Google promises that NEGATIVE LATENCY will help Stadia not feel laggy

Interesting thread:

There has been a prodigious amount of commentary/joking about this (including from plenty of game people I follow on Twitter) but at least a few people are saying that this isn't as outlandish as it might seem (click through to the thread):

I dunno; it all kinda sounds like snake oil to me, but the way Google talks so rapturously about this whole thing definitely makes me curious. I can't see any kind of game streaming being workable in an FPS (tried Prey with GeForce Now and it was barely playable when you needed to shoot stuff), but maybe it'll be fine for the Assassin's Creeds of the world.

I'll jump in quickly and use an example people will be familiar with, GGPO does this... to a degree.
The idea behind rollback netcode is that it tries to predict future actions and fills in the correct outcomes to reduce or remove perceived latency, it's very clever when it works and that's a simple explanation, but it's a system that can work well for a game like Street Fighter which due to its design is limited in complexity.
A much more robust system will be required to perform similar actions on Stadia but it is possible, especially as the games will be running on powerful server racks and significant resources can be spent on this process, but there's a lot more involved in doing this for Assassin's Creed with all its complexity than a 2D fighter.

"Faster in a year or two" isn't going to help them much if they launch next month with a service that noticeably lags the local installation experience. Their customer base will be long gone.

Games have been doing this for a while, to combat latency. This is the first I’ve heard of it as negative latency, normally it’s more like predictive analytics.

As stated above, GGPO does this. I believe Bungie hyped this sort of thing up for Halo 3, maybe?

Even if it does work as well as they say it does, Stadia still seems like a poor purchase. The bandwidth costs alone (20 gigs an hour per instance of Stadia, if I recall) are unacceptable to anyone with a cap, especially so if you share that cap with a few other people, like myself.

The further loss of game ownership and the history of Google shuttering their various projects are also a big part of what makes this so unappealing. It's a neat idea, for sure. Just not in Google's hands, nor with ISPs being the way they are.

Sounds like a marketing term and it is funny, but yeah, I've heard of this sort of thing in multiplayer games for years now.

This term describes a buffer of predicted latency, inherent to a Stadia players setup or connection, in which the Stadia system will run lag mitigation. This can include increasing fps rapidly to reduce latency between player input and display, or even predicting user inputs.

Why would they even bring the term 'negative latency' in the world?! Surely that sounds more snakeoilish than talking about the opportunity to predict multiple outcomes and selecting the right one for the situation?

'heh, my punches are so fast that it takes like NEGATIVE TIME for them to land heh'

Honestly, it doesn't really matter how they talk about Stadia, i just need to see it in person. You hear people who had a great experience with that Assassins Creed thing and people that felt like it was definitly a step down from playing the game locally. Even in the year 2019, with a decent internet provider that has no caps at all, i still have to reset my router every other day, so it still sounds kind of dicey to rely on the internet so completely.

As other folk much smarter than me explained: "negative latency" is definitely a doable thing

Beyond that: I think "Stadia will have lower latency for most users" is actually a very reasonable explanation that only relies a bit on gate keeping and "no true scotsman" fallacies

I usually spend the holidays with my sister and her husband. When we play CoD on his PS4 I can straight up feel a delay between squeezing the trigger and a bullet (or rocket) firing. I turn off every bit of postprocessing my TV has. They crank all that shit and then some up. To them "better" looking TV is the priority. Combine that with consoles GENERALLY having higher input latency than PC and you have a pretty big hit.

Contrast that with someone on their laptop or desktop. They have a very low latency display and fairly optimized controller drivers. Combine that with some good stuff under the hood of chrome and you are largely down to "just" the network latency. And for the folk who will reliably use Stadia, that is going to be fairly low because google has a data center every fucking where.

And when you consider multiplayer games: You already HAVE the network latency so that barely counts anymore.

---

For people who "care" about this shit, it is going to be worse. Those people already have refresh SKU consoles or gaming PCs and won't use Stadia. For the folk who just want to play Breakpoint on their laptop? Way lower than Breakpoint on their horribly misconfigured TV

I feel like this is just Google using fluff words, lol. We’ll see. I’m still not fully convinced that Stadia will work as well as Google has said it will.

I also generally don’t have confidence in Google supporting this well long term considering they tend to kill off products fairly rapidly when they don’t take off immediately.

I'm already drowning in new games and the ever growing backlog, so luckily I'll get to skip Stadia and observe how it plays out over the next couple of years.

Retroarch and it's emulators lets you try this type of thing locally which is real cool, you can play any mario game with better input lag than it ever had on an NES/SNES. Basically it emulates every single action you might do and when you press it it picks the one you did. https://www.libretro.com/index.php/retroarch-1-7-2%E2%80%8A-%E2%80%8Aachieving-better-latency-than-original-hardware-through-new-runahead-method/

Also Stadia during that test already very low latency as can be seen in this digital foundry test: https://youtu.be/VG06H7IQ9Aw?t=682

they can call their tech 'unicorn blood' for all i care- as long as the user experience is functionally good. will wait to see how stadia performs in the wild before jumping in.

I am excited to report to Google that I am willing to spend NEGATIVE DOLLARS on their Stadia platform.

I will eagerly await their check in the mail.

Negative latency: the controls will move before you even touch them! We are borrowing this from the Nintendo Switch, where it is called control drift!

I will move the stick before it thinks I even will, this should solve any latency issues. Just gotta think ahead of the game, man.

Hey alright, but unfortunately Google Stadia will be nothing but a distant memory "in a year or two".

I'm sure predictive analytics has been around forever and it's likely not how I'm imagining it, but it sounds really bizarre in the case of Stadia. Say I'm playing a single player game, maybe a platformer. Would the 'negative latency' predict that I would jump at 'x' time and perform the input whether I hit the jump button at 'x' time or not? If so, isn't that a bit like the game playing itself?

Not buying it, their assurances or their product.

To explain it, Google Stadia should probably use an ad with their massive server farm replaced with a solitary Doctor Strange sitting in a room computing every outcome of your actions "...in only one of which you win"

Negative latency is a good first step. But for optimal play, the next version of their product will require imaginary latency. Soon after that we will surely see quaternionic latency.

In five or six years when we are in a "Hey, remember Microsoft's E3 press conference? I wish we had gotten that" phase again, remember this thread and all the ones like it.

Bandwidth issues aside*, we have a system that was very widely tested during a free beta last year promoting the use of tech that has existed for years. And most folk refuse to even look farther than "ha ha, fuck that. It is impossible" style responses.

From a tech standpoint, google are surprisingly doing things right (and MS and Sony are pretending they aren't involved) and it doesn't matter.

A lot of that boils down to most of us not being the target demographic (I think this is cool as shit and might dick around a bit, but I have no desire to pay a monthly fee for it), but this is cool as shit from a tech standpoint and it is still kind of bonkers that google "tricked" everyone into beta testing it last year before actually announcing it as a product. Hell, I think it is kind of cool that Google has fucked the dog to death enough times that most of us figured 'Maybe it is a product. Maybe it is a stillborn experiment that we'll never hear about again. Whatever, I got AssOdd for free"

*: That is actually what I am least worried about. Games are so big that even 10-20 gigs an hour is probably breaking even for CoD and maybe even most Ubi games for most players. But more I am counting on Sony and MS and Google using their armies of lawyers and lobbyists to convince/bribe the ISPs to raise or abolish the caps. We'll probably just fuck net neutrality even harder in the process, but it will solve the impending issues of only being able to download one big release a month.

Reading past the buzzwordy sounding "negative latency" this sounds very much like trying to apply predictive heuristics to figure out what action a player might perform next, and already buffer that action before the player actually sends the signal to perform it?

So.. pretty much Minority Report but for player actions? Does that mean that playing games on Stadia will train an ML algorithm in the background, just like we are training self-driving cars when solving captchas? And here I thought this whole Stadia thing couldn't get any creepier/dystopian...