Star Citizen - Chris Roberts, lots of spaceship porn, lots of promises

[Timex]

This is not a “simple shooter” (and Planetside 2 would like to have a talk with you). This is game that promises to have hundreds or thousands of people playing in one server (or on multiple servers that “feel” like one), interacting with each other, and many of its promises actually depend on that kind of thing.

Just to be clear, this isn’t what they are talking about. The numbers they are shooting for isn’t going to be everyone in the game on one server.

The servers they are talking about here are running a single instance of a particular location of space. You are never going to see more than a few dozen other players, tops, in any one server like that.

Here’s an old design article talking about the instancing stuff:

Chris Roberts on Multiplayer, Single Player and Instancing - Roberts Space...

So, with that in mind, the issue you describe isn’t quite of the magnitude that you thought it was. It’s still an issue, which is why it’s being tracked as such… But it’s not the ultimate fail you might have thought it was.

[rhamorim]

The servers they are talking about here are running a single instance of a particular location of space. You are never going to see more than a few dozen other players, tops, in any one server like that.

Even then, that’s a problem, especially when you consider multi-crew ships. And if they want to have people invading the big ships and having firefights in them, 12-15 players is an awful low limit. If we’re talking about instances, they should shoot for something in the 32-64 range at least, if they want to deliver the kind of experience they promised (and that’s compromising on what they promised, actually). And I’m not even counting the “NPC population” they promised and the consequences of that depending on their implementation (oh boy, in technical terms, that will be a tough nut to crack).

And I’m not even getting into more direct engineering/non-functional concerns like what would be the impact of having low player counts per instance in terms of infrastructure and inter-server communication (and latency/bandwidth requirements). What little they said about that particular problem was little more than hand-waving at this point, and the kind of limitation I pointed out only makes the issue worse.

So, no matter which way you see it, it is in practice a big deal. This issue might actually invalidate the whole premise of that part of the project. I assume they’re doing something to correct that, but I do think that adding that kind of issue in the schedule report is the tacit admission that they are having bigger issues to fix this than they anticipated, and will release 3.0 in that state (because they have little choice), and that’s troubling to say the least.

[Timex]

So, no matter which way you see it, it is in practice a big deal. This issue might actually invalidate the whole premise of that part of the project. I assume they’re doing something to correct that, but I do think that adding that kind of issue in the schedule report is the tacit admission that they are having bigger issues to fix this than they anticipated, and will release 3.0 in that state (because they have little choice), and that’s troubling to say the least.

Yes, it’s clearly an issue, which is why they are tracking it. But the idea that it will somehow compromise the entire project is kind of silly. It’s a scalability problem. You run into those once you start doing testing with more players. They’ll just work the problem and fix it. They’ve hit similar stability issues in prior passes and dealt with them.

[rhamorim]

Yes, it’s clearly an issue, which is why they are tracking it. But the idea that it will somehow compromise the entire project is kind of silly. It’s a scalability problem. You run into those once you start doing testing with more players. They’ll just work the problem and fix it. They’ve hit similar stability issues in prior passes and dealt with them.

I’ve seen software projects go from possible to not viable and cancelled because of lesser issues (and many critical issues that make projects non-viable are exactly of the “non-functional” type, which includes scalability), but, for the record, I hope you’re right. I’d say it’s certainly a fixable problem (since there are games that have solved it), but I’m under the impression that it might be particularly difficult to fix in CIG’s case due to the engine they use.

[Timex]

At this point, the networking code is totally custom. There’s nothing left from the original CryEngine netcode.

[Juan_Raigada]

What about Minecraft for example was less transparent? That it was a smaller project doesn’t mean it was less transparent, that’s you appearing to not even try to be fair. It’s simply shifting the goalposts.

I think minecraft never had a production schedule up (I think even the design goals of the project were never stated until very close to release). These guys really are showing how the sausage (a perhaps too ambitious sausage with a shady business model, to be fair) is made. The transparency they have is astounding. Only thing I can think of to improve it would be to release the budget estimations per task/department.

This is obviously a problematic production, but it’s not a highly unusual one in the amount of problems vs. scale they are facing. If this was a big studio game, they might have been canceled by now (maybe not) because it’s just too big a game to be on it but they found a business model that allows it to struggle onwards even if it’s a suspect business model indeed. I also don’t trust their ability to finish it after having seen the speed of progress (the breaking point for me was about ten months ago) but actually seeing their production schedule makes me more confidant in the prediction.

Of course, dialing down the 100 systems to about 15 puts it back in the realm of possibility (IMHO their biggest issue is content creation, not tech. They are doing really cool things with the tech) , but that’s a very significant change of scope.

[Timex]

Given their planned method of discovering jump paths between systems, growing the overall game world over time should work pretty well after initial release.

[rhamorim]

At this point, the networking code is totally custom. There’s nothing left from the original CryEngine netcode.

Well, if they still have issues with the netcode even with most of the code replaced by theirs… it makes me wonder what’s going on. Well, I hope they fix the whole thing soon enough.

[Timex]

Well, if they still have issues with the netcode even with most of the code replaced by theirs… it makes me wonder what’s going on. Well, I hope they fix the whole thing soon enough.

Eh, they’re just hitting a bug. It happens. They’ll work the problem and fix it. To some extent, this is the point of the different tiers of testing they have. They’ll probably bit more stability issues as things progress.

[Ryan_Kelly]

Given their planned method of discovering jump paths between systems, growing the overall game world over time should work pretty well after initial release.

From what they’ve said, it doesn’t seem like they have any plans to abandon the original starmap, including all of the lore attached to it. Thus, if they were to bring some of these systems online after release, I doubt they would have the players discovering jump points that were already discovered centuries ago according to their own backstory. Especially when the systems are already inhabited by humans.

[dsmart]

At this point, the networking code is totally custom. There’s nothing left from the original CryEngine netcode.

Yeah that’s patently FALSE. You just made that up in your head. Even CIG have made no such claim.

Fact is, the netcode is the same baseline kernel from CryEngine 3.x and which they have been massaging since day one in order to get it to work well with more than 8 clients joining an instance without the server falling over. And that’s the hallmark of the shitty netcode that CryEngine has been famous for since the very first Crysis game.

CIG has been promising new netcode since early 2016. It was supposed to be in variants of 2.x. Then they discovered what a shit-show it was going to be to actually make any reasonable changes without affecting the normal release schedules for builds.

So they deferred it to 3.0. That didn’t work out so well either. So it’s no longer in any of the 3.x builds. And I suspect that they have dropped the idea of replacing it, and instead are going to continue kludging it. That aside from the fact that their switch to Lumberyard had done nothing but complicate things even further in this regard.

[dsmart]

Just to be clear, this isn’t what they are talking about. The numbers they are shooting for isn’t going to be everyone in the game on one server.

The servers they are talking about here are running a single instance of a particular location of space. You are never going to see more than a few dozen other players, tops, in any one server like that.

Here’s an old design article talking about the instancing stuff:

https://robertsspaceindustries.com/comm-link/engineering/12770-Chris-Roberts-On-Multiplayer-Single-Player-And-Instancing

So, with that in mind, the issue you describe isn’t quite of the magnitude that you thought it was. It’s still an issue, which is why it’s being tracked as such… But it’s not the ultimate fail you might have thought it was.

That’s all bollocks. All of it.

First, that’s not a design doc/article, it’s the classic literal diarrhea that Chris tends to write and spout in order to keep backers in the dream.

And none of it currently exists.

If you think that 6 years in, if they haven’t got a SINGLE aspect of that actually working yet, that they’re EVER going to do it, then you’re a fool. Go buy an Idris and prove me wrong.

The game is always going to remain an instanced session base game because unless and until they are able to get a decent networking framework working - which by all accounts they can’t - they’re not building any of that shit.

We’ve actually built it. With a combination of custom code and middleware (RakNet). I know it’s a nightmare. And it took the better part of 3 years to do. And it supports both open world and session based.

There are only two known middleware networking engines (RakNet and ReplicaNet. both of which I have used in my games) which are capable of coming anywhere close to doing what they want. And you still need custom code to handle the world entities, updates etc. Unlike graphics and AI, networking in games is the single most difficult thing to do. Bar none. Especially where real-time communication is required - in twitch based games.

At this point, the best that backers can hope for is that CIG at least gets 16 client sessions working with some degree of stability, because there is no way they’re going to go beyond that with any reasonable performance or stability. Heck, look at how Elite Dangerous did it. And that’s not even an MMO. And somehow you guys think that CIG is suddenly going to do anything like that, or even surpass it? Fucking LOL!!

Further reading:

http://www.scqa.info/transcript/?episode=10FTC-053

https://forums.robertsspaceindustries.com/discussion/comment/5284313/#Comment_5284313

[Grifman]

ls-up bankrupt so they can be a cautionary tale. (The only person I would feel for is Brian, and I may crowd-fund his $120 just so I can get that off my conscience.)

Why would you want something bad to happen? Should I wish for something bad to happen to you so that it would be a cautionary tale?

[dsmart]

From what they’ve said, it doesn’t seem like they have any plans to abandon the original starmap, including all of the lore attached to it. Thus, if they were to bring some of these systems online after release, I doubt they would have the players discovering jump points that were already discovered centuries ago according to their own backstory. Especially when the systems are already inhabited by humans.

They don’t. The starmap is the entire game world. Like everything else, that nonsense about "discovering jump points’ went out the window five minutes after Chris uttered it.

Heck, even in the upcoming much touted all-elusive 3.0 they had to “move” a planetoid all the way from a different system, Nyx, in order to have a planetoid and moons in the Stanton system. Why? Because 1) they haven’t built any other system yet 2) there is no travel between star systems - which is they rather than build an empty Nyx, and leave Delamar there, they decided to move it to Stanton.

It’s fucking hilarious, really. I wrote extensively about it already.

[dsmart]

Even then, that’s a problem, especially when you consider multi-crew ships. And if they want to have people invading the big ships and having firefights in them, 12-15 players is an awful low limit. If we’re talking about instances, they should shoot for something in the 32-64 range at least, if they want to deliver the kind of experience they promised (and that’s compromising on what they promised, actually). And I’m not even counting the “NPC population” they promised and the consequences of that depending on their implementation (oh boy, in technical terms, that will be a tough nut to crack).

And I’m not even getting into more direct engineering/non-functional concerns like what would be the impact of having low player counts per instance in terms of infrastructure and inter-server communication (and latency/bandwidth requirements). What little they said about that particular problem was little more than hand-waving at this point, and the kind of limitation I pointed out only makes the issue worse.

So, no matter which way you see it, it is in practice a big deal. This issue might actually invalidate the whole premise of that part of the project. I assume they’re doing something to correct that, but I do think that adding that kind of issue in the schedule report is the tacit admission that they are having bigger issues to fix this than they anticipated, and will release 3.0 in that state (because they have little choice), and that’s troubling to say the least.

It’s all pipe dreams. None of that shit is ever happening. The client instances can’t sustain more than 8 players reasonably well, and backers actually think they’re going to be flying around in capital ship chariots with a dozen of their best mates inside.

It’s hilarious.

[sharaleo]

I’ve looked for updated info on ‘StarNet’ and it seems difficult to come by. Whatever their plans for netcode customisation/re-write a year ago, I am going to be unsurprised to eventually learn it went largely out the window with the switch to Lumberyard and the networking features AWS has added to that.

[SlapBone]

Why would you want something bad to happen? Should I wish for something bad to happen to you so that it would be a cautionary tale?

I’ve explained multiple times in this thread why I want this game’s development to crater but your false equivalence is noted.

[Jason_McMaster]

The scope has significantly expanded since {insert feature} was announced on {insert date} therefore it is wrong to criticise CIG and this is in fact a very positive thing because {insert fatuous fawning}.

This made me smile

[dsmart]

I’ve looked for updated info on ‘StarNet’ and it seems difficult to come by. Whatever their plans for netcode customisation/re-write a year ago, I am going to be unsurprised to eventually learn it went largely out the window with the switch to Lumberyard and the networking features AWS has added to that.

Yeah, there’s that. However, the way CryEngine is designed, not only is it an arduous task to replace the networking kernel, but what AMZ has done to their Lumberyard implementation makes it largely insurmountable to even attempt it. Especially due to all the AWS hooks they also spent the better part of a YEAR implementing.

Basically CIG is never - ever - going to build an MMO from this. Like at all. And that “server mesh” nonsense they were going on about, which is similar to how ED does it with their instancing, is also out the window because they switched to LY.

I wrote about this back in May.

[KevinC]

I’ve explained multiple times in this thread why I want this game’s development to crater

That did strike me as a bit of a “but why male models?” question, but I mean that in a good natured / teasing way. It can be really hard to keep up with this thread sometimes.