The year is 2020 and Elite Dangerous is still P2P. Let's discuss PvE and PvP servers.

[Robert Maynard]

I'm sorry if this sounds cold and heartless but this kind of annoys me. Every single MMO on console requires Xbox live or PS Plus(?). A game being held back because some console players don't wanna pay $5 a month for online service is beyond ridiculous.

Whether one agrees or disagrees with the premise behind the sale of the game to every player is up to each player.

That this MMO does not require all players to play among other players does not mean that it needs to change to fit the "normal" definition of an MMO.

How is the game being held back, exactly, by players playing the game how they want to?

 

[Ian Doncaster]

but frontier just confronted the playerbase (the backers) with the decision to either pay for it with a subscription fee or go without. anyhow, they chose not to pay. don't ask.

Not quite - it was never offered as an option. I'm pretty sure they had the networking model all set out long before the Kickstarter ... sadly it didn't work as well over the entire internet as it did in their labs, though they managed to get a fair bit of it fixed up by 2.2.

The funding model bit they did change was a move from the original kickstarter "cash for credits" to "cash for cosmetics". Given their total inability to keep control of credit earnings, that was probably a good decision all round.

handling real-time player interactions with simulation of physics in a 6dof combat scenario is hard and costly proportionally to the complexity of the simulation, the volume of information per player and the number of players taking part. there are obvious practical limits to this, and it requires a dedicated network of servers to run. this is a significant cost that many multiplayer games (even f2p ones) just assume as part of the cost of operations,

To add to this, the big issue for Elite Dangerous - as compared to most other games - is scale.

If there's one player in a RES, the game simulation needs to handle the player, several NPCs, hundreds of asteroids and munitions and fragments. Adding an extra player or two to that isn't a big extra deal, though - more network bandwidth, but that's not a big blocker.

But because the Elite Dangerous game world is so large, the average number of players per instance is 1 (median, mode, and mean to within rounding error). So they basically need one full instance simulation per player. That's a lot of hardware - fine if it's distributed throughout all our houses, very expensive if they're running tens of thousands of CPUs constantly to keep up with players.

The games with a client-server model are generally designed to have the average number of players per instance be high ... which means a very small game world (by ED standards, anyway), which means no exploration as a profession.

It's basically not worth them switching to a client-server model for the tiny minority of cases where it would be better, when it would be more expensive ... and less good, because of things like input lag, higher bandwidth requirements, etc. ... for the ultra-common case of "a player, on their own in an instance"

 

[CaptainPugwash]

Not quite - it was never offered as an option. I'm pretty sure they had the networking model all set out long before the Kickstarter ... sadly it didn't work as well over the entire internet as it did in their labs, though they managed to get a fair bit of it fixed up by 2.2.

The funding model bit they did change was a move from the original kickstarter "cash for credits" to "cash for cosmetics". Given their total inability to keep control of credit earnings, that was probably a good decision all round.


To add to this, the big issue for Elite Dangerous - as compared to most other games - is scale.

If there's one player in a RES, the game simulation needs to handle the player, several NPCs, hundreds of asteroids and munitions and fragments. Adding an extra player or two to that isn't a big extra deal, though - more network bandwidth, but that's not a big blocker.

But because the Elite Dangerous game world is so large, the average number of players per instance is 1 (median, mode, and mean to within rounding error). So they basically need one full instance simulation per player. That's a lot of hardware - fine if it's distributed throughout all our houses, very expensive if they're running tens of thousands of CPUs constantly to keep up with players.

The games with a client-server model are generally designed to have the average number of players per instance be high ... which means a very small game world (by ED standards, anyway), which means no exploration as a profession.

It's basically not worth them switching to a client-server model for the tiny minority of cases where it would be better, when it would be more expensive ... and less good, because of things like input lag, higher bandwidth requirements, etc. ... for the ultra-common case of "a player, on their own in an instance"

Very well explained

 

[Rat Catcher]

It would be 'interesting' if, as an experiment to guage interest in such a scenario, Frontier changed the Launcher to offer such a choice with modified instancing which effectively made open 2 limitless sized PG's - then monitored and reported back to the playerbase the success or otherwise of the experiment...

I would hope that they would add some things to make a PvP server more appealing as well but what do you guys think?

Why should any 'sweetener' be added? Surely the option to play 'my way' rather than with a bunch of randoms who wish to play differently is the obvious attraction to both solutions?

I do think your proposal is reasonable, apart from the 'sweetener' option. (as it indicates that even you doubt that the population of a dedicated PvP mode would be large enough to make it 'worthwhile' for those wishing to play PvP exclusively.)

 

[Bottom Hat]

This is certainly an interesting thread, and is doing its best to avoid being a Californian hotel!

I've learned a lot about how the game works and servers etc, so thanks to the numerous posters who have given excellent reasoning and explanations.

(Not meant to sound like David Dimbleby, but managing it anyway!)

 

[VR Murgle]

I was lead to believe that E|D would spend it's life cycle with P2P connections. What being 2020 has to do with anything, I am not sure. The idea behind the P2P was to remove the need for a subscription fee. This subject was a very early topic of the Design phase conversation. Any talk of central servers would necessarily require part of this conversation to be about paying for the change. I for one wouldn't be backing an idea like this one.

It would be much easier, and cheaper, for we players to just accept that some people don't do what we do in game, and let it be. But....

You do realize that free to play MMO's were a thing before this whole peer to peer multi player non-sense was ever a thing right?

In fact of all the server based MMO's only a precious few are still subscription based.

 

[VirtueXOI]

Tbh i like my solo experience so far :x. And i'm not experienced enought in network to know how thing could be changed for a better game.

cant they make open on serv and solo the same way it is now ?

 

[ZeeWolf]

Elite does have dedicated servers for various things - the P2P stuff is for handling infinite possible "twitch" connections representing the whole of space and any combinations meeting and interacting within it.

Frontier outlined the network architecture they used in 2015

Source: https://www.youtube.com/watch?v=EvJPyjmfdz0

 

[zimms]

Frontier's point of view:

The year is 2020 and dedicated servers are still more expensive than P2P.

And that's where this story ends.

 

[Drelthar]

Where I would find a PvE server to be a bad thing is that it would be immersion breaking to me to be able to take damage from everyone except other players.

This is my gripe with any kind of PVE mode/PVP flag system or whatever other method.

I'd hate it.

 

[znôrt]

Not quite - it was never offered as an option. I'm pretty sure they had the networking model all set out long before the Kickstarter ... sadly it didn't work as well over the entire internet as it did in their labs, though they managed to get a fair bit of it fixed up by 2.2.

i was basing on hearsay on that particular since i personally didn't live the dd honeymoon. i could swear i have read several times here that there had been a vote, but that could have been fake news.

To add to this, the big issue for Elite Dangerous - as compared to most other games - is scale.

to a point, yes, but different compromises can be made. i agree frontier chose the compromises that suited their intention best: casual family play. which is ok, but the marketing of the game is highly deceiving which leads to this very same recurrent discussion.

i totally disagree that frontier couldn't have chosen other compromises, though. the infinite players in solo issue would actually be the simplest one to address. what you describe is similar to a solo experience, nothing prevents doing that client side (which is what it is doing now) and handing it over to a server the moment another player is involved. the vastness doesn't even need to be seamless (it already isn't with usses and czs and glide modes and such) so there are many ways to functionally/organically group players or even herd them into action. with a working(tm) system (true to the marketing hype) there probably would be a lot more action.

and finally, the current implementation doesn't live up to that scale hype. the vastness of the galaxy isn't important here, but the number/grouping of players, and it doesn't scale well with player numbers at all. there are problems even with fever numbers, i can play other realtime or twitch games perfectly, and can use p2p protocols just fine (albeit for what they are actually meant for XD) but have had really bizarre experiences with elite, and reports of bad experiences are plenty. wings (of 4!!!!) took ages to stabilize and multicrew isn't even there yet. and let's not open yet another fleet carrier subthread.

It's basically not worth them switching to a client-server model for the tiny minority of cases where it would be better, when it would be more expensive ... and less good, because of things like input lag, higher bandwidth requirements, etc. ... for the ultra-common case of "a player, on their own in an instance"

well, that is a very respectable opinion, but that 'minority of cases' turn out to spell the difference between a functional and a non functional pvp environment.

regarding lag/bandwith issues, we can discuss that too but i don't think there is evidence that a server based approach would necessarily perform worse, even could be the opposite. ofc all while complying with control and reliability requirements, which p2p doesn't. again, it all depends on the compromises made, and which compromises you choose is going to depend on your particular goal.

 

[ZeeWolf]

Frontier's point of view:

The year is 2020 and dedicated servers are still more expensive than P2P.

And that's where this story ends.

What do you think a "dedicated server" is and how does it fit into the architecture outlined by Frontier in the AWS presentation?

Do you mean on-prem or cloud when you talk about "dedicated"? How do you cost up the (actually expensive) AWS services spun up for the ED servers that support P2P connections vs some magial "dedicated" solution exactly?

I think that people hear "dedicated" and think "silver bullet" but it's important to understand what the actual problems are in the first place before you start suggeting fixes.

 

[MishaTX]

As to the Open PvE not being an option, I never quite figured out what FD's opposition to that was grounded in. Yes, I know, Mobius is a thing, I'm a member, but it's nothing compared to the population of Open. It's hard enough to run into fellow players in Open as it is, even though I try all the time. Only exception being the free-for-all hotspots where everybody is killing everybody for the heck of it, but that's not really my thing, no offense to those who love that sort of game.

As to dedicated servers vs. P2P, my main problem is that the latter is too damn wonky, witness the issues with instancing, sudden crashes and so on, not to mention that it means that my box is spending quite a bit of processing cycles trying to make the whole duct-taped-together budget thing hold together instead of improving my game experience which is, after all, why I bought the game ages ago in the first place. And even if it can handle it, I still can't connect with people who have issues, and I like those people too.

Of course, I certainly wouldn't want Solo to go away, mainly because it would basically shut out people on consoles who don't want to pay money to play the games they already bought, but why not embrace the healing power of "both"?

Is there any compelling reason why we can't have PvP/PvE/Solo? If so, please explain it to me.

 

[Old Duck]

Just wanted to make another thread about how garbage the instancing and networking is in this game and how amazing it could be if Frontier decided to go with dedicated servers.

All in all, I have found network and instancing to be surprisingly good in ED. We go through periods where the netcode is bugged, but that's not a design flaw, it's just somebody making a dumb mistake in the code. My only gripe with the current network model is that the number of players in an instance will greatly increase the number of NPCs in places like stations, which kills performance because all this extra telemetry that needs to be shared between players. It should be the opposite - in a busy station full of players like the old CGs, there shouldn't be hardly any NPCs - each player should REPLACE an NPC, not bring an extra dozen of his own NPCs. This would greatly improve performance and congestion (@Sarah Jane Avory).

 

[Ian Doncaster]

nothing prevents doing that client side (which is what it is doing now) and handing it over to a server the moment another player is involved

To some extent, and that in general wouldn't require many servers. It's more complexity, though - the client needs to (reasonably seamlessly!) be able to switch between full simulation and only handling input/output the moment another player enters the instance, and potentially back again once they leave again. Lag could certainly be compensated for a bit but it'd still probably be jarring suddenly going from "your input is immediately effective" to "your input takes a small but perceptible fraction of a second to cause a response", plus the chances of it being able to switch modes without a couple of seconds of synchronisation (which couldn't easily be hidden behind a loading screen) is minimal. Bugs which only show up when alone / when with other players would certainly be possible, too, though of course plenty of those happen now anyway.

It would, at the very least, be another compromise...

but the number/grouping of players, and it doesn't scale well with player numbers at all.

Agreed there. Wings and wing-sized instances are pretty stable now, at least in my experience, but took until 2.1 to work more often than not and 2.2 to work reliably. I assume they got a big shock between getting it to work on a LAN in a lab and getting it to work on a wide variety of consumer-grade hardware on the end of glorified copper string.

but that 'minority of cases' turn out to spell the difference between a functional and a non functional pvp environment.

Agreed. As I've said elsewhere many times, the Big Problem [1] with Elite Dangerous is that it tries to do too many things at once, and therefore does none of them particularly well, and "widescale exploration" and "functional pvp" are two things that I don't think can satisfactorily be fitted into the same multiplayer game.

If "multiplayer space games" was a mature genre like FPSes then there'd be combat-focused ones and exploration-focused ones and trade-focused ones and so on, and they'd all work a lot better at their one thing. In practice what there is ... Elite Dangerous trying to cover all of them at once, and a bunch of scenario-focused combat ones ... and maybe a few others in a few years.

[1] But also of course the big reason that Elite as a concept is so compelling and popular over the decades.

but the marketing of the game is highly deceiving

And agreed again - selling it as what it is, especially now that there's six years of people having played it and talked about it, would certainly help matters.

 

[ZeeWolf]

To some extent, and that in general wouldn't require many servers. It's more complexity, though - the client needs to (reasonably seamlessly!) be able to switch between full simulation and only handling input/output the moment another player enters the instance, and potentially back again once they leave again. Lag could certainly be compensated for a bit but it'd still probably be jarring suddenly going from "your input is immediately effective" to "your input takes a small but perceptible fraction of a second to cause a response", plus the chances of it being able to switch modes without a couple of seconds of synchronisation (which couldn't easily be hidden behind a loading screen) is minimal. Bugs which only show up when alone / when with other players would certainly be possible, too, though of course plenty of those happen now anyway.

It would, at the very least, be another compromise...

Those "servers" would have to be region specific, so it's however many servers for however many potential combinations of players multiplied by the number of regions to provide a reasonable experience. As you suggested, there's still massive issues around "seamlessly" moving between instances.

I doubt even a a budget the size of WoW would cover it.

If we're to propose dedicated servers, I think it's simpler to re-design the whole game around it. Just provide a small number of arenas to visit and a lobby to wait for a server. Drop all the exploration and model of the Milky Way (it just seems to upset folk anyway) heck it wouldn't be Elite, so drop the Elite name, maybe with the Odyssey stuff it could be more like what the kids want; FPS action in "battle royale" playgrounds that shrink until you are the last man standing! They could call it Odyssey-Bi-Weekly

 

[Rat Catcher]

Agreed. But let's start a competition - just for fun - and everyone shows his preferred marketing speech that a) correctly and in a non-deceiving way describes what the game has to offer and b) is still attractive enough for new potential users. I'm very curious with what ideas you'll come up here...

Maybe: "Elite: Dangerous* - Fly your own spaceships, drive across barren planets, 400 billion systems to explore... "

Notes:
* Other players also fly spaceships and drive across barren planets - everything has guns and these may be used against you.

 

[Rat Catcher]

Fail. Doesn't meet requirement b)
You were just kidding I guess. Ok

My marketing agency's motto is "We never lie!"... it isn't particularly successful... Perhaps I should have mentioned Trading too?

 

[Robert Maynard]

Maybe: "Elite: Dangerous* - Fly your own spaceships, drive across barren planets, 400 billion systems to explore... "

Notes:
* Other players also fly spaceships and drive across barren planets - everything has guns and these may be used against you.

• Other players are an optional extra - the game provides Solo mode and Private Groups to allow one to tailor ones game experience.
• All players both experience and affect a single galaxy state.
• There is no requirement to engage in PvP to participate in any game feature, except CQC.

 

[Rat Catcher]

• Other players are an optional extra - the game provides Solo mode and Private Groups to allow one to tailor ones game experience.
• All players both experience and affect a single galaxy state.
• There is no requirement to engage in PvP to participate in any game feature, except CQC.

It was a light-hearted one-liner to encompass the major points of the game whilst remaining accurate... every vessel & vehicle in ED when purchased has guns already fitted so I didn't alter fact there either. Had I been serious I wouldn't have bothered posting just 2 lines

...but I had forgotten trading...