Netflix Confirms It's Looking To Launch a Cloud Gaming Service

Netflix wants to extend its nascent gaming efforts to PCs and TVs, and it's looking to launch its own cloud gaming service to do so, VP of game development Mike Verdu confirmed at TechCrunch Disrupt on Tuesday. Protocol reports: "We're very seriously exploring a cloud gaming offering," Verdu said. "We'll approach this the same way as we did with mobile -- start small, be humble, be thoughtful -- but it is a step we think we should take," Verdu added. "The extension into the cloud is really about reaching the other devices where people experience Netflix."

Verdu didn't share many additional details, but suggested the company was looking to launch more than just casual games on TVs. He declined to say whether Netflix would build its own game controllers like Google has done for its failed Stadia service, but he said the titles wouldn't rely on TV remotes for input. [...] Verdu called Netflix's expansion into gaming a pivotal moment for the company, but admitted that it was a slow and deliberate multiyear effort. However, the company may already be seeing some rewards from those efforts. "We're seeing some encouraging signs of gameplay leading to higher retention," the company wrote in its letter to investors Tuesday. The report notes that Netflix has released a total of 35 games for mobile devices thus far, with an additional 55 games in its pipeline. "Fourteen of those games are being built by Netflix's own studios, Verdu said, adding that the company was going to launch an additional studio in Southern California soon."

Further reading: Netflix Password-Sharing Crackdown Will Roll Out Globally In 'Early 2023'

Buzzword Bingo Management.

[splutty]

I have no idea where they want to go with this, and I don't think they even know where they want to go with this.

"Start small" is rather indicative to that.

With all the 'cloud' services currently available, I doubt Netflix has the clout, power, knowledge or staying power to get anywhere..

Re:

[ls671]

I guess they should instead accept that growth isn't perpetual, accept to get smaller but get better at the same time. Concentrate on what they know and do best even if their business sizes down on revenues.

It kind of reminds me Facebook with their Metaverse. Maybe the analogy is wrong and the Netflix case makes more sense but all these tech linked companies should all realize that people are going to get tired of what they offer and eventually move back to more traditional pastimes or new ones that don't n

Re:Buzzword Bingo Management.

[fazig]

I think a fundamental problem with the "cloud gaming" approach is to try to make games that were developed to run on localized hardware and be played on that same localized hardware to run on a remote server and be played on a localized client .
Because unless the gameplay was designed to be very friendly towards higher input latencies, as well as some jitter in those latencies, the gameplay is likely to feel off.

With that in mind, it ought to be possible to have games that play more or less nicely in the "cloud", if they were specifically developed to be played that way. Of course that would mean investing into game development, which they are seemingly doing. So the question that remains is if their game developers can come up with games that people find enjoyable enough to be willing to pay money for them.

Given the stupid amounts of money crappy mobile games with their predatory monetization features make, there's certainly a market for this. Not a huge overlap with what you and I might consider "gaming", but still a quite substantial market.

Re:

[AmiMoJo]

Most games are designed with fairly high latency. For example, Street Fighter V has a built in 8 frames of input lag (133ms) which allows the use of rollback netcode to make online matches play exactly the same as local matches.

Even games which don't deliberately add latency tend to be designed with quite a bit of it in mind, since they have to account for the worst possible case of a low frame rate and very high latency monitor. The number of games that have low latency twitch controls is pretty low now.

Re:

[fazig]

Yes, that would be an example for a game that should play nicely enough in "the cloud".
Of the tick rate for the gameplay logic on the local hardware (not to be confused with server tick rate) is only 8fps .

But you're not likely to have a lot of fun with competitive Counter-Strike or similar games that were designed for fast paced gameplay and to work well with mouse inputs. Button presses and joystick inputs give a bit of leeway in how inputs can be processed. But mouse movement is a lot more stringent d

Re:

[AmiMoJo]

Street Fighter 5 ticks over at 60 FPS internally. It's just that it has an enforced delay of 133ms between it receiving a button press from the controller, and that button press being acted on.

That way as long as the network latency is less than 133ms, if the remote player presses a button and it takes 100ms to reach your console, it just adds another 33ms on so that the total latency is always the same for both players. They call it "rollback" because the game behaves as if the button was pressed 100ms ago

Re:

[fazig]

If the effects of the inputs only become apparent in 8fps intervals, then that is the effective tick rate for the gameplay logic as well. The frame rate at which it is rendered to the screen may much higher.

It's not uncommon to do it that way in video games.
Diablo 2 Resurrected comes to my mind off the cuff. That game preserves the original game mechanics by running the gameplay logic at the tick rate of 25fps (which back then was also the rate at which the game was rendered). So if you want to think abo

Re:

[AmiMoJo]

I'm trying to explain to you that it's not 8 frame intervals. It's 8 frame delays. 8 frame intervals would imply that it only polls for inputs every 8 frames, but it polls every 1 frame.

Re:

[fazig]

Probably a miscommunication here.

To clarify a hypothetical situation: Say the game is running at 100fps on the screen. Input A happens on the 1st frame of the screen frame rate. Input B happens on the 5th frame of the screen frame rate. Then A will take effect in the 1st frame+133ms and B will take effect in the 5th frame+133ms?

Re:

[AmiMoJo]

SF V is locked at 60fps. Beyond that, yes.

Button press on frame 1 -> actioned on frame 8
Button press on frame 5 -> actioned on frame 13

That way if the other remote player presses a button on frame 1, but due to network latency news of it only arrives on frame 6, the game can still action it on frame 8. The network latency is invisible.

Re:

[fazig]

Got it.

I also watched some internet multiplayer footage. Together with the "freeze frames" they seem to utilize whenever an attack "connects" that should be enough to for small scale multiplayer (like a Street Fighter game would have) over distances like US West Coast to US East Coast, assuming no other network issues. It it could even work from US West Coast to Western Europe well enough.

Re:

[Drethon]

Street Fighter 5 ticks over at 60 FPS internally. It's just that it has an enforced delay of 133ms between it receiving a button press from the controller, and that button press being acted on.

That way as long as the network latency is less than 133ms, if the remote player presses a button and it takes 100ms to reach your console, it just adds another 33ms on so that the total latency is always the same for both players. They call it "rollback" because the game behaves as if the button was pressed 100ms ago, i.e. it rolls back the state of the other player and applies the button press retroactively.

The actual internal game timing is handled at 60 FPS. The latency is 133ms from the end of the 16ms frame you pressed the button in.

And then very interesting things happen if someone hacks the timestamp of when the button press was sent?

Re:

[fazig]

Usually you try to check everything another client sends to you to see if it is plausible at the least, because Never Trust the Client.

Though in practice these mitigation methods have gotten difficult and ineffective, especially with all the peer to peer networking that's been used by consoles (among other things to keep costs down). Thus in practice where cheating is even a concern (it often isn't that much of a concern) you'll often find anti-cheat software at least partially running on the local machin

Re:

[timeOday]

I actually think cloud gaming is an idea whose time will come, sooner or later. However, netflix existing infrastructure was all built for movies, where latency doesn't matter (within a few seconds), so I would imagine it's deceptively difficult to get from where they are to cloud gaming.

The fact that google just quit the idea (Stadia) is what makes it so surprising to me that Netflix thinks the time is now?

Netflix has cacheing servers at ISP good for non l

[Joe_Dragon]

Netflix has caching servers at ISP good for non live content + your local device also does buffering as well.
Not at all build for live log lag game steaming.

Re:

[Gilgaron]

Yeah game design would be a key component. I've played with Steam local streaming and it worked well enough for a single player RTS like Total War but it is hard to see it working well enough for an FPS like Doom, and that's within my own network. Presumably some sorts of mobile type games would be lightweight enough, too.

Re:

[Tyr07]

I disagree, I think the fundamental problem with cloud gaming is that it's a product no one wants. They want to be in control of your entertainment and charge you for how long you're entertained and no one is interested in that.

Re:

[fazig]

Total DRM does appear to be one of the main motivations of larger corporations to push cloud gaming.

But beyond that there can actually be some merit to it.
For example when I think of multi-player physics simulation, it could be immensely beneficial if all the simulation (that includes collision detection which is usually done on the client side) is handled by a centralized server that makes sure the results of the simulation are all the same (common approach, where a dedicated server or one client that al

Re:

[Tyr07]

This is what servers do currently without a cloud gaming system. The only difference is the servers we currently use for gaming in a client / server setup are friendlier to people with poor latency.

Currently in an FPS or space sim to give some broad examples, all the physics is calculated server side often using lockstep to keep everything in sync. It has some leeway, roll back and compensation for people with latency.

We could accomplish the very thing you're talking about very easy with this kind of setup.

Re:

[fazig]

Lockstep is hardly used because that really fucks with the clients perceived performance, where the simulation stalls and has to wait until it's synced again, if the ping is jittery. There is no leeway.
It is used sometimes and it can work exceptionally well if latencies are low and most importantly consistent. This is something that is essentially impossible over larger distances with our current internet infrastructure.

So most commonly it's some predictive algorithms that are at work on the client side,

Re:

[Tyr07]

You made some very good points, definitely on the extreme end of performance having people have the same level of experience is nice, your arma example was a perfect example

The same thing applies to screen gamma and other settings. I know in Natural Selection 2 I turn off infestation because marine players do, and they'll hide behind it, you can't see them and they'll shoot you. However I wish I could leave add effects like that on and not be at a disadvantage.

I think this will swing back to though, the ext

Re:

[fazig]

The memory requirements on the server side wouldn't increase in that way.
What mostly scales with the amount of players is the rendering of the world into a given player "camera" that is inside the world. There you would need increased resources for the CPU side of the rendering pipeline as well as the GPU side.

But all the physics simulation, AI (NPC) simulation (like A* search with a dynamic navmesh), and a lot of other game usually does not depend on perspective. These parts could be done in a centralize

Re:

[Tyr07]

Client side prediction is usually limited and just generates the illusion of it happening in realtime, quite often the server is still the authority on what actually happened, which is why in some cases people "lag" back into position, or a shot they thought hit something didn't.

In some cases the clients predictions are actually submitted to the server but those games are ripe with cheaters. Like GTA 5 as an example.

Actually I think the memory requirement for the server would scale way, way more than even

Re:

[fazig]

In most cases the clients predictions are submitted to the server in some fashion.
It's just that the server has the last say on the matter, so it will likely discard them and have the client with the "out of sync" prediction roll back, roll forward, or like Overwatch does it, temporarily speed up or slow down the local simulation on the client.

Though, why would you render the entire world in full detail all the times?
You only have to render what the individual player cameras would see. So essentially f

Re:

[Tyr07]

Something I forgot to add.

With the current system, you're decentralizing the rendering allowing individual systems to contribute their system to offload rendering from the server side, which makes scaling possible as each client adds to the pool of hardware to do that rendering, so servers only have to detail with updates on where players are, where they are looking, if they're running into a collosion object, bullet tracjetories using vectors and more.

Actual rendering is such a whole different beast it com

Re:

[thegarbz]

Because unless the gameplay was designed to be very friendly towards higher input latencies, as well as some jitter in those latencies, the gameplay is likely to feel off.

I've heard that many times, but honestly having played FPS games on Stadia (despite not having a Stadia datacentre in my country), it didn't feel off in the slightest. It stuttered occasionally but then a lot of poorly optimised games do that on local hardware too.

Now that said I'm not a professional e-sport champion, so I wouldn't use Cloud gaming if I'm trying to win some competition. Those people may notice a difference, but for the 99% of the rest of us, if you have a local datacentre chances are input

Re:

[fazig]

Sensitivity to input latency varies between people. Sensitivity likely also changes with age.
Of course some people may even just imagine it, where if you put them into a blind test, they wouldn't be able to tell the difference, but still insist on something like 1ms less input lag making a great difference. Or tangentially related insist that a 360Hz refresh rate monitor is so much superior to something like 144Hz. LTT (which I usually do not follow closely) did a not very scientific test about it here htt [youtube.com]

Their servers are god like

[rsilvergun]

So I can understand why they'd want to leverage them. But given that Google just failed spectacularly, leaving a ton of indie devs high and dry, I think they'll have a damn hard time getting content.

Re:

[thegarbz]

Google's failure can be largely pinned on it's fucked up business model expecting you to buy games on its platform rather than either a) licensing games for anyone to stream (basically every other Netflix business), or b) playing games you already own (like NVIDIA Geforce-Now which hasn't shutdown).

Re:

[quantaman]

I have no idea where they want to go with this, and I don't think they even know where they want to go with this.

"Start small" is rather indicative to that.

With all the 'cloud' services currently available, I doubt Netflix has the clout, power, knowledge or staying power to get anywhere..

How many people started watching Amazon Prime TV because they had a Prime subscription for shopping, and then they kept the Prime shopping subscription because they were watching Prime TV?

If Netflix releases some decent free games accessible via Netflix subscriptions I could see that helping them keep subscribers. It's all in the execution of course, but it's not the most terrible idea.

game streaming service pricing model

[Anonymous Coward]

The question that will decide it's fate right at the start is the pricing model.

This, of course, depends on what the game companies will agree to.

The reality is that to cover the costs of the running the service and make a worthwhile profit they need to charge about $2.25 per hour, with
the AAA game companies getting about a 70c per hour cut.

The Google Stadia model of paying $50 up front for one game won't make a profit in the case of a customer that has it running for a few hundred hours making use of a com

$2.25? what about $6.99/hr for in room N64?

[Joe_Dragon]

$2.25? what about $6.99/hr for in room N64?

Re:

[torkus]

Adult Me: $2.25 an hour to play games seems pricey and unsustainable
Teenage me circa 90's: 25c per 2 minute arcade play... where's the quarter-changer?

IDK if $2.25/hr for at-home gaming is sustainable but I guess the market will tell.

Your complete divided attention please

[pereOlthwaite]

I suppose 100ms of input latency is fine if you're playing Animal Crossing, which roughly half of Netflix's audience generally are.

Re:

[thegarbz]

Why would you have 100ms of input latency? Are you playing cloud games on a 56k modem? Do you live in the middle of the pacific ocean?

Re:

[pereOlthwaite]

Google appears to think that the 85ms of steam-gaming input latency on their new stream-gaming-specific line of Chromebooks is some kind of an achievement. There's a whole lot more to it than bandwidth and ping.

Good idea, Netflix!

[atomic_toast]

That's a great idea because it's been so successful for everybody that has tried it.

Obvious choice

[Locke2005]

Because Stadia worked out so well for Google!

Re:

[thegarbz]

Because Stadia worked out so well for Google!

Stadia had a fucked up business model which made no sense, making you re-buy your games if you wanted to play them on that platform. They were were a victim of their own very specific design flaw. There are other services which haven't shut down and didn't make such stupid decisions https://www.nvidia.com/en-us/g... [nvidia.com]

Re:

[Locke2005]

So you're saying failure had nothing to do with the lag introduced by every controller command having to go round trip to the server and come back as streamed video, introduce several hundred milliseconds of lag to game play? That's the reason I never tried stadia.

Utilization

[Ritz_Just_Ritz]

They've gotta do something with all that infrastructure as their streaming customers bleed out to greener/other pastures. :)

Don't want full priced games locked to an Netflix

[Joe_Dragon]

Don't want full priced games locked to an Netflix sub fee that you can't take to an different platform, may lose access to if you don't pay for Netflix?
Hell they may be forced one day to add ESPN and then if may be like if you don't pay the starting at $29.99 basic TV free + the $5-10 games fee you may lose all of your paid in full games.

ISP's with local servers are in an better place to

[Joe_Dragon]

ISP's with local servers are in an better place to offer this from an LAG standpoint.

Couldn't they just buy Stadia?

[Provos]

Seems like an opportune time - google wants it gone, Netflix wants something to do 'cloud gaming.'

No.

[muh_freeze_peach]

Did they fix the input latency? Did they fix the compression artifacts? I guess this kind of crap would be okay for casuals.

Perfect timing

[Opportunist]

Google's competition is just on the way out, everyone else who ever tried it already folded, so now's the perfect time to monopolize the market!

Why Streaming?

[KermodeBear]

I don't understand the obsession with the whole video game streaming model.

They need to maintain a large number of high-powered machines with expensive graphics cards, and pay a lot of cash for very high quality bandwidth (extreme low latency). Then, they need to find consumers who A) are interested in the product, and B) have an extremely fast network connection. More than a few ms and input latency is noticeable, making games unplayable.

Google couldn't pull it off. Not saying that Google has all the answe

Re:

[Gilgaron]

Probably retention of user base vs Google trying to get more people to subscribe. Kinda like how Prime Video has a large user base because it is free alongside Amazon Prime.

No one wants this

[Tyr07]

Companies want this, companies want you to pay for everything you do, every minute of your life should some how be money towards them and their services. Playing a game you only paid for once? No way! You're still enjoying it? You need to give us money!

Every company is literally to look to charge you not for the service or products you buy, but literally want to charge you for being entertained, and the longer you're entertained, the more they want you to constantly pay them.

It seems to drive them nuts if s