We're Excited To See What Google's Project Stream Can Do

Is video game streaming going to change the industry as we know it? On Kotaku Splitscreen we discuss that question and much more.

Kirk was on jury duty last week, so Maddy and I recorded and edited the show all by ourselves like grown adults. First we talk about Bloodborne, Danganronpa, beloved Nintendo of America president Reggie Fils-Aime retiring, and the rumours of a Nintendo-Xbox love affair. We take some listener questions on Anthem and games as a service (35:10).

Then it’s time for off-topic talk (52:40) on X2 and Russian Doll as well as my surprise Music Pick Of The Week. It’s a good one.

Listen here:

Get the MP3 here, or read an excerpt:

Jason: ...There’s definitely something to the idea that [Microsoft and Nintendo] are working together.

Maddy: They’re buddies now—what is going on? The console wars are over! There’s some sort of truce happening. I don’t know, it’s messed up now. Whole new world.

Jason: I think the part of this equation maybe people aren’t paying enough attention to but will be soon is streaming and what that will do to games, because that’s going to change everything. I’m incredibly excited to see what Google shows at GDC.

Maddy: We’re gonna be there! We’re gonna see it.

Jason: Yep, you and I will be there, and hopefully we’ll get to talk to some folks over there. They announced that they’re doing this big GDC keynote. All shall be revealed, that’s their big thing. I had reported in the past that they have this streaming thing, they have some hardware - I don’t know if the hardware’s going to be there; I’d heard rumblings that it might not be - but the streaming platform, if it works as advertised, I think that could change everything we know about the way video games are made and played.

Maddy: People have said this about streaming and games for a while, though.

Jason: But there hasn’t been a company on the level of Google coming in and saying, ‘Hey, we’re going to bring this to mass market, make it so everyone, no matter what equipment they have, can pick up their phone, PC, whatever, and open up a Google Chrome tab and play Assassin’s Creed on it.’

Maddy: That sounds too good to be true to me, but hey.

Jason: Did you try the Project Stream beta test at all?

Maddy: Nope, did you try it? Was it awesome?

Jason: Yeah!

Maddy: Then I’m no longer sceptical. I’m totally bought in. I’m ready to play Assassin’s Creed in Firefox or whatever.

Jason: Well no, not Firefox. Never Firefox.

Maddy: Of course not! Chrome. Google Chrome.

Jason: There’s a tiny bit of latency that I noticed, and I was on a high-speed internet connection. Obviously this isn’t going to work if you’re on bad internet or your bandwidth is capped. But when I played it, it ran pretty smoothly. It was at 30 frames per second, not 60, but it looked good. I could play the game.

And I was baffled by the fact that I was sitting in the office on my Macbook playing this in a Google Chrome tab - it was just mindblowing. So I’m excited to see what they have to show, and what their capabilities are, and what they can actually do with this thing. I know they’ve been funding games, so maybe we’ll see some of those at their presentation.

Maddy: It’ll be fun. It’ll be fun to be there.


    Google Chrome? I'm gonna have to download more RAM then.

    Physics says nope.

      There are ways to do it, it all comes down to bandwidth and latency. One part says the physics should be fine, other says probably not.

      Get a big enough data pipe, and you can shove as much info down it as you want. It comes down to kb/s, and with a modern connection that makes for a lot of data that can be sent. The physics should be OK with that, our connections can handle the load. We can stream 4K now for example.

      Its the latency that will be a bigger issue. No matter how you cut it, you're still probably sending the data hundreds or thousands of kms, and the same distance back. One way will be reasonably fast, its just inputs, but the return leg will have all the graphical data attached. Lag higher than 35ms will be felt, and make games feel janky. That's roughly 30 fps.

      That's where the physics probably fails at the moment, and unless these games are hosted a lot closer than currently, something that wont change. Doesn't really matter how fast the internet connection gets, that distance is still an issue.

      That's the issue. The concept itself is fine, and I'm interested to see how they solve that. Pinging your nearest AWS cluster should show a fast enough round trip for it to work, but man that's some work. On the other hand, Microsoft does have all their Azure data clusters, so theres no reason they cant leverage off those and make it work.

        You're not taking two things into account though.
        1) No matter how big the pipe, if that last few metres (or any part) is small then that bandwidth is for nought.
        2) Video streaming uses buffering to cater for the inconsistencies of streaming data because it's just downloading a known sequence of packets. Games can potentially change every 1/60th of a second which means you can't buffer that data and need to send it as is.

          Sure I was. On point one, that last few meters is relatively insignificant. You can push data down a copper line at Gb/s if its short enough. It wouldn't be, but 100m would handle enough traffic for 4K streaming to be viable. FttP, FttC, FttB, and maybe DOCSIS 3.1 HFC should be fine. That's a big portion. Even FttN if you are close enough will be plenty.

          At the moment, FttN has a copper loop far greater than that, which is the issue for most, but even the longer lengths should be short enough to handle the speeds needed. ADSL2 ran at peak 24 Mbps when loops were under ~700m, which is getting out to the peak of what FttN might lump on people.

          In general, under NBN most people should have a short enough copper loop that the physical limitations of copper aren't hit. Fibre optics limit is up around 15kms, so not a problem. This is my sisters area of expertise. She's very good at it, and we talk about it often enough. A 24 Mbps connection translates to about 3mb/s data that can be streamed, and that's plenty.

          As for the buffering, spot on. Which is part of the latency side of things. It's also down to the initial connection, which is going to set certain limits, but that round trip is the big concern with this. 1080 streaming was being done pre-NBN though, so that bandwidth should still be enough.

          Contention will be a bigger issue when you're sharing that bandwidth with several other devices in the house, but you can handle that with data prioritising.

          But the data round trip to the servers wont change much. Most of that trip has been fibre optic for the past 30 plus years, so changing the stretch from the house to the exchange is a very small portion of the overall distance. Latency went from ~250ms to ~50ms for US East Coast servers for me, which still isn't fast enough - that's 20fps - so the servers will need to be closer.

          In other words, in Australia. Or whatever country people are located in. Which is less a problem than it used to be as AWS, Azure, Apple, etc all expanded around the globe. Handling that will be key to this, and at the moment I'm not sure if it can do it or not.

          That latency is the equivalent of buffering for similar reasons to your list, and was why I mentioned it. If it can do 35ms, it might work at 1080p, if it cant handle that, it wont handle it at any reso.

        Yeah, my concern is that they will try to use algorithms and AI to "predict" inputs and smooth out input lag.
        This whole streaming thing just seems to me like the answer to a question nobody asked. CPU's, GPU's, storage etc - all getting cheaper and cheaper. Why farm processing out at this point? Presumably, the answer is that they want to make more money for themselves and less for hardware component makers.

          Its a revolutionary idea though, and for that alone I want to see where they can take it.

          If this works, it could be akin to going from buying DVD's to subbing to Hulu or Netflix. End result is the same - you watch a movie - but the access is fundamentally different. I dont think its a big step to do similar with gaming, and as I see it the tech is so very close to it happening.

          I'm not sure its quite there yet, but they need to at least try so they can find the flaws and fix them. If they can.

          I dont think it damages hardware makers that much anyway. We'll still need most of the expensive components for it to work - the GPU will still be needed, as will the CPU and RAM. All you lose is storage, and thats cheap as chips. Most storage makers earn their big bucks in industry anyway, not gamers.

          What you lose is the upfront cost of the software, and we probably move to a sub service more and more. Which again would mimic Hulu and Netflix. Which didnt seem to break things too much.

          If it doesnt work, there are still things that can be learned by trying.

            I thought the point was to outsource all the computes and processing?
            I'm getting old so i hate new things and just want everyone off my lawn. I do like movie/tv/music streaming but for some reason game streaming really doesn't sit well with me. It feels like were be ushered down this crap path by people who don't really care much about the user experience and just want monthly subs on top of microtransactions and paywalls.

              In effect, this is really just a virtual desktop, its not a new concept. The difference this time is the amount of data needing to be sent back to you. Which shouldnt be any different to a video stream of the same resolution.

              The issue I see is how far you are from the servers. Too far, it really doesnt matter. You'll visually see the flaws when theres a 1/2 second gap between acting, and your avatar moving. But given how widespread Amazon, Microsoft, and Apple's servers are these days, its not something unsolvable.

    As always, game streaming is only ever going to work and revolutionise gaming like people imagine if game development companies suddenly have almost presidential power in every country and implement local data centers, roll out high speed internet to the entire country and provide ISPs with the infrastructure to manage and handle the demand.

    But hey, you guys all have high speed internet connections right?

    As someone who was a part of Google Stream, I was initially skeptical on it's connectivity. How could the internet read my inputs and manage my moving screen.

    Well it did it. I never saw any latency. I'd play ANY game on it.

Join the discussion!

Trending Stories Right Now