CNET también está disponible en español.

Ir a español

Don't show this again


Innovation comes cheap, says Google engineer Kevin Marks

The software engineer, who works on the OpenSocial and Social Graph API projects, talked to CNET at the Future of Web Apps conference in Miami.

MIAMI--When the Future of Web Apps conference wound down Friday night, a few things were clear, not the least of which is the fact that open standards are a big deal.

Google engineer Kevin Marks gave a talk at FOWA about application program interfaces (APIs) and Google's role in the developer community. Marks, a veteran of blog search start-up Technorati, now works on the search giant's OpenSocial initiative, which is working toward a universal standard for social-networking standards and is slated to launch on and Hi5 as well as Google's own Orkut soon. He also works on the Social Graph API, which aims to consolidate data across disparate social networks.

On Friday, amid a frenzy of chatter about open this and open that at FOWA, Marks took some time to chat with CNET about OpenSocial on MySpace, that wacky Silicon Valley exuberance, and his view that a tough economy won't hurt innovation--because the cost of innovation has gone down.

P.S.: No, I wasn't given any restrictions on what to ask, and we didn't touch upon the subject of health care, in case you were wondering.

You've been speaking to a lot of developers at this conference. What has your most consistent advice for them been?
Kevin Marks: I say, "OK, stop and think about your application. Do you really need to be a standalone site? Do you really want to write user registration code, or would you be better off taking your application and bringing these other sites where there are lots of users already and where they have already expressed both their personal information and their connections to other people?"

Would you encourage them to use OpenID?
Marks: OpenID is closer to the Social Graph API, because OpenID is like, "I own this URL."

Do you think there's any kind of conflict between the Social Graph API and OpenID? Will they be able to co-exist?
Marks: No, OpenID and the Social Graph API are perfectly complementary. OpenID is one of the things that we index as a connection, and if you use OpenID to say, "Yes, I own this URL," that means that you can call the Social Graph API and see what assertions spread out from that URL. So it'll give you more confidence that the person is who they say they are. It's not me giving the site your URL and impersonating you, because you verified that URL.

There's obviously a lot of excitement about openness, but with all this talk about it, do you think there's too much exuberance? Do you think people are overlooking anything that's going to make it a tougher process?
Marks: No, I think it's a healthy trend. The point is that open standards are better than closed standards, and open-source software is better than closed-sourced software in lots of ways. There's a lot of good to doing stuff in public and doing it in the open and getting community feedback, and that's a big part of OpenSocial particularly. We started that out, we announced it in November, and basically said, "OK, this is out in public now, we want to have a public debate about what we're doing, this isn't some secret thing we're going to hide for a couple years and then ship. This is out there."

We've been iterating with different developers, going around and finding what we need to change and updating that for the last few months, and now we're at a point where we've got some code that we're ready to release on three very large social networks over the next month. So that process has been out there, and it's been getting more open over time.

Have you been addressing a lot of questions about why the Orkut launch was delayed?
Marks: There's delayed and there's delayed. That team was making the decision "can we support this now?" and "is this ready to do?" I haven't seen a lot of complaints about that. What I have noticed a bit is they were waiting for a big launch to come back and look at this again to write their apps. I'm expecting that will change once we have Orkut, MySpace, and Hi5 within the next month. Those are three very active "sandboxes" and there are a whole lot of developers in there.

The MySpace platform launch is coming in a matter of days. Can you give any hints as to what we might see, what might surprise us?
Marks: I think the surprises will be how users interact with it, because that's the stuff you can't know until you do a big launch. I can't give you any code surprises, but what we'll find is that users will start using it and developers will start realizing that it fits in different places and there are different things you can do.

One of the things that MySpace has that is interesting is that you can install applications both on your profile page and on your user page. So you can have applications that are sort of performing to others, and applications that are shown only to yourself so that you can analyze things. If you think about the social networks, there is this split between public performance and private interaction. Some sites are all public performance and everything happens on the profile, and some sites there's much more of a reflective view of showing the user what's going on. MySpace has both those pages.

And OpenSocial has these abstractions that will tell you where your app's running and what the context is. So you can write the same app but it will give you different things in both contexts. It'll do one thing when it's on your profile showing to the world, and another thing when it's on your page just showing things to you. One will be outward facing, one will be inward facing.

Do you think that Facebook is gravitating toward a more open model?
Marks: From our point of view, we think it would be great if they all did OpenSocial. From their point of view, they've got an API that fits their site very, very well, because they designed it around that. OpenSocial we designed to be this abstract generalization that fits a lot of sites, and that's a lot of the value it brings developers...For Facebook, that may not be as attractive to them, but I suspect it will be attractive to developers. We've already seen somebody build a "run OpenSocial inside Facebook" thing as an experiment, and I expect we'll see more of that...Bebo's running both APIs so I expect that will be an interesting place for people to experiment as well.

Do you get a lot of requests about interoperability, so that for example an application on a MySpace profile could communicate with the same application on a Bebo profile?
Marks: It's one of those things that people talk about, bringing the users between sites. But that's one of the things that actually quite hard to do, because there are two boundaries to overcome. One is that there are differences between the sites, and the other is the users' privacy concerns, which is why they've got different accounts on different sites anyway. The Social Graph API works with the publicly articulated things that are out there and connect them between sites, but that's there to work. Doing that between the private ones is a harder problem because you've got a permission barrier in each case.

It's something that we could potentially do, and the part of the Social Graph API that does the profile ID mapping stuff and canonicalization could be used to do that, but you've still got to ask the user and connect them and things like that. And if the stuff's not public, you've got to not just ask the user, you've got to ask the user's friends about bridging the stuff. Sometimes people blur the difference between open and public. You want your code to be open, but you don't necessarily want all the data to be public because people have explicitly given it to the social network with the trust that they'll treat it in a certain way.

What's it like working with an OpenSocial partner that is very concerned about maintaining a very uniform look, feel, and attitude on its site--like LinkedIn, which is very focused on keeping things strictly professional?
Marks: Each container obviously has the ability to police which apps run on their site, and we expect to see some variations there, with some being wide open and some having "white lists" and some having "blacklists."

Have you turned down any requests from particular sites that want to be OpenSocial partners, either something that's controversial or something that just doesn't fit?
Marks: No. Would we do that? I can't even think of how we would do that, or why. It's an open standard, an API. They can check the code out and build it themselves. I expect we will see all kinds of different sites doing it. One of the interesting things has been seeing that Oracle was interested, and was interested, which you don't think of in the same breath as Facebook or MySpace, but they have a large collection of information about people that's correlated together and it makes sense for them to have an API to do that.

You gained a bit of blog buzz for saying, "Before you think about your business model, think about your pleasure model." Where I'm from, in New York, that sort of thing makes us roll our eyes and call it Silicon Valley bubble-speak. Do you ever get criticized for that view?
Marks: I first said that when I was talking to a bunch of nonprofits. I was at a conference before I joined Google, and these (nonprofits) were talking about how they can work on the Web and work on their business models. And I was like, "What? Where did that come from? You're charities! You're not supposed to be 'businesses.'" A lot of it is that people think they have to put a business in it, and show revenue, and put something out...(but) you have to work out what it is you're doing that will make people want to use your site and come back to it, and why that's useful and interesting. And then, later on, you can say, "If I've got a lot of interest in this, I can probably make some money from it."

Now it does sound fantastic, because you've got to invest a lot of time and money to do something, so why would you start a business that way? Part of the point of this Web stuff is it's lowering the barriers to entry. You can build an application much more easily. You can put something up and see if people like it or not, and tweak it. One of the points of OpenSocial is to make that stuff even easier because you can build an application without having your own server, you can run it inside the social network itself, you can let it store data in the social network's site, and later on you can decide, "OK, this is interesting, I've got a bunch of users in this app, I should connect them to an external server, I should work out ways of serving advertising or something."

That's a lot of where this comes from. It's not that in the Bay Area you have loads of money all the time. It's actually that the threshold for launching some of these things has gone down a lot, so it's no longer "I have to go and get myself a hundred million dollars" and spend two months, two years, three years doing product development and then launch the stuff...You can start stuff and play with it. And that's a lot of the message of this (Future of Web Apps) conference, is that that is a way of thinking and working.

And you think that's still going to hold true if the economy continues to be so volatile? Obviously a lot of people are talking about Google's numbers in January and Web 2.0 in general when it comes to this.
Marks: A lot of these sites were built after the dot-com crash. If you look at the history of Flickr, it was built in a firm in Vancouver where their consulting business was a bit quiet and they could build stuff on the side. That's part of this. The disruptive stuff is always at the edges and the margins and being done by people on the outside, and Google and other companies will try to do that internally.

Google has a very strong culture of internal innovation and will see that stuff, but there will always be people with a bright idea, and if they've got the ability to execute on that, and we can provide open-source software that helps them do that, that improves the whole thing. A lot of this stuff is the "improving the Web" thing, and that sounds like hippie West Coast nonsense or whatever you want to call it, but the point for Google is, if we can improve the Web and make the Web better, more people will use the Web, and anyone who uses the Web uses Google.

And there are still no plans to profit off of OpenSocial?
Marks: It's not a "Google open social Web," it's an open social Web, and this is part of our help to catalyze the standardization and help it converge in the same way that we're working on HTML 5, we're working on TCP-IP standards and a whole bunch of other standards and open source projects in Google, because they're complementary to our core business. By making the Web and the Net better, there's this nice feedback, and Google is far-sighted enough to do that and has enough money that it can keep that cycle going.

We're investing in solar energy. If you look at that, and ask why we'd be doing that, well, we use a lot of energy. If we can make energy cheaper, that's good for us, and if it makes us good for everyone else in the world, that's also a nice side benefit.