For the traditional pvp setup with a host, picture that host as the hub of a many-spoked wheel, where each spoke is the connection to a client. Clients send data to the host that says (I want to move here) or (I'm shooting now) and the host makes it happen and then sends to everyone else (this client moved there and he's shooting now). There are more predictive ways to do this where you send bits of information to the other clients as well, but the key element is that the host dictates to every other system what happens in the game. If you lag while sending "I'm moving" to the host, the host won't get that data so it will assume that you're not moving, and tell the other clients as much, so everyone else sees you standing still and not shooting even though you think you're running and gunning. If they shoot you, you die on their systems and in the game, and as soon as you catch up you will teleport back to where you started lagging and be dead. This system sucks if you have a shit connection, but it is not exploitable via network interference (lag-switching). Also, lagging is generally immediately obvious to the lagger because they stop getting info from the host, which means they see everyone else stop moving at once. The major advantage to this architecture is that if you shoot someone, they will die, even if they are lagging. The only thing that determines your effectiveness in combat is your own connection to the host. Typically, a client-host system also has a hidden "host reliability" tracking setup so hosts who have poor connections or who frequently drop out of games or quit them are weeded out and eventually only reliable players are allowed to be hosts.

You're still mixing assumptions about networking model in with trust. There is no reason that a host-based gunplay model couldn't stall kill judgement based on confirmation from the client being killed, for instance (although I'd be surprised if a hard version of that is the issue here).

You're right, but the design of traditional console FPS host-client setups (as accepted for esports) is that the host's judgement is absolute and everyone else gets updates from the host. I know this first hand as I worked on many of the call of duty games, specifically online multiplayer. Dedicated servers with absolute server judgement is better because you know you can trust the server, whereas you only have a good idea whether or not you can trust a host, based purely on match history and connection quality and speed.

From my observations (I'm a 13 year pro tester, these are not lay observations), it appears that System A would actually send "I have created bullets between points A and B" where A is the player's gun and B is where system A thinks those bullets hit something. However, if System B doesn't agree that player B is standing where System A thinks he was standing, it simply won't return any damage as registered.

AFAIK with precision weapons, it usually isn't based purely on shot vectors, for precisely the reasons you mention. Player's views are so divergent that you'd basically never be able to hit anything. Bungie made a good post about this back in the Halo 3 days: with Halo 2's hitscan weapons the client would send requests of the form "I shot this player in this part of their body", and with Halo 3's projectiles it was in the form of "I shot at this player and I was aiming this far ahead of them."

Since bandwidth is such a critical issue, I'd actually expect that most games only ever reconstruct shot vectors on the host. So the host might say "no, there seems to be a wall between you and the enemy's elbow based on where I think you are and where I think they are."

The trick is that we know, in Destiny, that if a system believes there to be a person in a particular space, shots fired on that system will stop moving at the space where they believe that person is. Imagine a scenario where there are three characters, A, B, and C. System B is lagging badly. On Systems A and C, it appears that all three characters are standing in a straight line with character B in the middle. On System B, Character B is actually back at his spawn, waiting for his connection to improve, and is no where near characters A and B. If character A fires a golden gun shot at where he sees character B, that shot will deal no damage to character B (because character B isn't there on System B) but also will not continue on and hit character C instead (because character B is there to block it on Systems A and C).

This fact is a major source of frustration when playing Destiny: The shot acts as if both the position of character B are true and false at the same time, and in both conditions the outcome is the least desirable outcome from the standpoint of the person who pulled the trigger. It is further compounded by the fact that supers and heavy ammos are so rare and valuable that wasting even one shot can be the deciding factor in a game.

I believe that a true host-client structure where only reliable hosts were chosen would virtually eliminate the odds of scenarios such as this from occurring.

Adding on: Destiny's architecture is such that in situations when one player lags, the game resolves this situation such that only the lagger knows where he really is and all other players have false information. In more traditional models, the reverse is true. The lagger is not sure where he really is, but everyone else sees the correct location as dictated by the host. It's a utilitarian philosophical point, but one I feel is worthy of mention.

There are plenty of things that could potentially still throw this off with lag, however. For instance, one thing you mention is that Destiny allows prediction to go very far. One possibility would be that, when a player stops sending packets about where they are, the host doesn't repeat these to other clients, and potentially falls behind in communicating updates on that player's location at all. In that case, small differences in that person's position and orientation at the start of the lag could very quickly diverge into huge differences across different player's consoles. That could effectively make the player invulnerable since the host would find other player's shots somewhat nonsensical.
(Obviously this is totally hypothetical and might not be much of a problem, although all this stuff would be easier to check if we had replays.)

Now, there is one player who is the "physics host" of the game, and that player's system handles the clock countdown timer, ammo spawns, and - presumably - non player object motions such as ammo drops (though it is entirely possible that ammo drops are 100% contained within their own players' systems).

Those are mostly things that don't require high responsiveness but require stability, things that according to the networking presentation would be handled by the dedicated server (or, as you mention, some of it could be handled locally).

I agree on all of these points. As you mention in your hypothetical scenario, while the lag issues used to be a major problem from the game's inception until some months ago, they appear to have been largely resolved. I'm not sure what changes were made in the netcode to solve these issues, but I imagine it was purely due to more connection-based matchmaking priority.

Shoot, they could even put some of that microtransaction money toward 'live' events happening in these locations without having to mess with the core D2 levels.

Yeah, I thought that was pretty funny myself. ;)

Shoot, they could even put some of that microtransaction money toward 'live' events happening in these locations without having to mess with the core D2 levels.

There is a cost to keeping old stuff, namely having the systems in place to support it. If you keep old stuff, you have to keep a lot of other things that are best left behind.

Surely importing the level geometry (and enemy spawns) won't tie them down. They don't need to keep all the public events and missions and all of that. Just allow us to go back, maybe give us a reason as part of an exotic quest. Shoot, they could even put some of that microtransaction money toward 'live' events happening in these locations without having to mess with the core D2 levels.

I expect Bungie to build all new maps for most of the story missions & patrol in D2. I hope they also import the old maps from D1, expand some areas (similar to what they did w/ the plague lands, but not have it be a separate patrol area) & add quests and side story that involve revisiting old areas which have new content in them. A drawback I see with the importing of old maps... If they do import the old maps then the community may rage that they are selling us old content again. Bungie will need to make sure the new maps & content are clearly worthy of a game in and of themselves & present the ability to visit old areas in the right light... Seriously, you KNOW people will rage about it even if it's just "as a bonus".

edit: formatting believe it or not.

Make old content a PS4 exclusive. Then Xbox and PC players will rage about not having it

For the traditional pvp setup with a host, picture that host as the hub of a many-spoked wheel, where each spoke is the connection to a client. Clients send data to the host that says (I want to move here) or (I'm shooting now) and the host makes it happen and then sends to everyone else (this client moved there and he's shooting now). There are more predictive ways to do this where you send bits of information to the other clients as well, but the key element is that the host dictates to every other system what happens in the game. If you lag while sending "I'm moving" to the host, the host won't get that data so it will assume that you're not moving, and tell the other clients as much, so everyone else sees you standing still and not shooting even though you think you're running and gunning. If they shoot you, you die on their systems and in the game, and as soon as you catch up you will teleport back to where you started lagging and be dead. This system sucks if you have a shit connection, but it is not exploitable via network interference (lag-switching). Also, lagging is generally immediately obvious to the lagger because they stop getting info from the host, which means they see everyone else stop moving at once. The major advantage to this architecture is that if you shoot someone, they will die, even if they are lagging. The only thing that determines your effectiveness in combat is your own connection to the host. Typically, a client-host system also has a hidden "host reliability" tracking setup so hosts who have poor connections or who frequently drop out of games or quit them are weeded out and eventually only reliable players are allowed to be hosts.

You're still mixing assumptions about networking model in with trust. There is no reason that a host-based gunplay model couldn't stall kill judgement based on confirmation from the client being killed, for instance (although I'd be surprised if a hard version of that is the issue here).

From my observations (I'm a 13 year pro tester, these are not lay observations), it appears that System A would actually send "I have created bullets between points A and B" where A is the player's gun and B is where system A thinks those bullets hit something. However, if System B doesn't agree that player B is standing where System A thinks he was standing, it simply won't return any damage as registered.

AFAIK with precision weapons, it usually isn't based purely on shot vectors, for precisely the reasons you mention. Player's views are so divergent that you'd basically never be able to hit anything. Bungie made a good post about this back in the Halo 3 days: with Halo 2's hitscan weapons the client would send requests of the form "I shot this player in this part of their body", and with Halo 3's projectiles it was in the form of "I shot at this player and I was aiming this far ahead of them."

Since bandwidth is such a critical issue, I'd actually expect that most games only ever reconstruct shot vectors on the host. So the host might say "no, there seems to be a wall between you and the enemy's elbow based on where I think you are and where I think they are."

There are plenty of things that could potentially still throw this off with lag, however. For instance, one thing you mention is that Destiny allows prediction to go very far. One possibility would be that, when a player stops sending packets about where they are, the host doesn't repeat these to other clients, and potentially falls behind in communicating updates on that player's location at all. In that case, small differences in that person's position and orientation at the start of the lag could very quickly diverge into huge differences across different player's consoles. That could effectively make the player invulnerable since the host would find other player's shots somewhat nonsensical.
(Obviously this is totally hypothetical and might not be much of a problem, although all this stuff would be easier to check if we had replays.)

Now, there is one player who is the "physics host" of the game, and that player's system handles the clock countdown timer, ammo spawns, and - presumably - non player object motions such as ammo drops (though it is entirely possible that ammo drops are 100% contained within their own players' systems).

Those are mostly things that don't require high responsiveness but require stability, things that according to the networking presentation would be handled by the dedicated server (or, as you mention, some of it could be handled locally).

Their ability to manage old content is questionable. Cruel tells me that if you play the game from the beginning now, with Questification, that it is not only a flow killing series of back and forth to the tower, but the game also has you starting Taken King missions before you even do the Crota raid. Which makes no sense because you haven't killed Crota yet!

I made a similar post in a reply to a thread on reddit somewhere a few days ago.

The Quest system sucks. It's worse than it ever was, if you're playing from the beginning. It makes LESS sense. It basically just opens up everything from every expansion at once and floods your quest log with new things, with absolutely no indication of how they fit together.

Nor should it. If I'm year ten I am playing year one stuff that is just lazy. New new new please.

How the hell is it lazy? I'm not saying they shouldn't add content. They should. They should spend just as much time and effort as they have to date, if not more, adding content to a stable platform without taking it away. By year 10 the game should be 10x bigger than it was in Year 1, and that shouldn't require giving up one square meter of what has been built.

I'm with you. I think Cody's reaction is a perfect example of what I'm worried about Bungie having to manage if they include old content. People seeing it as a negative instead of a positive. "Bungie's too lazy to make new stuff," even if they did make as much new stuff in D2 as they had in D1.

There is a cost to keeping old stuff, namely having the systems in place to support it. If you keep old stuff, you have to keep a lot of other things that are best left behind.

Their ability to manage old content is questionable. Cruel tells me that if you play the game from the beginning now, with Questification, that it is not only a flow killing series of back and forth to the tower, but the game also has you starting Taken King missions before you even do the Crota raid. Which makes no sense because you haven't killed Crota yet!

How the hell is it lazy? I'm not saying they shouldn't add content. They should. They should spend just as much time and effort as they have to date, if not more, adding content to a stable platform without taking it away. By year 10 the game should be 10x bigger than it was in Year 1, and that shouldn't require giving up one square meter of what has been built.

It is only 10x bigger if you are just starting out. When I have conquered and have little interest in going back to year one stuff because it's been mastered and I'm sick of it, it doesn't matter if it's kept. Even worse if it is slightly tweaked and rolled into the activities you are expected to do.

I've already done it. I want to do something new.

I'm with you. I think Cody's reaction is a perfect example of what I'm worried about Bungie having to manage if they include old content. People seeing it as a negative instead of a positive. "Bungie's too lazy to make new stuff," even if they did make as much new stuff in D2 as they had in D1.

My reaction is simply, "Why would they include it?"

Presumably, Destiny 1 won't go away. If I want to play Destiny 1 stuff, I'll play Destiny 1. I don't want new things to be held back or limited in any way by old stuff. If what this article hints at is true, it seems like that's exactly what they're avoiding. I'd be more than a little disappointed if Destiny in 10 years is basically exactly the same as Destiny now, just with more stuff.

Nor should it. If I'm year ten I am playing year one stuff that is just lazy. New new new please.

How the hell is it lazy? I'm not saying they shouldn't add content. They should. They should spend just as much time and effort as they have to date, if not more, adding content to a stable platform without taking it away. By year 10 the game should be 10x bigger than it was in Year 1, and that shouldn't require giving up one square meter of what has been built.

I'm with you. I think Cody's reaction is a perfect example of what I'm worried about Bungie having to manage if they include old content. People seeing it as a negative instead of a positive. "Bungie's too lazy to make new stuff," even if they did make as much new stuff in D2 as they had in D1.

By year 10 the game should be 10x bigger than it was in Year 1, and that shouldn't require giving up one square meter of what has been built.

IF it means building on a bad foundation, we should absolutely give it up. Destiny is fun as it's built, but it can be so much more than what it currently is. If we have to give up all Destiny 1 content carrying forward to make it happen, it's a price I'm more than willing to accept.

Yeah. I'm kinda in the same boat. I totally believe in what Narcogen is saying. But this is one of the times that Bungie has kinda slipped up in their production of a game. If they can say, well Destiny 1 was a trial that didn't go so well and from here on out create an awesome game. I'm okay with that. It would still be nice if they kept the same content but put it on a better platform/foundation.

By year 10 the game should be 10x bigger than it was in Year 1, and that shouldn't require giving up one square meter of what has been built.

IF it means building on a bad foundation, we should absolutely give it up. Destiny is fun as it's built, but it can be so much more than what it currently is. If we have to give up all Destiny 1 content carrying forward to make it happen, it's a price I'm more than willing to accept.

Wasn't the whole point of making Destiny the way it is was to be able to continue to add onto it, like with WoW, so that at the end you'd have the equivalent of all the content of Halo 1-3 except under a single umbrella, playable start to finish, with significant content dropping once or twice every year over a decade instead of three times with gaps of three years?

Was it? Because I certainly don't recall ever hearing that--just a whole lot of assumptions being thrown around and taken as gospel by the community at large. They talked about a ten year journey with your Guardian being carried forward throughout the journey. I don't recall them ever saying all the content from day one would carry forward forever.

I'm honestly not sure how "a ten year journey with your Guardian carried forward" is consistent with "throw out your old gear and your old characters and by the way all the planets are new too."

In that case, why bother with incremental updates at all? Why not do monolithic releases like before?

Nor should it. If I'm year ten I am playing year one stuff that is just lazy. New new new please.

How the hell is it lazy? I'm not saying they shouldn't add content. They should. They should spend just as much time and effort as they have to date, if not more, adding content to a stable platform without taking it away. By year 10 the game should be 10x bigger than it was in Year 1, and that shouldn't require giving up one square meter of what has been built.

I would kill to have a seamless way to play all the Halo games, in a single engine, with coop that works half as well as Destiny's does. And if in year ten I say hey, wouldn't it be cool to run a Plaguelands patrol for good time's sake, do I want to hear that everybody has to go hunt for a disc, or that they've deleted Destiny 1 off their hard drives to make room? Or that server support for that area is gone?

As is true most of the time, I hope like hell nobody at Bungie thinks about this game the way you do. The way you complain about it, so far I've been lucky, but who knows...