Players who are not using PC shouldn't be allowed to host PUBLIC missions

[Nekomian]

As pointed out earlier in this thread, this topic has kinda been beat to death before in another post, but some nice points are being made here too.

On 2024-01-25 at 8:23 AM, (XBOX)C11H22O11 said:

Why not? A switch player can join a PC player and get their spawns, haven't heard anyone complain about their game being slower when that happens

I actually do get complaints when I host from switch players who are clients - namely their system gets really hot and sluggish because of all the rendering of entities it's needing to do. Even if it's not tracking to the same degree that the host does, it still has to render them all on the device itself and it seems to struggle with that.

On 2024-01-25 at 11:47 AM, trst said:

As I understand this isn't a "non-PC" issue but an outdated console hardware issue. The spawn rate issues aren't present on the PS5 and Series X and is only a problem with the PS4, Xbone, and Switch. So if you have gripes with the system it should be properly directed at least.

Really though it's an issue that DE will need to deal with sooner or later.

Absolutely this ^ . Especially with mobile ports coming, they'll need some way to address this. It's not a "console vs pc" issue at all, it's just limitations of how entities are being managed by the hardware. Note I'm specifically saying managed, the hardware itself isn't necessarily the problem.

On 2024-01-25 at 5:57 PM, Aldain said:

Wouldn't a better method be to let people mark "I do not want to be a host" for matchmaking specifically (premades wouldn't be effected)?

I'd have that on at all times since I know my connection is not that strong.

Of course, then that would create a problem if more people click off of being a host than don't, but it would at least do something to help.

This solution has been proposed before, but it has too many issues - it complicates matchmaking to an annoying degree and further segments the available pool. You'll have situations where there's no hosts available (like most of the star chart tends to be), and if you enforce "no host" rules then they can't play the mission. If you override it, you either force the player to go solo (defeating the purpose of public matchmaking) or end up in the exact same spot, where that person who may not want to host ends up hosting. There's always going to be a greater need for hosts than clients (since they get outnumbered 1 to 3, assuming full squads) and creating further imbalance will just exacerbate the issue.

On 2024-01-25 at 7:55 PM, Voltage said:

It's even worse when every player on a different platform has the exact same logo. There's no way to know what hardware a player has without asking, and from my own experience, you usually get an angry response when you ask someone their hardware before you invite them.

They obfuscate this because of platform elitism - there's already this stigma some players have against NSW, and this would cause them to be singled out. They're just playing the game same as everyone else, and it should work the same regardless of platform (from a player perspective, obviously the builds themselves are not the same).

On 2024-01-26 at 4:10 PM, ---Merchant--- said:

Computers are connected to lan, this means pc connection is always more stable than wifi devices like switch and phones. Which should give them priority even ignoring the spawnrate issue.

This simply is not true - many consoles are connected via ethernet, and MANY more PCs (specifically laptops) are connected over some god awful decade old router wifi. It's a F2P game, so the barrier for entry is very low on PC - you just have to own something that's made in the past decade. Many PC players are under-spec'd and on wifi, this isn't a console-specific issue. Hell, when I first played this game I was on 802.11g wifi that dropped out constantly and on some cheap dual core acer aspire laptop with integrated graphics.

All that said, they should just unify spawns across all platforms and not punish players with "no host" bans for literally just playing the game. Whether that means lowering PC to the common denominator (not ideal) or optimizing routines on hardware to work similarly across all hardware, it needs to be done for a consistent player experience. This has its challenges, but it's very unlikely to be a hardware limitation by itself and more of an optimization one. Consoles at the very least have set specs and they can consistently reproduce target goals / rates they'd like to achieve - if this means tuning routines or lowering graphic fidelity then those are trade-offs they'd have to make to ensure player experience is consistent (in terms of gameplay here, obviously NSW is not going to look like a PC on all high settings with 4K res).

[Xylia]

Personally I don't get why the game can't run an internet and system test, and come up with numbers and get some sort of aggregate host score, but not display that to other players to avoid elitism.

Basically, for example, Player A benchmarks, say, 50FPS on the highest settings at 4k and has 900mbit. He gets a hosting score of 99.

Player B is running an old PC, is only getting 30FPS on high settings at 1440p and has 150mbit internet. He gets a hosting score of 30.

Player C is running a Playstation, and their internet only does 20mbit. They get a hosting score of 10.

When you group up, the game goes "Who has the highest hosting score?" ... obviously, Player A does, so Player A gets host.

[Nekomian]

19 minutes ago, Xylia said:

Personally I don't get why the game can't run an internet and system test, and come up with numbers and get some sort of aggregate host score, but not display that to other players to avoid elitism.

System test I can understand (just like a CPU / GPU benchmark of sorts), but there isn't any generic "internet test" that would sufficiently work for determining a host. Differing NAT or firewall restrictions can cause problems with some users being host while others work fine, and network speeds alone don't determine things like variable latency (i.e. "best host" has wifi that drops out every 30s that the test doesn't pick up) or other routing issues that could be present. It's why host migrations fail sometimes too, because the new "eligible host" the game has picked has some restriction that let them be a client but not a host to everyone.

Some people would argue that even a system with the best PC specs shouldn't be host with a spotty connection, while some would say even a perfect connection system with 2013 hardware shouldn't host; there's no good way to determine what's best because it's different for everyone and so variable, so I think the existing system just uses RTT latency averages on matchmake and that's it (there's some ELO calcs and like, grouping applied too I think tbf, but they seem to be less weighted than latency) - if there's no eligible session, it makes you host. It's crude / simple, but prevents situations of people in a "waiting for eligible host" queue and weighted metrics that might improperly assign hosts without waiting 30s to maintain network connections long enough to see if there's dropouts.

[Xylia]

You could have the internet test be something that is run upon first booting the game up.

The game does simple speedtest-like thing while you load the game up, or after you log in, and it will note your speed, ping, and whether or not the data download and upload was interrupted. This could be done in the background, and most players aren't going to just jump straight into a mission within the first 30-60 seconds upon logging in, and even if they did, then a score of 0 would be assumed.

EDIT: Also, the game could simply ask you a few questions about your network. "Are you on wifi? Are you on ethernet?" with a  "Don't know" option for those who don't know what that means, which might go at least some way to picking players who have more stable connections.

Or, perhaps some people KNOW they have good hardware and good internet connection, and can go into Options and select "I'm a good host" and upon selecting that, the game will run a test and if they meet certain criteria, then they are more likely to be chosen as a host.
 

Like, myself, I have 900mbit down *AND* up, which I've speedtested several times randomly and I'm getting the full 900mbit and I'm on ethernet. My PC isn't a 4k rig, but I run it on 1440p and I have never seen framerate drops in-game, even in multiplayer matches.

 

EDIT: Tested it just now....

Edited January 29 by Xylia

[Nekomian]

1 minute ago, Xylia said:

You could have the internet test be something that is run upon first booting the game up.

The game does simple speedtest-like thing while you load the game up, or after you log in, and it will note your speed, ping, and whether or not the data download and upload was interrupted.

This works for the one connection you make to a server - you would need to do this for every single potential client on every connection, including NAT traversal for difficult to connect to clients. Again, it's not as simple as a "one time speed test" to identify eligible hosts. I could have amazing ping to a server near me and horrible connections to everyone who connects to me via P2P UDP. I'd reread the first paragraph of my last message, or this article that goes into detail about NAT traversal if you'd like (that further exemplifies determining hosts in P2P scenarios solely for networking purposes is difficult, and this would need to be accounted for in those calculations): https://tailscale.com/blog/how-nat-traversal-works

[Xylia]

Eh, I think if you have a mostly stable connection, that is better than no testing at all, and just giving host to a rando without any sort of testing whatsoever.

You're trying to come up with the perfect scenario, I'm merely suggesting that we pick the "best person for the job".

You got 4 people and if you test the 4 people, you pick the one who has the best speed with the fewest/no interruptions during their speed tests.

Pretty sure that 9 times out of 10, you'll get a good solid host (as long as they also have good hardware too). That would be way better than the chaos we have now, where it doesn't seem to do any testing whatsoever.

Edited January 29 by Xylia

[Nekomian]

6 minutes ago, Xylia said:

You're trying to come up with the perfect scenario, I'm merely suggesting that we pick the "best person for the job".

I'm confused, should you not determine who the best host is? That would be the perfect scenario, determining who everyone can connect to and has the most stable connection out of a pair of 4 clients. It doesn't have to be perfect, but it does need to be considerate of more than just speed and latency.

7 minutes ago, Xylia said:

You got 4 people and if you test the 4 people, you pick the one who has the best speed with the fewest/no interruptions during their speed tests.

Pretty sure that 9 times out of 10, you'll get a good solid host (as long as they also have good hardware too). That would be way better than the chaos we have now, where it doesn't seem to do any testing whatsoever.

It already does it this way, but not for determining session hosts - it sends out packets to all potential hosts on matchmake and determines the best RTT and lowest variance (any latency variations or jitter among the packets). It's in game logs for diagnostic purposes if you or anyone else wants to look.

The problem is this does not account for a lot of situations that are real-world and cause interruptions and/or disconnects. 3 seconds of network testing to one person (more accurately, each host of a session) isn't sufficient enough to detect dropouts, and they may not be the "best host" either, just whoever was available. The proposed solution doesn't really solve anything, we'll still have the same problems unless it accounts for additional conditions, as well as now having to wait for 4 people to queue up in a mission to test them all (because a session cannot start without testing all the people to determine who is "best for the job" without all the people present).