about the amount of information the persistent world server would have to save in the database.

8 minutes ago, SkyFall2489 said:

vote, lock, or master controlled warp, then there's one bubble

You can't put the responsibility of consensus on the players.

[Pthigrivi]

15 minutes ago, Vl3d said:

You can set as target and warp to rendezvous as usual. No waiting. When you warp you exit Kerbin multiplayer, do maneuvers in single player map mode and enter the target bubble when you arrive. Both of you then sync to share a solar system configuration. Let's assume the sync is hidden.

But not if the vessel you’re targeting is owned by someone else. When you you exit multiplayer that object stays in multiplayer and doesn’t time warp with you, which means instead of gently merging your orbits it stays nearly fixed in place and you zoom past it. 

[Vl3d]

7 minutes ago, Pthigrivi said:

When you you exit multiplayer that object stays in multiplayer and doesn’t time warp with you, which means instead of gently merging your orbits it stays nearly fixed in place and you zoom past it. 

No, the position you see warps with you because it's "on rails" (calculated on your PC using parameters from the server). The only issue is that your ss configuration in map viewer is different from his. So probably if you enter his bubble you sync to his state (the server state).

1 hour ago, Vl3d said:

PS 2: @Pthigrividon't confuse yourself trying to solve time travel.. multiplayer is simple - you're either playing together synced or you're playing alone with "images" of other players "on-rails". There is no time travel, there's only the local PC game state and the server game state. You switch between the two when playing. But remember: warp drives are not allowed. So the Luna/Dark implementation where players just pop in and out of existence in front of you and break the simulation by teleporting is not acceptable. Also for me having players "work together" on deciding when to sync or warp is not acceptable. The system should do it automatically for them.

 

Edited November 28, 2022 by Vl3d

[Pthigrivi]

1 minute ago, Vl3d said:

No, the position you see warps with you because it's "on rails" (calculated on your PC using parameters from the server). The only issue is that your ss configuration in map viewer is different from his. So probably if you enter his bubble you sync to his state.

"On rails" just means moving through space naturally from your perspective. What you see in map-mode is reality, at least for you. If each player has their own perspective of a given SOI's events then you basically have what t_v described with many nested realities--one for each player in each SOI, not a persistent SOI multiplayer world. You don't enter or exit multiplayer, you import the positional data for other players vessels at the time you entered the SOI and then carry them along within your own bubble reality as you time-warp. When you synch to another players station your vessel is plucked out of your reality and dropped wherever that other player sees their station. If you stay in that SOI for any period of time and other players variably warp, move vessels to other orbits, dock and undock with their own and other players' vessels your perspective of that SOI is going to progressively diverge from what other players see making cooperation and coordination more and more difficult the longer you play and the more players are active. Is my vessel at Pe or Ap? Is my teammates' vessel in a 100km circular orbit or a 1000km elliptical orbit? Nobody knows because everyone is seeing different things. Importantly you also don't share a clock. As mentioned this makes it so there's no way to tell who won a space race in game-time, but it also creates some weird issues with supply lines. I could say I'll deliver 1000 tons uranium to a friends base, make a tiny vessel that carries 2 tons of uranium, and then timewarp a few decades. The other player would see 1000 tons all delivered instantly. And if time and quantities are meaningless in that way and exploits like that are so easy it kind of breaks a lot of the rules you'd expect to follow when competing. 

[Vl3d]

1 hour ago, Pthigrivi said:

As mentioned this makes it so there's no way to tell who won a space race in game-time, but it also creates some weird issues with supply lines. I could say I'll deliver 1000 tons uranium to a friends base, make a tiny vessel that carries 2 tons of uranium, and then timewarp a few decades. The other player would see 1000 tons all delivered instantly. And if time and quantities are meaningless in that way and exploits like that are so easy it kind of breaks a lot of the rules you'd expect to follow when competing. 

1. First who gets there wins.

2. Even talking about having a space race means DRM-enabled server hosted multiplayer.

3. The resources example is not useful to prove any point related to multiplayer. In my opinion mining should be directly proportional to time spent in game / effort. Mining continuously while time warping breaks competitive multiplayer.

[razark]

11 hours ago, Vl3d said:

1. First who gets there wins.

The question is how do you define "first"?  By the clock in the game, or the clock on the wall?

If A takes a week of real time to get to the Mun in 3 hours of in-game time, and B takes 4 days to get to the Mun in 5 years of game time, which one was "first"?

 

If you're going by the clock on the wall, then it's not a space race, it's a timewarp race.

Edited November 29, 2022 by razark

[Vl3d]

6 hours ago, razark said:

The question is how do you define "first"?  By the clock in the game, or the clock on the wall?

If A takes a week of real time to get to the Mun in 3 hours of in-game time, and B takes 4 days to get to the Mun in 5 years of game time, which one was "first"?

If you're going by the clock on the wall, then it's not a space race, it's a timewarp race.

Two types of races then.

One is a PvP real time flight race.. if you plant a flag on the Mun before the competitor, you won. Warp as much as you can, winner is decided by "first there wins".

The second is a "play when you can" space program race involving the whole process of design and tech progression. Winner is decided by in game time.

I would prefer the second - you play as slow as you want in real time, but what matters is the mission time. The winner is decided thus: first one there AND in-game-mission-elapsed-time lower than the competitor.

Edited November 29, 2022 by Vl3d

[Pthigrivi]

9 hours ago, Vl3d said:

The second is a "play when you can" space program race involving the whole process of design and tech progression. Winner is decided by in game time.

This is the trick though because you don't share a clock even with your own teammates. If you have two teams with 4 players each and on one team those players have spent 3 days, 5 days, 4 days and 4 days and on the other team the players have each spent 1 day, 2 days, 2 days, and 8 days which team won? 

[Streetwind]

I think when he said "in game time" he literally meant the date/time in the game.

If the challenge is to land on the Mun as quickly as possible, then out of eight players, the winner is the one who manages it on the earliest ingame timestamp. Doesn't matter how much IRL time anyone invested.

That way, someone who can play only thirty minutes per evening can meaningfully compete with, and even win against, someone who spends all day in front of their computer. It all depends on how efficiently you go about the task.

[Pthigrivi]

2 minutes ago, Streetwind said:

I think when he said "in game time" he literally meant the date/time in the game.

If the challenge is to land on the Mun as quickly as possible, then out of eight players, the winner is the one who manages it on the earliest ingame timestamp. Doesn't matter how much IRL time anyone invested.

That way, someone who can play only thirty minutes per evening can meaningfully compete with, and even win against, someone who spends all day in front of their computer. It all depends on how efficiently you go about the task.

No I understand. The trouble is in his version of MP players don't exist on the same timeline. That means they are all going to have different in-game timestamps even when cooperating. That's fine if its 1v1v1, but not if its 4v4, because each player within the team has different in-game timestamps making it difficult to tell which team won. Depending on how players on a team divvy up responsibilities those number could be all over the place leaving no easy metric for comparison. 

[darthgently]

29 minutes ago, Pthigrivi said:

No I understand. The trouble is in his version of MP players don't exist on the same timeline. That means they are all going to have different in-game timestamps even when cooperating. That's fine if its 1v1v1, but not if its 4v4, because each player within the team has different in-game timestamps making it difficult to tell which team won. Depending on how players on a team divvy up responsibilities those number could be all over the place leaving no easy metric for comparison. 

The solution involves trivial arithmetic 

[Pthigrivi]

25 minutes ago, darthgently said:

The solution involves trivial arithmetic 

Okay, is it the average? Or is it the team with the player with the shortest time? or is it the team with the shortest total time? Remember these players aren't on the same timeline. When they synch players within a team are in a sense time-traveling to the point of the synch, because time does not elapse for them and their world to make up the difference. On a 4v4 race to Duna you might have 3 players who spend a couple of in-game days delivering modules to orbit while the 4th flies the mothership all the way from kerbin to the surface. The team they're competing against sends multiple, simultaneous, smaller deliveries that combine in Duna orbit before landing. If they were living on the same timeline team 2 might win because it took them less time to get to the surface on the shared clock, but based the average or collective time spent in an incoherent MP scheme team 1 would win by a factor of nearly 4 not because they actually got there faster, but due entirely to the way they divvied up the job. 

Edited November 30, 2022 by Pthigrivi

[darthgently]

11 minutes ago, Pthigrivi said:

Okay, is it the average? Or is it the team with the player with the shortest time? or is it the team with the shortest total time? Remember these players aren't on the same timeline. When they synch players within a team are in a sense time-traveling to the point of the synch, because time does not elapse for them and their world to make up the difference. On a 4v4 race to Duna you might have 3 players who spend a couple of in-game days delivering modules to orbit while the 4th flies the mothership all the way from kerbin to the surface. The team they're competing against sends multiple, simultaneous, smaller deliveries that combine in Duna orbit before landing. If they were living on the same timeline team 2 might win because it took them less time to get to the surface, but based the average or collective time spent in an incoherent MP scheme team 1 would win by a factor of nearly 4. 

You are making it too complicated.  Duration of individual race is simply end time minus start time, regardless of synch. 

Start time is first launch or whatever is agreed upon. 

End time is when conditions are met, as with a contract. 

How a team score is computed could follow any number of existing tried and true models.  The simplest being the sum of individual team member scores, assuming same number of team members.

Multiple ribbons could also be awarded, fastest touchdown, MVP, etc 

End conditions could even leverage the existing Contract system to generate the appropriate timestamps in log, game contract archive

Edited November 30, 2022 by darthgently

[Pthigrivi]

1 hour ago, darthgently said:

Start time is first launch or whatever is agreed upon. 

End time is when conditions are met, as with a contract. 

This is true if players exist on the same timeline. Thats not what's happening in the bubble-universe MP models like the one Vl3d is describing. In those models each player has their own timeline which splices into others whenever they interact with a new SOI or another player's vessel. Player A is working in and around Kerbin and Player B flies out to Jool and back. When Player B arrives back at Kerbin they synch in real-time. That means a few in-game days or hours have elapsed for Player A and 6+ in-game years have elapsed for Player B. Neither "Start time" nor "Conditions met" are fixed or comparable points in time for all players. In my example above when Team 1's mothership arrives 300 days have elapsed for one player, but zero time has elapsed for the other 3 because each of those team members exists in their own distinct timelines. Because you don't have a single, coherent reality you end up not measuring victory the way a space-race works--an actual race against a shared clock--but based entirely on how a team happens to have allocated jobs. Teams in which 3 players do nothing but design work in the VAB and one player does all the flying will always beat teams that fly and combine multiple vessels because 0 time elapses for 3/4 of the team. 

Edited November 30, 2022 by Pthigrivi