Gaikai

A few months ago, I had posted here asking about startups, and appreciated the responses given. It helped me look deeper into an opportunity that came up, and guided my decision to join.

With the Game Developers Conference starting next week, the company I joined will be showing the service is alive and running. Gone are the selective beta gates, and instead we are letting anyone within reach of our servers to try it out. But what is it exactly?

Gaikai is a service that is basically remote desktop for high end applications, like video games. A client connects to our servers, and the server runs the game, collecting input from the client and streaming the video and audio back to the client. We aren't the first in this space, but we have a different approach then our main competitor. Instead of selling games direct to end users, and having them always play via the service, our approach is one of a demo service. We want to enable game companies and other high end application producers to allow end users to try a demo without friction. Instead of requiring a large file download, install process, and possibly a high spec machine, our tech enables a company to just embed the demo in their web page. This allows someone to try a demo for free nearly instantly, and then they can make a more informed buying decision.

One of the important questions from the startup thread was about business feasibility. Having worked in the games industry for 5 years, the demo issue is a big problem. As games get more complex, it becomes harder to package up something that represents the full experience while also keeping file download sizes small. There is a surprising number of users who will sign up for an MMO trial, but never make it in game to give it a go. With the Gaikai service, file sizes no longer matter to the end user, and it's a much quicker turnaround time from interest to actual hands on game time. Seeing the tech in action during my interview proved to me it could be done well. So right there I knew there would be interest, and Gaikai would have customers in the form of game companies.

The other part of feasibility is of course the ability to eventually turn a profit, and ensuring there is enough initial investment to keep the service alive until that day arrives. Gaikai's approach is to tap into the existing marketing budgets of games. We charge based on per minute usage of the service, and the costs are competitive to running banner ads on major sites. The team has run the numbers, and seems to have found the right balance between how many servers are needed and what can be reasonably charged for the service. The initial two rounds of investments have been completed, and some big names were involved, including Intel and Limelight networks. So far, things look solid on this front.

David Perry is our CEO, and he's been around the industry for a long time. He brings a lot of good business sense to the table for us, and also knows the right ways to run a development shop. He cofounded the company with two technical people, one of them being the creator of the base tech that powers the Gaikai service. It's a good blend of technical knowledge and business sense at the top, and after interviewing with them, it was clear to me they had a pretty stable startup.

If you want to give our service a try, we have a few ways of doing so:

1. EA is running a beta of a Sims 3 trial
2. Gaikai is running a short gaming related survey before allowing access to a Dead Space 2 demo
3. Linden Labs is running a trial of Second Life
4. Gaikai.com also has Mass Effect 2

#1 and #2 will show our testing process (bandwidth, available plugins to use, proximity to a data center), and is the best way to go to see if you can use the service currently. Numbers 3 and 4 represent how the service will work as more customers roll it out. The idea there is that the same tests happen silently in the background while a user is browsing a site, and if the tests pass, the user is presented with an in page popup inviting them to play. If conditions aren't right for the instant service, the user is never notified they missed something, and will proceed with the normal way of evaluating a game.

For more insight on the company and our future plans, David posted more info to his blog.

There is also a recent episode of The Jace Hall show where Jace interviews David Perry. It's on youtube here. The Gaikai part is at the beginning, and resumes at the 4:26 marker. Fair warning, the overall show is not safe for work. Oddly, it also has Jace over at my former employer playing DC Universe Online.

Moderators, this seems to be spam.

Hey, I have to pay for this BBS somehow

In all honesty, I'm pretty excited about the possibilities with tech like this, and the job melds my joy of games, and enterprise scale technology. The longer I'm here, the more I feel like I made the right move. I was pretty happy at my former job, but this one has opened me up to a lot more unique experiences. I know we have a few other gamers here, so I'm curious to see what their impressions are.

Now to hope it makes it big. Having EA out of the gate is already pretty major. I'm looking forward to where things go from here.

I also need to find a more interesting place to live. While I'm happy at work, suburbia is not the most exciting place to be.

I hope you've got this pinned up in your offices somewhere

Peter

Is this like onlive ?

Awesome, glad you ended up at Gaikai.

I was beta-ing OnLive for a while, which I considered quite awesome, but the problem is that they didn't have many *games*. And then they went live, and then no new games showed up. I played through Borderlands and that was *it*. They were supposed to get Dragon Age Origins, and it never materialized. So I'll have to check out the game selection on Gaikai and see what's up there.

Exactly. Just a different company with a different game selection. Hopefully a larger one.

Tried Mass Effect 2 Demo, ran well.
Tried Sims demo, got "Error, we broke the internet" message and Gaikai exited before I got a chance to try it.

That happened to me when I tried to full-screen the Mass Effect demo.

I tried the sims demo but they told me my connection wasn't fast enough. 10 down 1 up according to a couple speed test sites.

It gave me the "broke the internet" error only after it had already pre-approved my connection speed.

I am not a gamer so I didn't really play the demo once it started. Little problem with not knowing the key mapping.

I tried the Dead Space 2 demo. Everything seemed to work fine. During the prolog, Comcast dropped my bandwidth, and the playback glitched, but the server seemed to adjust and carry on. That might have sucked if I'd been actually gaming and gotten slagged during the glitch.

Edit: The survey was a bit fun. One of the questions asked me to admit that I filled in required personal data with garbage to avoid entering personal data. I did so admit. The at the end, the survey required an email address, I of course put in spam@gmail.com

haha, I had forgotten about that comic. It's going up on my wall Monday


Similar concept, but a different business model. Our goal is to enable game publishers, and retailers a way to embed instant demos into their web sites. After playing, you would need to still buy the game at a shop or online.


The tests that run look at the practical bandwidth and latency stats to the data centers we think are near you. It's possible you are either too far away, or the internet routing from your provider to our servers is not as optimal as it could be.

Every time someone attempts to play (either via the direct beta test sites like the Sims 3, or just taking the silent test on Gaikai.com) metrics are gathered to decide where to build out a datacenter next. It's also used to try and address any odd routing we notice off our network. The service is still technically in beta,

When Engadget talked about the service back in closed beta, they also were allowed to publish one of the heatmaps we generate from those stats.

The map is out of date now, but gives you an idea of how we are moving forward. They pulled in some pretty experienced people from the industry to make sure each aspect of the service is done well.

Interesting map. The huge gap in the middle is curious though.

The midwest is very sparsely populated, and also tends to have very poor internet options. I didn't really understand how sparse until I had a vacation out to New York, and then drove out to Ohio with Rob R. He also made comments about it when I picked him up from the Denver airport and drove back to Colorado Springs.

Even taking that into account, the way the drop off line goes almost straight north/south is curious. Maybe the mid-west is sparser then I think it is...


...So it is, wow:


It is as if the settlers gave up half way across and flew the rest of the way. Thanks for the education Tom.

Looks pretty much like any population density map.

It also a lot more vertical, being the area of the Continental Divide.

The early settlers didn't think much of farming on top of big rocks.

Okay, first: that's not the midwest. The midwest is Ohio to Kansas and points north. The area you're talking about is more accurately the Great Plains, though even that is not a perfect descriptor.

As to why it's so sparsely populated, read about the Dust Bowl.

Depends on where you turn for the definition of midwest. The official census definition is where the Ohio to Kansas part comes in, but I know plenty of people who think of eastern Colorado, and Oklahoma to be part of the "midwest", more from a cultural standpoint.

Bitt, my knowledge of US geography is admittedly dreadful. Can you point me towards a map that has areas like the mid-west, great plains etc labeled?
Which reminds me I need to get my map of the states back on the wall.

http://en.wikipedia.org/wiki/List_of_regions_of_the_United_States is a good resource for looking up what people might mean when they say rust belt, or the now rarely used frontier strip.

http://www.architectureschools.com/m/regions.gif

Here's one with different names for the regions to make it more confusing for you

http://www.mapsofworld.com/images/usa-regional-map.jpg

I think the regions are kind of arbitrary.

Thanks Tom & Matt.

Why would you play computer games, when you can hang out in a corn field!!!!

Specially since you can play baseball with legendary baseball players.

"Out past the corn fields where the woods got heavy, out in the back seat of my 60's Chevy."

http://www.tested.com/news/gaikai-goes-live-offers-streaming-video-game-demos/1915/

The first one doesn't even list the Pacific Northwest as a seperate region- it just lumps it all in with California boooo. It also makes questionable decision and puts Missouri in with the South- also very iffy and likely to inflame nerves. Missouri was a border state during the Civil war for a reason...

Yeah I don't see how Minnesota is Mid-West either. It's north east of the center of the county.

Which is interesting, because that's only going half way with the potential of the service.

During the Beta, I was incredibly interested in the OnLive service because it would have enabled me to play games that I wouldn't otherwise have been able to play because of underpowered hardware. That was its selling point for me.

For instance, I played all the way through Borderlands on OnLive with fairly high graphic settings, when I wouldn't have been able to do so on my work laptop normally. I played through all of Crysis:Warhead on OnLive with a decent frame rate when it would have been nothing but a slideshow on even my desktop machine.

OnLive worked incredibly well for this purpose (actually playing games), *AS WELL AS* providing live demos of the games. The demos were interesting because they weren't manufactured-demos, they were just a limited-time use of the *actual* game. That, too, was a big deal to me.

If OnLive had had more actual *games* available to them, I would have bought some. I had actually *pulled the trigger* on a pre-order of Dragon Age: Origins, and then poof it disappeared from their games lineup on launchday and I got credited for it.

I guess what I'm saying is that Gaikai is missing the boat on one possible monetization of the service. I'm sure they're studying OnLive's business model and deciding that people like me are in a minority, and that it's not worth their time to try to go after me as a market segment. Which I can't fault them for.

I just think that if services like OnLive and Gaikai could strike deals with publishers for larger game libraries, then that side of the monetization would be viable.

A lot of that is related to the westward expansion. It's all relative to the established population of the eastern seaboard where everything else is West.

The second one isn't any better. Alaska is lumped in with the Pacific Northwest, despite being separated by 800 miles of Canada (and that's just to get to the lowest point, which isn't even close to the bulk of the state). North and South Dakota are lumped into the Rocky Mountain Region, even though the Rockies end half way through Montana, and have much more in common with the Central Plains, than they do the Rockies (here's a hint... on the western-most edge of North Dakota, is the Little Missouri National Grassland). Heck... you're probably closer to the Canadian Shield, than you are to the Rockies.

Whoever put those regions together has obviously never been to the places they're categorizing. They're also missing the border between WA and ID, and have an extra "n" on the end of the "Great Lakes Regionn". Not a stellar example of map-making.

Well, it's good to remember that political and physical geographic borders frequently fail to coincide. To use a recent example, I think that Milwaukee is definitely part of the midwest, but the plains and highland areas of Wisconsin probably are not.

Oh, certainly. But some of those regions don't seem to correlate with either.

Exactly. Attempting to define "regions" based on proximity of state boundaries to particular geological features is foolish. Regions are murky because they depend on history and culture as much as they do geography. Sometimes, if you're lucky, those coincide with state borders.

Bumping this back up, as it's been a very wild ride working at Gaikai over the past year. Looking back at my post from February, it's pretty amazing to me to see the progress made as we come to the end of this year. Back then, we were just doing some basic demos, now we have full agreements in place that have lead to being integrated into places like Walmart, EA, Origin (EA's online digital store), Escapist Magazine, and most recently, Youtube.

The tech has also advanced, with our service able to handle even the heavy hitters like Crysis 2 at full quality. The web site changed a few months back to really help showcase what we are doing, complete with every game we currently host being on one page (http://www.gaikai.com/games). No more hunting for stealth launches or blog posts, it's open to everyone within range of the expanding network.

Oh, and we recently moved into a new office that has some nice perks. The temptation to just go into work and play is very high now.

It's up on my new cube wall now too, though I photoshopped it to say Gaikai :-)

Dumb question: what's the latency on this thing? For twitchy games, is the latency enough that you can really notice it? Also, what's the bandwidth you need for a "high quality" game at "high resolution"?

That's really cool that you guys are doing well. Changing jobs is never an easy decision, and going to a startup carries even more risk. Glad it's working out.

I've never worked for a startup, but I did get a lot of great advice from people here on the BBS when I made my own decision to change jobs back in 2004. I'm very glad I pulled the trigger, even if there was some risk involved in moving to a new city and starting out as the new guy in a group full of people that were much smarter and more experienced than I was. Now I'm one of the older hands in the group, and the work is still challenging and rewarding some seven years later.

Very glad it is all going well, Tom! It is an awesome concept, and it looks like it is growing well.

Small companies can be so much fun! (admittedly I got sucked back into corporate world - now doing the same as I was when running my own company. Cons: the things you expect in a company of 170,000 people, like less fun, less excitement. Pros: much lower risk)

Surprisingly low latency, considering what they're doing. The technology is impressive. I played through all of Crysis:Warhead and Borderlands (okay, I was using OnLive for that but I've played Gaikai games to and they are similar), and although I could feel the latency it was never really annoying. I couldn't believe how quickly this process completes:
- My mouse click
- upstream to the server
- Game's own inherent frame-by-frame latency
- Game screen rendering to buffer
- Video encoding
- Video frame sent downstream to my PC
- Decoding on my end
- First frame of gun-firing animation on my end.
That's a huge amount of shit to wade through, and it felt to me like it was in the 100-200 millisecond range. That's just amazing.

I wouldn't do Unreal or Quake deathmatch that way, not because I don't think I could do it (it would certainly work), but rather because 100-200 milliseconds is the difference between win or lose when you're up against other human players. I suppose if all the other players were on the same streaming service so that they all had the same latency, then I would have no problem deathmatching that way.



DSL or cable modem are fine, as long as you're not streaming Netflix or doing a big FTP/BT download on the same line. If I had someone else doing something major on the household net connection, I wouldn't be able to play.

In the end, I was very interested in the technology and was seriously considering playing a lot of games that way. It's a perfect way to experience single-player games. The only reason I'm not doing it right now is that OnLive hasn't added any decent new games since they launched the service a year ago. It's simply a content thing. If they add some new games that I want to play, I'll log right back in and pony up the dough. And of course, Gaikai is positioning themselves differently, doing pretty much just game demos instead of full games, so that's a completely different thing. I'll try out any Gaikai game demo without hesitation, knowing that the technology works and that I'm getting an accurate representation of the game.

Without going into NDA territory, I will say the tech behind some of our our competition is different then the tech behind Gaikai. And they are all evolving and finding new ways to handle latency. On the Gaikai side, I've been able to see the improvements made over time to keep shaving little bits off here and there.

Bandwidth wise, as long as you can stream an HD (720p) video from YouTube without issues, Gaikai should also be just as smooth. Bandwidth can be very finicky based on how the internet works though, with packet loss and different routes impacting the practical bandwidth between a user and our servers. WiFi is also somewhat problematic at times due to how quickly the available bandwidth can change. Makes it all that more challenging to engineer these systems to be reliable

Tony, I'd be curious in your direct experience with Crysis on the different solutions out there.

It was the Crysis:Warhead sequel, and with OnLive, my only complaint was that the machines that were actually running the game weren't top-end, so the graphic settings weren't cranked. In other words, from my point of view, one of the best reasons to play Crysis on a cloud system was that the cloud server could be beefier than my puny home system. Which it was, but not by much. So I didn't really get to see Crysis in its full glory. Other than that, it worked just fine. I could feel the latency a bit, but for a single player game, that only means a slightly less crisp control feel.

Clearly what needs to happen is that the backend should be a quantum cluster that performs every possible action every quantum of time and then provides you with whatever result happened due to the action you actually made.

Sorry, should have been a little more clear. I'm wondering about your experiences between when you played the Crysis:Warhead game vs Crysis 2 on Gaikai. It's only the demo on Gaikai, but I'd be curious to hear if you notice any subtle differences between the two.

I didn't do a back to back comparison. I'll poke at Crysis 2 on Gaikai and report back.

I am trying from outside the US though.

As long as you just test the service and not the game itself as Crysis 2 was ridiculously disappointing.

I had a go on it last night and it worked fine in the UK.

Gaikai sounds like exactly what Sam needs

http://www.bbc.co.uk/newsbeat/15627247

My problem with Crysis 2 on Gaikai was that once you get through the splash screens and intro videos and tutorial you're left with maybe 3 minutes to run around and get shot at before the demo time ends.

This looks like it will be interesting:
http://www.engadget.com/2012/01/10/gaikai-partners-with-lg-to-power-smart-tv-gaming-service/

That is cool.

It was fun watching this come together. The newer TVs in the pipeline have really slimmed down, and the electronics inside appear to be so simple due to the SoC technology being built in. One more device down in the vision shown off on the Gaikai banner

Nice development again. It seems Gaikai will now also stream complete games and not only demos. There will also be Facebook integration. Cool!

Sadly I still can't run Gaikai demos from home. Despite being able to happily play most current online games without any issues, Gaikai tells me I have packet loss:-(

Still think it looks exceptionally cool, and when my broadband gets above 1Mb I may be able to try it...

Ok, it's taken me a while, but now I've tried Gaikai for the first time. It works fine for me. I like the idea and the interface. I also like the cloud idea. I may be wrong, but Crysis 2 ran very smoothly on my Core2Duo system. And my video card is probably the oldest part of my entire system (a gforce 6600GT, but I don't game any more so I don't need anything more powerful) but still everything was fluent. I also like the ease of having the games at your fingertips should you decide you want to play a game. I see this going a long way, I really like this concept.

http://www.scei.co.jp/corporate/release/120702_e.html

Err... congrats?

Hope you get a big fat check Tom!

It will be interesting to see where it goes from here. I left late last year, so I won't be a Sony employee (again). I do think it will be interesting to see what a console maker does with it. And out of the three console producers, Sony is probably the more interesting one as far as Gaikai tech goes, due to their market presence in TVs, tablets, phones and computers too.