OnLive Turned On

OnLive launches this week, and company CEO Steve Perlman explains how the startup plans to ramp up the cloud-gaming service following launch, the business model behind the tech, and the ever-present issue of latency.

For Steve Perlman’s OnLive, the moment of truth arrives this week. On June 17, just after the conclusion of the E3 convention in Los Angeles, the long-awaited cloud-based gaming service will finally go live for the initial influx of OnLive subscribers.

Originally announced at Game Developers Conference 2009, some have heralded the service as an industry game-changer -- gaming on a cloud means that data is bounced back and forth between a remote server and a user’s machine. High-powered data centers handle the brunt of a PC or Mac game's system requirements, and all a gamer needs to play high-end games is a good internet connection and a screen. It can theoretically reduce the hardware barrier, at the same time providing game makers a direct link to consumers.

But there are still many who are skeptical if OnLive can deliver -- that it can change the way we receive and play games. The company has demoed the service and amassed support from prominent game makers, but onlookers want to see OnLive in the wild to see if it really works.

In the ramp-up to launch, we caught up with OnLive CEO Perlman, who doesn’t promise to change gaming as we know it, but rather expresses a somewhat cautious approach to a brand new way to deliver games, in terms of tech and business. And he’s tired of talking about lag (but he still talks about it at length anyway).

So,the launch date is June 17, and that's official.

Steve Perlman: Yes. It'll be open, I think, about an hour after E3 closes.

What kind of marketing push is OnLive putting behind the service's launch? And how are you guys getting the word out beyond the core gamer?

SP: We're starting out everything with a ramp. One thing we've learned with beta is this is different from anything else that's ever been done before.

We just got a lot of questions in the beginning. Some are really, really basic. Some people are conceptual. You know, "What's going on?" It's not something people have seen before.

Plus, we just added a boatload of new games. Any time you've got many thousands of anything hooked up together, you want to take it a little bit slow.

We had this pre-registration program, and we invited 25,000 people that subscribed. So, we finally cut off and said, "That's it." What we're going to do is let the first-come-first-serve in the pre-registration. Let's get out and get onto the service.

We're going to gradually roll that out over a series of days. The people that sign up starting on June 15th, when is when this AT&T program is set up to start, then [after] we'll begin again making it through that list of people.

So, we will be running it well below maximum capacity on June 17th when we open up the doors. It's just sort of a necessary step. We want to see if there's any systematic bugs or errors or something crept in, maybe if it was something we did, or whatever. We'd rather do that with 10,000 rather than with a 100,000 people. Then, we need to go in with ramping it up.

This is like introducing a new kind of very large jetliner in a way, and we expect the takeoff to initially have some turbulence -- you know what I mean. Just some bumpy stuff. Once we've got to cruising altitude and it's smooth -- then, at that point, we will, for example, start a larger marketing campaign with the service. As soon as we feel like things are really solid, then we'll roll out the MicroConsole [which will allow OnLive to play on televisions].

How far into the future do you expect to do this ramp-up before you get to cruising altitude?

SP: Well, I don't know. The very important qualification is it's going to depend on how things go. The other thing is what demand there is. We ended up having more people in the pre-registration [than expected]. So, we'll see.

How many people were in the beta?

SP: Well, we had hundreds of thousands of people that signed up, but not every single person got invited because we had [for example] 10 people who were on the same ISP, the exact same area, the same block, it doesn't do us much good to have multiple ones testing, you know what I mean. So, it was pretty spread across the country.

We cycled through them. For example, you have a new group of people who are updated now. I saw an email kind of fly through, "By the way, it's time for your beta, now that it's a week before launch." [laughs]

The particular configurations were something we had to bug everyone to make sure that everyone we had and were able to verify. So, it's kind of like that. The people who were nice enough to go and sign up, we have a large pool to draw from. That's what we've done.

I don't know how many thousands we've gotten in. I have to actually go through our records to figure it out. The pool is hundreds of thousands that we drew from

Different people tend to do different things at the same time. We actually have had several different, we'll call them "farms", server farms, set up. We're testing MicroConsoles, and we're testing a version of MicroConsoles.

So, we're testing Macintosh, different versions of the algorithm. The compression algorithm that we use, integrates both what in traditional terms we call "error correction" but more appropriately is "error concealment" because when you have a very, very good network connection going into a university or a business, it has packet loss for all the packets that get dropped. And certainly consumer connections have a lot.

So, most of what we've been doing over the last few years and most of the stuff that has been going on in beta has been going just through tens of thousands of homes. I don't know what the number is, but I guess at this point thousands of different models of computers going through all the different compatibility scenarios. I guess it's hundreds of ISPs. I'm pulling these numbers out of my head, a vision of where are numbers are now. Very large number of ISPs.

There are a million things that could go wrong.

SP: Yes. So, when we go live, we now have over a hundred algorithms that are actually deployed depending on the particular scenario you have. We keep adding them. We added a couple more last week, for example. It worked fine for those people; it just worked better with these tweaks. And we'll continue to evolve it.

In fact, when you're connected, for example, as soon as you start playing at 3 PM and 5 to 6 PM comes round, and a lot of people get onto your cable connection, you'll get a different algorithm for us. We swap the compression algorithm, so you probably won't notice it, but it actually is streaming differently because of course it needs to overcome different obstacles.

You guys had a recent announcement with BT.

SP: Yes.

Now you have a new partnership with AT&T. Are you currently in talks with companies like AT&T, Comcast, Verizon about bundling OnLive with their services [to get OnLive in more homes]?

SP: In the U.S., we have all different kinds of discussions with U.S. operators as well. I think you can take the fact that AT&T decided to partner with us in launch and offering this promotion, and the promotion works across with any connection, needless to say... What they're demonstrating so far is in addition to being an investor, they are also a strategic partner.

That's part of why they wanted to be with us as we roll out from the very beginning. Their brand is associated with OnLive on the consumer side.

And Verizon and Comcast... Bigger providers might be taking a wait-and-see approach and see how it goes?

SP: Well, we haven't announced anything with those providers. What I can say is the following, that in our data centers, we have about a dozen different connections coming in from different providers. They're called T1 providers. The reason we do this... I'm giving you a little bit of the ingredients.

When we connect to your home, if we just go through the internet, we might get a route that is very, very roundabout, right. You know, it might not be direct from our data center to your home. So, what we do is we actually go and connect you through a number of different providers in order to find a route that is minimal latency, minimal congestion, and best throughput.

So, what we've ended up doing is partnering with I think most -- I guess at this point, yeah -- most of the major ISPs in the United States. So, we actually have relationships with all these ISPs in addition to AT&T. AT&T is one of them. We're able to go and make direct routes to people's homes.

So, again, it's not a consumer announcement. If we were to make an announcement about that, it'd be more like our Dell announcement, you know what I mean. It's a business announcement because it's on the data center side. But the impact it has on the consumer is that we're able to find better routes and get into your home.

And frankly, we can't get to every ISP. There's a lot of little ISPs out there, you know. But we can certainly get into, at this point, the majority of broadband connections in the U.S. with a direct route.

I see that you guys are partnering with Dell. The quality of the service kind of relies on how close the data center is, right?

SP: The biggest issue is actually not how close the data center is. The biggest issue is what sort of latency you are having in what we call the last mile, the connection between your ISP's either central office, in the case of a telephone operator, or the head end in the case of a cable operator.

We have a data center in Silicon Valley in Santa Clara. At this point, we've overflowed the data center, so we have multiple data centers. But anyway, so we have this cluster of data centers -- that's a better way to describe it -- in the Bay Area. So, suppose you're in, I don't know, in Sunnyvale, which is extremely close to these data centers. You might have a 20 millisecond lag if you have sort of a worst case ISP in that last mile.

And somebody else, who is, for example, in Las Vegas is hundreds of miles away from a data center. If they have an ISP that only has a five or ten millisecond delay... The delay in the internet is probably less than 10 milliseconds or less from Las Vegas to Silicon Valley. So, that person would actually have a lower latency connection than somebody who is closer and has a higher latency last mile.

So, a lot of people don't realize that. A thousand miles on the internet, using the techniques that OnLive is using to find an optimal route... I mean, obviously, that's as the crows fly a thousand miles, right. If you're routed a roundabout way, then you much further than a thousand miles. But a thousand miles on the internet is about 21 milliseconds, okay. The worse case we see pretty much is around 25 milliseconds.

Are you getting tired yet about people asking you about lag and OnLive? [laughs]

SP: I am. So, the other thing is we're going to be putting out facts. There's a lot of misconceptions about latency. It's a brand new technology, and it's fair for people to ask. We're going to need to understand this, right.

For example, we've measured monitors that have 80 milliseconds of latency. The monitor. [laughs] In fact, we spoke to that monitor manufacturer, who will remain nameless [laughs], and they understood what we were doing. They got excited about it, and they came out with a new line of monitors that have 9 milliseconds of latency. They're some of the best monitors out there now, okay.

So, people just don't think about it. But you know, if you have an 80 millisecond monitor, anything you're playing on it, a local game, you're going to see some lag in a local game. But people don't think about it because they figure, "Okay, it's local, so lag is not on my agenda to thing about."

Some of the mice you can get that are 15 milliseconds of latency, or you can get mice with 1 millisecond of latency, you know what I mean. It's a huge swing. So, what we're going to do is put the facts. We're going to say, "Look. Here are the kind of things that introduce latency. Please let us know. If you're going to be on a lagging experience, come tell us. Tell us what your system is so we can go and try to test that equipment."

Then what we're going to try to do is put up a list of different equipment and how we've measured it. Not every laptop, but most laptops have pretty low latency screens. So, you know, again, not every trackpad, but trackpads are probably not the best thing to use for gaming. But nonetheless, at least it's built in, and there's a good chance the latency is not too bad on it, right. So, give that a go before you kind of condemn the whole system. [laughs]

These are things that we need to educate people on. The bottom line is this: if you have a good connection to OnLive and your gear is low latency, you have a low latency experience. It works. It really does. It's never going to be low latency as having the exact same computer capability locally, right. I think that's an obvious thing, right. There's a load of latency introduced by the internet.

But the thing that we need to get across to people is that the latency is not exactly what you expected. Sometimes the latency is actually not the internet. Most of the latency is in the last mile. Actually, most of the latency, if you don't have optimal equipment, is in your gear, your monitor, and your mouse. And then the next place you look is in the last mile. And actually, the third place you look is in the internet.

Moving on from latency, there is the business model. So, I'm just going to quote Dave Perry [head of cloud-based gaming company Gaikai]. People compare Gaikai and OnLive, but really, to me, they seem like quite different things at this point as far as the business models go.

Anyway, back in March, he was criticizing your business model. He was saying you're "paying $15 only to have the opportunity to buy the games. So, $15 gives you no games. If you decide, 'I don't want to keep paying that subscription,' you've just lost access to your games bought at full price." What would your argument be about that?

SP: What we said in March is we were going to charge $15 a month at a maximum, and then we would announce promotions before we launched. And in fact we did. So, OnLive is free, and then after the first year, if you want to stay, and it's completely optional, it's $4.95 a month, and it's month to month.

Plus, it's not in the press release, but it's in our terms of service. You can suspend anytime you want. So, if you were going on vacation for three months and you're not going to have internet service or something, we will hold onto your games and your game saves and anything else you've got, right. So, I don't think there's any real cost to using OnLive, and I don't think there's any real risk.

I think in retrospect, I probably should not have even said anything about pricing. We had a lot of people pushing us to say what kind of price it would be. So we said, "Look. How are we going to tell you how much it won't be more than, and then say that it expects to be less than that?" The reason we had to be very cautious about announcing a price back in March is that until then, beta only had back catalog games.

So, we looked at usage patterns, but we really didn't know what usage patterns would be if you have a brand new game that someone's likely to play through the end, right. And the other thing is we had a bunch of different equipment that we were doing from an engineering point of view, and we expected that to go through to reduce the cost, but you really don't know and you can't really get a quote on components, on servers, and you know, we have custom silicon in there, until you get to very large volumes.

So, we said, "Alright. Let's go put out a very cautious number." We figured nobody would be too upset if the price came down. So, in fact, we were able to bring the price way down. And so here we are. And to the point where AT&T is really happy to sponsor the first year.

So, I think that concern that you need to spend money in order to use OnLive evaporates. First of all, demos are free, so there is really no cost in that. And again, for the first year, actually for the first two years, I think you can take $4.95 a month, you know, for 12 months, which would be the second year, and amortize it over two years, it's less than the cost of Xbox Live if you get Xbox Live on an annual basis. And you can't cancel that on a month to month basis.

A lot of the concerns that people have about the cost of the service, month to month, probably should have evaporated at this point. In some ways, you're asking me a question about, "How you would respond to a misconception about old news?"

At this point anyway, [the free year is] a promotional thing, and we might be talking two years from now. It sounds like you guys want to keep it as low as possible, possibly through other sponsorships or partnerships in the future.

SP: There will be such amazing announcements coming. So, we're not worried about people thinking OnLive is too expensive. That's fair to say. [laughs] You know, we have a zero million unit install base right now, so publishers need to be concerned about maintaining consistency across all the different places where they sell their games. 

Games usually are priced the same as they are with other things, with Steam-type systems and so on, or Direct2Drive, etcetera, or retail. But the opportunity with OnLive to package games in many different ways that are difficult to do through either downloads or physical media.

So, you'll see things like, instead of a rental for some number of days, you'll see, if you will, you can pay for a certain number of levels. "Oh, you really do like the game? Well, how about some more levels." What today we think of downloadable content, you know, DLC, I think you're going to begin to see as like these are just option packs that you click and they go right away, you know what I mean. There's no downloading.

And then there's going to be, you know, models like they have with Facebook and Zynga, where the games themselves are free and they're paid for by in-game purchases. I mean, you'll see ad-supported games where there's no cost at all. They're promoted with ads.

And you'll see ads that become games -- somebody has a product or an offering, and you actually get to try it out. I mean, an example would be like a new car, and you click on it, and you take the car for a drive, and it looks just like the real car, you know what I mean. "I think I'll try it with a larger engine option," and see how that happens. Or try different tires and see how well it corners on a really serious race track.

The ways we can package the OnLive experience are just innumerable. There are many, many different possibilities. You know, we are a start-up company. [laughs] We're not Sony or Microsoft. With Sony or Microsoft, or Nintendo, they have enough money in the bank that they can go and say, "Alright. It's going to be three months before launch, it is going to be free for the first people, then $4.95. Then after that, we're going to have this program. Then after that, we're going to have that program."

Unfortunately, I don't have the privilege to do that. I have to kind of wait until all the ducks are lined up before I say, "Okay, here's the promotion that we have." I think people should take what they're seeing with this announcement on June 15 as an indication of the direction OnLive is going.

Like you said, it's a new technology and people have questions, but I think one of the best things that you guys are doing is saying, "Hey, we've got Konami. We have Capcom. We've got Epic and all these game publishers." It kind of gives people an indication that maybe you guys are onto something here if they're agreeing.

Can you kind of describe in general how OnLive went about convincing these publishers to release their games on the service? And how did that go in convincing them to support and put their games on OnLive? It seems like they've been receptive.

SP: There's been secret data centers that we've set up. [laughs] And we announced one of them with BT. People don't realize that damn thing has been up since 2009. All this time, we've been in Europe.

What we did is we certainly talked to publishers, but you don't really push this one through the publishers. It was the developers. It came from the bottom up with the publishers. The developers, we gave them test accounts on these private servers.

And it was kind of funny because we had to maintain confidentiality between each of them. So, they're all going to be surprised with this announcement. They have no idea how many publishers are coming because we kept them all separate up until this point.

So, the way OnLive was picked up by the developers is very simple. They got on, and they started to play it. That's it. They started to use. And that answered every question they had. They took the thing home. They tried it from their own connections. They used it. They hammered on it. They asked us questions. They dug into it.

Some of them put in network impairment simulators, which would go and add packet loss, add latency and the equivalent of congestion, jitter, etcetera. And that's why. You know, this is their brand. Their reputations are going along with these games, right? They're not just going to throw them out there on anything.

And so it was after that that, one by one, they came in.

The developers are bubbling over. And the business guys who are running the companies, are bit more cautious. You know, they're sensible. They have to maintain their existing relationships, and they should. But the developers, what they're like is, "Not only do I get rapid distribution but I can watch people while they're beta testing. I can actually see people while they're playing my game." Which is so cool.

And then the developers, they spoke to their parent companies.

SP: They said, "This is cool. We should be on this thing. We should try it." And then we talked to the parent companies. The business proposition is very good for them. The thing with used games... Before the days of GameStop, the publishers have always had bargain bins. When games get older, they charge less for them.

The difference now with GameStop is those dollars, when the price goes down, don't go to the people taking the risk on creating the content. They're going into GameStop, which certainly has some retail risk and so on, for overhead, that's certainly true. But none of it is going to the publisher.

They don't have a problem with online games being cheaper, you know. It's whatever the market will bear. They're not going to charge more for a game than people will pay. But what they want to be able to do is make it so that the tail of the life of these games comes back to them. Because otherwise, they're very, very motivated to make games that have very short experiences, so they can make as much money as they can before it enters the used game market.

With OnLive, it gives them an opportunity to do that where you can't copy the game. There's no piracy. There's no used games. It gives them the opportunity and makes it so that the dollars that are paid for for games funnel into game development.

They're still subject to competitive forces. If there's a better game out there, people will go to the other game. And if people stop playing the game, they're going to have to lower the price. So, the market will determine the price of the game, and they have to create a game.

So, it's not going to change anything from the point of the consumer and the cost of games to them. It will change things from the point of view of how many games they're going to get, what kind of risk the publishers are willing to take on new ideas, and how sophisticated those games are going to be. That's what's going to change over the next year or two.

You haven't revealed the pricing for the actual games yet.

SP: No, I can tell you. The range is from about $5 to... We have one game that's $60 that's a brand new, very hot game. But most of them are in the middle. It looks just like you'd expect. It's a smorgasbord of games. You're going to find some casual games, a couple casual games, and then... Some very high-end new games. And then there's a bunch of demos that are free.

And then there are some games available for rental, too, which is kind of cool. And you can keep extending the rentals, if you want. So, if you rent the game and you want to play a game because you've got a whole weekend to spend or something, go for it. So, it's another way to get into a game that's less expensive.

Latest Jobs

Manticore Games

San Mateo, California
Senior Software Engineer - Mobile

Sony PlayStation

San Diego, California
Sr. Online Programmer

The Walt Disney Company

Glendale, California
Associate Marketing Manager - Walt Disney Games

Insomniac Games

Burbank, California
Accessibility Design Researcher
More Jobs   


Explore the
Subscribe to
Follow us

Game Developer Job Board

Game Developer Newsletter


Explore the

Game Developer Job Board

Browse open positions across the game industry or recruit new talent for your studio

Subscribe to

Game Developer Newsletter

Get daily Game Developer top stories every morning straight into your inbox

Follow us


Follow us @gamedevdotcom to stay up-to-date with the latest news & insider information about events & more