Could what Google calls ‘negative latency’ make Stadia cloud gaming viable?

The term is a little silly, but the idea behind it doesn’t sound too bad—we’re excited to hear what the remote desktop graphics community says about this.

We’ve been following Stadia since Google officially announced it earlier this year after testing it out as Project Steam in late 2018. With Stadia launching November 9, we’ll soon get to see first-hand how Google’s cloud gaming team handles issues desktop virt has for decades.

With under a month to go before Stadia’s release, Edge magazine sat down with Google VP of engineering, Majd Bakar, who spoke about the one aspect at the top of most everyone’s mind: Latency.

‘Negative latency’

This one term is pretty much all anyone has taken away from the interview, because it’s sort of a silly-sounding phrase, but there may be something to the actual idea of it.

What will likely make or break Stadia, along with any other cloud gaming service is latency, with a big component being one’s location to a Google datacenter. But, in the future, Stadia might be able to improve the experience with something called “negative latency.”

Talking a big game, Majd says that possibly within a year or two Google Stadia could “have games running faster and feel more responsive in the cloud regardless of how powerful the local machine is.” They added that Stadia could even surpass the local console experience with what Google calls negative latency.

So, what is negative latency? It’s a term that covers a few features aimed at optimizing the cloud gaming experience by reducing latency. This might manifest a few different ways such as providing users with a super-fast framerate to reduce latency, or use the compute power of the datacenter to predict a user’s actions to make it appear as if there isn’t any input lag.

Some aspects of negative latency and how well it’ll work for Stadia don’t make apparent sense to me just yet—but remember this is a future thing (albeit a close future)! Like we’ve written before, there are plenty of opportunities for the more experienced and knowledgeable VDI/remote graphics community to step up and break down whether this is feasible or a pipe dream.

But, there might be something to what Majd described. While providing a super-fast framerate might seem at odds with preventing latency issues, the increased framerate could prevent the user from noticing any issues in case there’s a hiccup between the user and the server. As for predicting user input, there is already some precedence for it. The Evolution Championship Series uses GGPO (Good Game Peace Out) netcode that simulates frames based on the player’s previous actions to predict what they might do next and provide a smooth experience for remote players.

I spoke with Rachel Berry, and she felt that predicting behavior might be naïve, at least when it comes to multiplayer games. She noted that one possible way this could work, however, would be maybe sending multiple predictions at a lower depth and then upscale the quality, which might be enough during frenetic action scenes. Back in 2015, Rachel was involved in a demo showing that 55fps using HDX and Citrix XenDesktop on a low-cost thin client--imagine what Google can do with whole datacenters.

Still plenty of questions and things to learn about

As neither Jack or I are remote graphics people, some of the technical aspects still escape us, so we’d be curious to hear from someone who is more capable of breaking this down and saying whether they see this as a potential success or doomed to failure like OnLive.

Not just any organization can pull off cloud gaming, but both Google and Microsoft at least stand a chance. (Admittedly, Google is still questionable given their habit of ditching projects and never actually having dealt with game development before. But, they have opened a brand-new game studio to support Google Stadia.)

The biggest hurdle for any cloud VDI experience is whether the vendor has enough datacenters near their customer base to keep latency to tolerable levels. Luckily, this is something both Google Cloud and Microsoft Azure have covered. (for example, Workspot found that most everyone is within 50ms of an Azure datacenter—check out Microsoft’s tool.)

I hope to try out the different cloud services myself to see what the experience is like. I did get to try out Stadia back at RSA 2019 and didn’t notice much input lag while playing Assassin’s Creed Odyssey, but did notice a graphical downgrade during frames where I was moving—standing still it look as good as what you get on consoles.

And then, down the road, it will be interesting to see Google’s “negative latency” in action over the course of a gaming session—what if a user takes an action that isn’t predicted, will one see an inconsistent experience?

Join the conversation


Send me notifications when other members comment.

Please create a username to comment.

well our mobile network is getting faster and also home / work internet so I dont see any issues with this. If they can do it as great as Netflix does it will be a success. Nvidia also have their Geforce now platform they offer to telcos, which is a great platform if you wanna do game streaming. Their are rumours that Microsoft and Sony as well is coming with streaming platforms for gaming
Watching video and playing games are different beasts, but I do think you can build off the lessons Netflix and others have learned. Microsoft has xCloud, which went into public preview last week and Sony technically has PlayStation Now.