Join 3,495 readers in helping fund MetaFilter (Hide)


Wandering lonely as a cloud-gamer.
March 26, 2009 6:33 AM   Subscribe

Beta-registration has already started for Onlive, a revolutionary cloud-gaming service that promises to put an end to costly PC hardware upgrades, videogame piracy and the entire console industy and game retail sectors. There's just one small problem: it can't possibly work.
posted by permafrost (65 comments total) 10 users marked this as a favorite

 
See, if you have a technology which allows you to compress data efficiently enough to play a modern console remotely over the web, surely you'd not bother with games and simply spend the TRILLIONS OF FUCKING DOLLARS YOU HAVE MADE PATENTING YOUR TECHNOLOGY.


Hey guys, I've invented teleportation and I'm totally going to put Blockbuster out of business!
posted by fullerine at 6:43 AM on March 26, 2009 [20 favorites]


I guess they figure with all the fraud in the financial sector, it's high time to have another round of gaming/IT fraud?
posted by jkaczor at 6:45 AM on March 26, 2009


Obligatory
posted by DU at 6:47 AM on March 26, 2009


What they don't tell you is that the games are all Atari 2600 titles.
posted by swift at 6:47 AM on March 26, 2009 [7 favorites]


Imagine a world in which you can stream games directly onto your television or computer at 60 frames per second, with graphical settings at maximum. ... Essentially, OnLive's servers will be running the game, and sending a videostream through your Net connection. In turn, your controller/button inputs, are sent via your Net connection to OnLive's servers.

Actually, YouTube proves this will totally work...as long as the viewable area is no more than 4 pixels wide.
posted by DU at 6:50 AM on March 26, 2009


Would these dataservers have a greater environmental impact than running games off your own PC?
posted by doobiedoo at 6:57 AM on March 26, 2009


I'll believe that this has a ghost of a chance of working the day that Twitter manages to go 72 hours without getting "overloaded."
posted by Damn That Television at 7:05 AM on March 26, 2009 [2 favorites]


The "why this can't work" article nicely sums up my immediate reaction:

[I]n order to make OnLive perform exactly as claimed right now, the company has to have achieved the following:

1. OnLive has mastered video compression that outstrips the best that current technologies can achieve by a vast margin. In short, it has outsmarted the smartest compressionists in the world, and not only that, it's doing it in real-time.
2. OnLive's unparalleled grasp of psychophysics means that it has all but eliminated the concept of IP lag during its seven years of "stealth development", succeeding where the best minds in the business have only met with limited success.
3. OnLive has developed a range of affordable PC-compatible super-computers and hardware video encoders that are generations beyond anything on the market at the moment.


All this announcement needs is Duke Nukem Forever as the flagship title.

THAT SAID, I could totally see this working for games that aren't so video intensive, and which don't require the split-(sub)second responses demanded of modern action games. Stuff like Civilization should work particularly well under this scheme.
posted by mkultra at 7:09 AM on March 26, 2009


I wonder what's in the mind of those investors... As long as we have lag playing games on our machines how can they realistically expect not to have lag while transporting that amount of data?
posted by Lighioana at 7:09 AM on March 26, 2009


Stuff like Civilization should work particularly well under this scheme.

Yeah, but they're selling it as a replacement console, and the demos they've been doing feature all sorts of high-end games. Streaming Crysis to my TV with no lag sure seems like a pipe dream to me.
posted by graventy at 7:15 AM on March 26, 2009


Perlman has enough credibility that I'm willing to give OnLive some benefit of the doubt. But I don't understand how it could possibly work, either. Even if you can run a server farm for high end games cheaply, and even if you have magic video compression technology, you're still dealing with 80-200ms Internet latency that you simply cannot make go away. But, well, OnLive's put their stake in the ground and was demoing high quality racing games and saying "we can sell this". I look forward to seeing them try, and I hope it works.
posted by Nelson at 7:19 AM on March 26, 2009


The most important question is: Will it run on my Phantom?
posted by thanotopsis at 7:21 AM on March 26, 2009 [1 favorite]


Stuff like Civilization should work particularly well under this scheme.

That would greatly reduce the benefits to the user of the system though. If you can play a new game without a high-end graphics card and a fast CPU, then it's great. If you can only play games that would run on my machine anyway, why not just send me the game image and let me run it myself?
posted by burnmp3s at 7:22 AM on March 26, 2009


I was thinking that maybe they weren't really planning to have the "high-end games" (by which I guess you mean the high-end hardware requirement games, although I generally consider twitch games to be the LOW-end, gamewise). But thinking about it, that's all they could possibly mean. I mean, I can already play puzzle games, platformers and the like on my "obsolete" hardware. The only purpose this service would have is to keep me cutting edge without an upgrade treadmill and the only need for that is the "high-end games".

And in any case, it's self-defeating. No matter how you slice it, a game will run faster when the hardware is in my living room. So take whatever advances Onlive makes in compression or whatever and stick it in a box and sell it to me. Instant Onlive killer.
posted by DU at 7:26 AM on March 26, 2009


It's easy to make industry shaking announcements.

All you need is Scriptaculous
posted by srboisvert at 7:28 AM on March 26, 2009


We can already play high-speed games at home; why would we want to abstract hardware this way? We wouldn't be just playing against the game and/or other players, but also against Internet lag and provider malfunctions.

One benefit would be, in theory, avoiding the driver issues that plague PC gaming -- but that problem is handled pretty nicely by consoles.

This looks like the proverbial solution in search of a problem. Even if it actually DID work, and that doesn't look likely, would gamers really care?
posted by Malor at 7:30 AM on March 26, 2009


The most important question is: Will it run on my Phantom?

No, but it will run all the games released for the Phantom.
posted by cog_nate at 7:32 AM on March 26, 2009 [2 favorites]


We can already play high-speed games at home; why would we want to abstract hardware this way? We wouldn't be just playing against the game and/or other players, but also against Internet lag and provider malfunctions.

Yeah, but...it's in the cloud, man. The frackin' CLOUD!
posted by Thorzdad at 7:40 AM on March 26, 2009


This will work beautifully, and better than promised.

By 2015
posted by Burhanistan at 7:41 AM on March 26, 2009


Last year Teradici was showcasing this technology for the high end VDI market. In fact as a proof of concept, they were running Crysis at much higher resolutions than 720p. Last I heard VMWare and Teradici were trying to push a software implementation of their transportation protocol. Right now it requires a discrete card for each deployment, which just moves the machine from users desks to the data center, you don't see the consolidation and cost reduction you would typically see from a VDI deployment. Not that this isn't great within itself, but it only helps if you're users are sweating because they have 3-4 workstations in their office or cubicle.

I'm getting off topic, but the point being is that I've seen this technology work, I've played with this technology, but I'd be surprised if most users will have the low latency that this requires. Really, anything over 15ms and you begin to notice it and when it hits the WAN, while it works 80% of the time, the other 20% were annoying enough to render it unusable (again the correct deployment is simply to get a private line between offices, which in my case was cheaper than running a data center in the branch office).

What's most surprising is that I've never heard of them and I sort of have my pulse of the virtualization community. Does anyone know if they're a spin-off of Citrix/VMWare? There's only a few big players in this market, and everyone is throwing money at them. I'd be very surprised if this was a completely independent company that developed this technology all by themself.
posted by geoff. at 7:44 AM on March 26, 2009


This looks like the proverbial solution in search of a problem. Even if it actually DID work, and that doesn't look likely, would gamers really care?

Uh, the millions of gamers and non-gamers that don't bother with games because they are a clusterfuck of video card GPU specs, console wars, and constant PC upgrades.

I watched the OnLive demo and was blown away because it's perfect and what I always wanted. No need to pick a console. No need to pay $60 for a game I might play for ten minutes and hate. No need to have an overclocked PC covered in fans howling like a leaf blower just to get decent graphics.

I tried having a high end gaming PC sometime around 2001 and it was a constant pain the ass and made my hardware feel like a treadmill. I was rebuilding the thing every four months because better games came out that I couldn't get enough of a good framerate on.

I have a PS3, and often wished I had an xbox360 (mostly to play against friends). I've missed out on loads of major games because they never came to my console or they were heavily delayed and I lost interest by the time they finally came out.

I would love it if OnLive worked, but it seems too far-fetched. Perhaps they can price it too high, so that they're not trying to support 1 million simultaneous players.
posted by mathowie at 7:45 AM on March 26, 2009 [4 favorites]


We can already play high-speed games at home; why would we want to abstract hardware this way?

To reduce clutter and electricity usage in the home. I've got 3 consoles hooked up to my telly at the moment, with another couple lurking around in storage. If there was a unified gaming platform (constantly getting upgraded without my input) that could run every game around then I'd be happy to use it.
posted by MUD at 7:49 AM on March 26, 2009


Maybe we should tell OnLive that it won't work....
posted by Smaaz at 7:49 AM on March 26, 2009


Perhaps they can price it too high, so that they're not trying to support 1 million simultaneous players.

One of the links above (or maybe one I didn't post) mentions that Onlive have been cagey about pricing. Assuming they make it to launch, I'm anticipating some sort of pricing tier system. If you want to play Crysis 2 on Half Life Ep 3 on launch day, you'll end up paying a fairly large percentage of the retail price. Access to older titles will be correspondingly cheaper, since there'll be less demand and they won't need as much bandwidth/processing to stay playable.
posted by permafrost at 7:57 AM on March 26, 2009


...a revolutionary cloud-gaming service that promises to put an end to costly PC hardware upgrades, videogame piracy and the entire console industy and game retail sectors.

Heh. Very droll.
posted by KokuRyu at 8:08 AM on March 26, 2009


If the game freezes up, who is going to eject the cartridge, blow on it, and then put it back in and press reset? Cloud people?
posted by tawny at 8:08 AM on March 26, 2009 [20 favorites]


It's not that far-fetched -- the round-trip latency to Akamai from my house is 10 ms.

As for the hardware cost, you presumably have more total users than peak users, so it should at least be cheaper than having each user buy their own console.
posted by you at 8:22 AM on March 26, 2009 [1 favorite]


No need to pay $60 for a game I might play for ten minutes and hate.

The current consoles (well, PS3 and XBox, at least) have demos available for most games, which should solve that problem.

That said, you'll still probably be paying a monthly fee AND 60 or so bucks for your games. The monthly fee for access to the service itself (XBox Live-style) makes sense.

The 60 bucks because game companies are still at the mercy of retailers, and can't afford to undercut them and lose meatspace sales. I'm sure they'll be able to have Steam-like sales on old games, but new stuff is bound to be regular over-priced.

They're probably also assuming a downloadable game style of DRM, which might not be acceptable with the recent announcements of Steam DRM changes and Goo allowing resales.
posted by graventy at 8:25 AM on March 26, 2009


OnLive boss Steve Perlmen remains adamant: "Perceptually, it appears the game is playing locally... what we have is something that is absolutely incredible. You should be sceptical. My first thinking was this shouldn't work, but it does."

Can anyone think of a situation where that "I know it's unbelievable, I wouldn't believe it if I were you either" line has actually panned out with something that was actually unbelievable-but-true? I'm having trouble coming up with specific examples but the phrasing fills me with this pervasive sense of a series of predictable disappointments. The world seems to be depressingly believable, aside from those occasional situations where it turns out to be unbelievably more horrible than I thought.
posted by nanojath at 9:30 AM on March 26, 2009 [2 favorites]


"Uh, the millions of gamers and non-gamers that don't bother with games because they are a clusterfuck of video card GPU specs, console wars, and constant PC upgrades."

Those people buy consoles instead.

Onlive may run tolerably well in ideal conditions, but getting a low latency, high speed net connection is often more challenging than buying decent hardware, and who cares how high the detail setting is if you're having to put up with compression artefacts? It seems like one of those businesses that's relying on having enough gullible investors and consumers.
posted by malevolent at 9:40 AM on March 26, 2009


Can anyone think of a situation where that "I know it's unbelievable, I wouldn't believe it if I were you either" line has actually panned out with something that was actually unbelievable-but-true?

Quantum mechanics.
posted by The Bellman at 9:41 AM on March 26, 2009 [1 favorite]


Or the transistor.
posted by Mitheral at 10:01 AM on March 26, 2009


Can anyone think of a situation where that "I know it's unbelievable, I wouldn't believe it if I were you either" line has actually panned out with something that was actually unbelievable-but-true?

Turbochargers seem like perpetual motion machines -- sooo you're going to force air into the intake using a pump driven by air coming out the exhaust. And this will make more power. Rrrrrright... (Same deal wrt belt-driven superchargers.)

Tons of things with social dynamics seem counter-intuitive and/or damn near like mind-reading or voodoo thought projection -- friendship & leadership, attraction & dating, conflict resolution, body-language and other sub-communication, etc.

This seems like a great AskMe that would definitely get deleted.
posted by LordSludge at 10:22 AM on March 26, 2009


I was going to say relativity. But as mentioned above, if Onlive has made a discovery as fundamental as relativity, QM or the transistor, why are they running a game company?
posted by DU at 10:22 AM on March 26, 2009 [1 favorite]


The telephone.
posted by ook at 10:28 AM on March 26, 2009


Obligatory

That's the link for the current comic; here's the permanent one. And here's Tycho's two cents.
posted by Chrysostom at 10:30 AM on March 26, 2009


Steorn used the same marketing strategy ("it's crazy, but it just might work!") for their perpetual motion machines. Apparently, they're still at it, despite every public demonstration of their technology mysteriously failing to work.
posted by teraflop at 10:35 AM on March 26, 2009


I don't know how relevant this is, but about 7 years ago while I was at Georgia Tech, I used ssh with an X tunnel to play UnrealTournament, running on a GNU/Linux PC in my dorm, in the computer lab on a Sun workstation. It wasn't great, but it was playable.
posted by nzero at 10:35 AM on March 26, 2009


What they don't tell you is that the games are all Atari 2600 titles.

What they don't show you is the 800-node compute cluster running Pong.
posted by Blazecock Pileon at 10:38 AM on March 26, 2009



Can anyone think of a situation where that "I know it's unbelievable, I wouldn't believe it if I were you either" line has actually panned out with something that was actually unbelievable-but-true?

Quantum mechanics.


And the great thing about quantum mechanics is it's true and not true all at the same time!
posted by fusinski at 10:47 AM on March 26, 2009 [2 favorites]


Wow, I don't know whether this is ready for gaming, but it seems to have great potential for GIS and CAD software. Rather than needing an entire power workstation to use such software, a company or institution could buy a high resolution display, basic box, and reasonably fast connection, pay a subscription fee, and not have to worry about upgrades for a long time.

Someone upthread asked about the comparative environmental impacts (actually, I think that might have been a snark). I'd rather centralize the massive physical material and energy investment required for that much computing power. This provides opportunity for efficiency in energy use and for applications such as combined heat and power to take advantage of the inevitable massive thermal waste. The centralization of physical resources makes it easier to upgrade with minimum waste and makes closed loop manufacturing easy once manufacturers start reclaiming the basic materials required for computer hardware. Since the company would be selling a service (gaming) rather than a good (gaming hardware and software), they would be responsible for all material and energy waste rather than the customer. Of course, since this service significantly expands the pool of gamers by making gaming cheaper, overall energy consumption might increase despite greater efficiency and less waste of materials.
posted by Derive the Hamiltonian of... at 10:51 AM on March 26, 2009


Wow, I don't know whether this is ready for gaming, but it seems to have great potential for GIS and CAD software. Rather than needing an entire power workstation to use such software, a company or institution could buy a high resolution display, basic box, and reasonably fast connection, pay a subscription fee, and not have to worry about upgrades for a long time.

And we could call this brand-new technology a Mainframe Computer! And create a brand-new programming language for it called FORTRAN! This is truly the wave of the future!

On a serious note though, the general strategy employed by OnLive, to put everything on the server side and use an incredibly fast and wide connection to throw it to the client is usually a dumb idea. A basic box these days does a lot more than high-end machines could ten years ago, so using them to sit there and do nothing while a server chugs away at doing all of the work is usually a waste of resources that produces a substandard result. It's a tough balancing act between client and server, and the evolution from basic HTML only to JavaScript/ActiveX/Java/Flash to Ajax/Silverlight/Air has been a gradual progression away from doing all of the work on the server side for web applications.

PC Gaming could probably benefit by handing off more of the work to servers, and MMOs along with distribution systems like Steam or GameTap have been making progress in that direction, but OnLive's plan is at best overestimating the benefits of going to a completely server-based solution.
posted by burnmp3s at 11:18 AM on March 26, 2009


this is my reaction to the "why it can't possibly work" article as I read it:

-It's wrong in its very first claim.

To give the kind of performance OnLive is promising (720p at 60 frames-per-second) realistically its datacenters are going to require the processing equivalent of a high-end dual core PC running a very fast GPU - a 9800GT minimum, and maybe something a bit meatier depending on whether the 60fps gameplay claim works out, and which games will actually be running. That's for every single connection OnLive is going to be handling.

what it needs is a barebones rackmount mobo with a hefty gpu and onboard sound and whatever their network or fibre channel networking is. it needs one of these for every simultaneous connection that runs a graphics heavy game in the location each user is a part of. they say in their interview that if you're talking about older games or peggle or something like that, they have smaller less heavy duty units to keep from monopolizing the heavy hitting hardware. but the key point is: they only need to support the users who are online at any given moment in the location of each data center they run. I forget how many they're planning for the (north american exclusive) initial release, but I know it's at least 2 data centers (1 for each coast) if not significantly more. I don't know if they'll successfully do it, but I imagine that they spent 7 years figuring out a way to do it. I hope it works. I don't know if it will, but I'd say that it's at least possible.

moving on:

It sounds brilliant, but there's one rather annoying fact to consider: the nature of video compression is such that the longer the CPU has to encode the video, the better the job it will do. Conversely, it's a matter of fact that the lower the latency, the less efficient it can be.

this is armchair tech swagger. I can sympathize will his desire to call bullshit, I really can, because OnLive basically walked on stage saying "we have this magic box. you put a game in the magic box and it instantly comes out the other side as video delivered to your home so that you can play the game remotely in real time." the process behind it is completely obscured and no one can know how the hell they worked it out. they claim they did it, created this revolutionary video streaming service that does what was formerly impossible, but who the hell knows how. but let's be clear: they claim to have created what was formerly impossible. they didn't come out and say "using only current gen video compression software and hardware, we did blahblahblah..." they came out and said "we spent 7 years working on the video end of this. what we've done is more than revolutionary. we've reinvented online streaming with our crazy magic box." now, they could be lying. that is absolutely possible. But to say "they must be lying because that's not how modern current gen video works" is like saying "man can't fly because flying involves wings and we don't have wings." well, we didn't till we built some, but hey look we worked it out. now we fly every day. if OnLive are lying, then yes obviously it can't work. but until we find out what's inside the magic box, none of us can know if they really worked it out.

I can only see one way to make this work and guarantee the necessary quality of service, and that's to adopt an IPTV-style model. The OnLive datacenters will be licensed to ISPs, who will have them at their base of operations. Latency will be massively reduced, the connection will be far more stable, plus the datacenters with the PCs and hardware encoders can be distributed worldwide in a more effective manner. ISPs will be cut into the deal the way that retailers are now with conventional game-purchasing.

But even in this scenario, practically, I still can't see it happening. Microsoft's IPTV venture still hasn't materialised anywhere outside of the USA, so what chance does OnLive have of brokering a deal?


well, onlive went up on stage and said "we've worked out deals with a lot of the nation's ISPs that'll keep everybody happy." so dude here is saying "no way. if Microsoft couldn't do it, these guys have no chance." okay, dude. they must be lying, then, I guess.

in the end, I think this guy could very well be right. It is entirely possible that OnLive have just blown smoke up everybody's asses so well that we're all living in a fantasy world. It seems a logical leap to my mind to then say "well, that means it's impossible." It sounds to me that this guy both wants and doesn't want it to be true. He wants it to be true, the way any gamer would want this to be true. But he doesn't the way any competitor wouldn't. As in "why didn't I think of it first?! well, because it's clearly not possible!"

and maybe it isn't. but the point is that we don't know just yet.
posted by shmegegge at 11:18 AM on March 26, 2009


I love how the EuroGamer article consisted of thinking of stupid ways to implement things, and then proving they wouldn't work.

Most of the issues involved here are easily solvable with custom codecs implemented in hardware at each end of the connection and multiple server-farms full of cheap hacked console motherboards.

The only real question is whether the end-to-end latency (maybe 200 milliseconds) will be too high to feel good.
I'm going to give the OnLive guys the benefit of the doubt at this point. I can see that they have had smart people designing for low latency since day one, and I think they have a shot.
posted by w0mbat at 11:30 AM on March 26, 2009


now, regarding onlive:

what strikes me as odd that no one's mentioned is the following:

we're talking about a stream that requires 1.5 mbps just to run SD games, and 5 mbps to run HD. Most people REALLY don't have 5mbps, but those people do have xboxes that run in hd.

what we're ultimately talking about here are people who have an HD experience in their home, being forced to either upgrade their internet (if it's even possible where they live!) to a 5mbps connection at substantial cost, or accepting a reversion to SD gaming.

part of what onlive said is that they've worked out deals with the ISPS to ensure they get the bandwidth they need over the last mile of content delivery (i.e. the connection to your home cable or fiber modem.) to ensure these incredible speeds. they hinted that they endeared themselves to the ISPs because they'll inspire more isp customers to get 5mbps connections in order to enjoy the HD experience.

but how much more would a person pay for that experience when they've already got a box that does hd for a one time fee, instead of a costly subscription? even if they DO have to buy another one for another one time fee 4-6 years down the line? hell, you can play grand theft auto iv without any internet connection at all on an xbox that costs under $300.

they're implying that people will happily make what amounts to a substantial supplemental purchase just to enjoy this service the same way they enjoy their xboxes and ps3s. I can't help but wonder if that's true. I'm thrilled as hell at the prospect of Onlive, but I know I can't upgrade to a 5mbps connection. I actually recently had to shut off all but the most basic cable package that I needed in order to get cable internet in my house because my bill was too big. that 25 or 35 bucks a month is a big deal these days. much as I want to play crisis (my pc is a piece of shit), do I want to play it in 480i? maybe. I'm not sure. Do I want to pay out the ass to play it at 720p? almost certainly not. but that's me.
posted by shmegegge at 11:31 AM on March 26, 2009


and that's on top of the monthly subscription model for their service!
posted by shmegegge at 11:33 AM on March 26, 2009


I don't think the video-compression issue is really the killer; the deal-breaker is latency.

Sending a raster image down the wire is obviously ridiculous with current technology, but I'm not convinced that transmitting OpenGL instructions or some other form of vector data -- which is then rendered by video hardware on the receiving end -- is impractical. Somewhere not long ago I was reading an article about NVidia working on a technology like that for graphics-rich thin clients. There are still significant technical hurdles (loading all the texture materials would be problematic), but I don't think it's completely in the realm of science fiction.

The latency issue ... that seems like a greater problem. Unless they're planning on deploying their servers all over the Internet, so that no user will ever be further than a couple of network hops from one, the whole idea of sending user input/button-presses up the wire to the server, then sending the results back down in the form of display output, is not going to work very well. Even moderate latency, like 100ms or so, can be very noticeable when it occurs on every single keypress.

Increasing endpoint bandwidth is expensive, but it's not really a Hard Problem. Decreasing latency, on the other hand, is very difficult and depends on many parts of the network that aren't under your direct control. You might be able to pay for some sort of packet prioritization, but if your packets need to go somewhere and the only link from A to B is congested, they're going to wait in line -- and that means latency. (Plus even if you somehow gave your packets ultimate priority, you would still have a 'latency floor' dependent on then number of routers it passes through; it takes time to perform the routing itself, even if the link isn't congested. Good luck fixing that.)
posted by Kadin2048 at 11:36 AM on March 26, 2009 [2 favorites]


burnmp3s; You seem to know more about this than I, but anecdotally I've seen many, many university computer labs and business setups that consist of a room or office full of twenty or thirty individual high powered workstations. Sure, my data is stored on the network, but the box next to me is doing all the heavy lifting when I render, zoom in, whatever. Maybe all really big organizations already put everything on a central computer, but there is certainly a lot of market space at the university computer lab and small office level for something like this.

Regarding today's comparatively high-powered basic boxes; they're designed to stand alone. I imagine that if this technology takes off, you would start to see really inexpensive workstations designed specifically for this application, freeing up money for, say, really nice displays and maybe some improvements around the office.
posted by Derive the Hamiltonian of... at 11:38 AM on March 26, 2009


I think they can make it work, especially if they downgrade some of the specs (30fps or 45fps instead of 60, for example). What I don't think they can do is make it economical.

Most of the time, they will only need enough servers to support some average number of users. But when major releases come out that everyone wants to play, they're going to need a lot more servers. But those servers will be idle the rest of the time. Then, when the next major game comes along, they're going to need to buy a whole new set of servers to handle the improved graphics, physics, AI, etc.

It's the upgrade treadmill all over again. If they never needed a 1:1 ratio of subscribers to servers then they could offer a cost savings compared to everyone owning their own machine. But new AAA releases are going to require them to have very nearly a 1:1 ratio.

The easiest way out of this is tiered pricing: if you want to play on launch day you have to pay extra. But then the hardcore gamers who want to play at launch would probably find it cheaper to maintain their own machines...which is what they do already anyway!

As a practical matter, of course, even 720p is on the lower end for modern PC games, and standard def is almost laughable. I can see this being a way to bring PC games to people who prefer consoles, though. If they stop targeting hardcore gamers and refocus on casual gamers who don't necessarily need, want, or care about having the best graphics and latest high-end titles, then I can see it being economical. I strongly doubt it can be economical and still compete with PC ownership among hardcore gamers, though.
posted by jedicus at 11:44 AM on March 26, 2009


IMPORTANT IF TRUE!

This simply doesn't pass the sniff-test for me. Maybe they've spent the last seven years designing video compression that outstrips any other attempts at it by a country mile. I don't know, it seems unlikely, but it's possible. What I'm pretty sure they haven't done is get Comcast to finally provide me with service that remains consistent from one moment to the next, allowing me to play a racing game remotely. The lag issues on the user-side are simply too massive and completely out of OnLive's control to be made up for by any technology they come up with server-side.

I'd love love love to eat my words on this. This is the type of thing I've waited my whole life for. It looks to me, however, like they've been working for seven years without a product, the economy went to shit since the last GDC, and so now they've got to present something in order to not lose their investors or even get nailed for fraud, they know that what they've got won't work, but they'll slap a "Beta" on it indefinitely while looking around for new investors.

But again, I hope I'm wrong.
posted by Navelgazer at 12:01 PM on March 26, 2009


You seem to know more about this than I, but anecdotally I've seen many, many university computer labs and business setups that consist of a room or office full of twenty or thirty individual high powered workstations. Sure, my data is stored on the network, but the box next to me is doing all the heavy lifting when I render, zoom in, whatever.

My point was that your solution to this, putting all of the logic on a central computer and having dumbed-down PCs that only act as an interface and a display device, already existed for much of the early history of computing. The big central computers were called mainframes and the input and display devices were called terminals. In fact, today's technology directly evolved from that setup.

It's not that IT experts are unaware of this option or that we are too attached to the current system, it's that the current system is better than that system was. Terminals were replaced by PCs because it's not actually much cheaper to make a dumb input/display device rather than a full-blown PC. These days a $40 router can run Linux and do a lot more processing than those dumb terminals ever did. And if you use a laptop for example you can work from home, bring your computer on a business trip or any of the other things that get much more difficult with a system that puts all of the computing power on the server.

In the current IT world, most applications are balanced between the client and the server, using each side to do what it's best at, and that seems to be the best solution anyone's thought of at this point.
posted by burnmp3s at 12:05 PM on March 26, 2009


c'mon people, you know this will work. you just have to BELIEVE.
posted by sexyrobot at 12:10 PM on March 26, 2009


one last thing about that eurogamer article, and then I'll stop harping on it I swear:

he keeps talking about how unlikely it is that in 7 years onlive would have outpaced "the best minds in the business." just to be clear, go ahead and look Steve Perlman up, to realize that he IS one of the best minds in the business. besides having been instrumental in developing quicktime and (more questionably) webtv, his revolutionary tech just won an oscar.

it's frustrating to hear someone (even though he writes for eurogamer, which I love) whose experience in the fields he's talking about exists entirely in the span of time these guys have been working on this one project start talking about how unlikely it is that one of the actual innovators in the field would, you know... innovate.
posted by shmegegge at 12:11 PM on March 26, 2009


Whatever, I've been running Crysis at Quad-HD resolutions on my wrist-watch since the late 1980s.

I'm still on the loading screen.
posted by blue_beetle at 12:25 PM on March 26, 2009 [1 favorite]


Honestly, I think the biggest problem OnLive will have is attracting a customer base that is willing to simply have all their games available only online. There are still too many people without internet connections, collectors who need to have the physical game disc/(whatever), etc. etc. for this to become the glorious one unified solution that will end all console wars forever and ever as OnLive and some bloggers seem to think.
posted by bookwo3107 at 12:45 PM on March 26, 2009


burnmp3s: Thank you.

I'm honestly surprised that, for specific high-powered processing operations, running a whole room full of obsolete-in-two-years $5k a pop beast workstations drawing several hundred watts each at full load and putting kilowatts of additional load on the HVAC system is less expensive than setting up a room full of inexpensive, low power terminals that pay a subscription fee to connect to a central processing bank that can take advantage of all the perks of being centralized (efficiency projects, rolling upgrades, etc...) and would have constant demand for the resources it sells. Businesses would buy most of the system during the 9 - 5 hours and research labs running simulations an similar clients would pay for the rest of the day. Or gamers. Whoever. The waste heat from a monumental processing bank like that could probably be used to heat surrounding infrastructure or drive some serious industrial processes.

But, this is just a cool thought experiment for me. I don't pretend to have the technical knowledge required to make the comparison I'm suggesting.
posted by Derive the Hamiltonian of... at 12:51 PM on March 26, 2009


As far as the difference in computational power needed at peak versus average, there must be services like Amazon S3 but for processing, right? The tiered pricing model works here, because you just buy more cpu time when demand goes up.

720p is plenty... unless you have the latest hardware you're not going to be running any higher than that for the newest games anyway.

I would love this service, if it works. I haven't played PC games since college, and fled to the consoles because as I've gotten older I have more money than time. I just want to pop the disc in and have it work, no dicking around. But I dearly miss the whole genres of games that don't get ported to the consoles.

I don't have 5Mbit down, but I would get it for Onlive. It's not like buying a $300 drum set for rock band; I also get delicious internets!
posted by danny the boy at 2:39 PM on March 26, 2009


Actually, Derive, the heat problem is significant. Datacenter density is essentially heat-limited now. A bunch of GPUs and high end CPUs running full-tilt is gonna play hell with the cooling.
posted by Nelson at 2:41 PM on March 26, 2009


720p is plenty... unless you have the latest hardware

What you mean like a sub-300$ xbox? There are plenty of 360 games that run native at 1080p. So the point remains how is the onlive service going to deliver that and make it worth my while?

They'd have to practically defy the laws of physics at rock bottom prices.
posted by lumpenprole at 3:46 PM on March 26, 2009


What you mean like a sub-300$ xbox? There are plenty of 360 games that run native at 1080p

Which ones? Most xbox360 games run natively at 720p, and are then scaled up to 1080p. Remember the drama around how Halo 3 only ran at 1152×640?

But more to the point, even at a lower resolution, console graphics are tuned down from their PC counterparts in order to get a decent framerate. Unless you think the graphics capability of the xbox 360 is somehow on par with an up-to-date gaming PC?

All I am saying is, there is a business case for the service, even if it doesn't happen to appeal to you. I want to play PC games that don't exist on the console. They are promising that ability, with graphics that are comparable or better than the console experience. That by itself is worth it to me, but I love the idea that at some point in the future games could become truly platform agnostic, not just format and media agnostic.

I am talking about playing Crysis 4 on a $150 netbook that has no hard drive, at the airport. Or whatever.
posted by danny the boy at 4:21 PM on March 26, 2009


Knowing Perlman you can bet that the technology involved is very clever. He is not a bullshit artist and he works with the best.
posted by w0mbat at 5:03 PM on March 26, 2009


Which ones?

Well, these ones admittedly, they are mostly crap but still.

(Ratatouille, the game at hi-res. Really?)
posted by lumpenprole at 5:25 PM on March 26, 2009


Oh, and "playing Crysis 4 on a $150 netbook that has no hard drive, at the airport", keep dreaming pal.

The day the airport streams that to you is the day I eat my own hat. I hope to be pleasantly surprised by onlive, but there's no way you'll be doing that at a public wireless node anytime soon.
posted by lumpenprole at 5:26 PM on March 26, 2009


I'm generally of the opinion that this won't work. I would be extremely skeptical of someone having a positive gaming experience running Crysis at high-def across a 100 mbps lan, based on everything I've experienced running virtual consoles and the like. When you claim to be able to do it across the internet, that's not a bold claim - that's an indication that it's time to call the nice people in white coats to take you on a nice vacation. Nothing I've ever seen tells me that this is even remotely possible, to say nothing of desirable.

Or, to put it another way: I have 70 ms. of ping to a good internet server. That's not very much, but it's still perceptible. If my controls suddenly had that kind of lag - particularly if my mouse did - it could easily make the game a lot less enjoyable. Now add the overhead for encoding and decoding video, encoding and decoding your controls (I imagine the mouse will be especially fun to control remotely), and all that kind of stuff. Good luck making it playable after all that. Plus, some people will inevitably have bad pings, internet slowdowns, crappy wireless, etc. Now all of that can ruin single player gaming as well as multiplayer - and do so even more effectively!

Also, another thing; you can buffer streaming video but not if it's realtime! Every latency spike, every network hiccup, ever tiny moment of lag is going to be not only noticeable but incredibly disruptive. Even if you could stream this kind of thing over the internet, which you can't, you'd be talking about doing it real-time with no buffering at all. I've never seen an internet connection in my entire life that could do that.

This plan is just ridiculous, and it will not work. Not at all, not even a little.
posted by Mitrovarr at 6:54 PM on March 26, 2009


Shamwow.
posted by Frasermoo at 9:21 PM on March 26, 2009


« Older "How do you talk about something like gangsta rap ...  |  Wikirank... Newer »


This thread has been archived and is closed to new comments