In Depth: New Facebook CTO Bret Taylor Discusses Platform Governance, Social Gaming, Viral Channels, and Credits
Five weeks ago, Facebook CEO Mark Zuckerberg appointed Bret Taylor, who joined the company when Facebook acquired FriendFeed in August 2009, as its new CTO. We sat down with Taylor to talk about Facebook’s vision for the future of the Facebook Platform, and how it’s affecting everything the company is doing.
Taylor shares his thoughts on the state of the Platform, new ways developers should expect Platform governance to evolve in the months ahead, the new group at Facebook that is responsible for the health of the games ecosystem, new types of communication channels that Facebook may launch, and the Credits rollout transition. In the second part of this interview, coming tomorrow, Taylor discusses the the state of the Open Graph Protocol and Facebook’s long term Platform vision as it relates to mobile and regional growth.
Taylor is moving from his role has the head of Platform to CTO at a time when Facebook is nearing 500 million monthly active users as a whole, over 1 million websites have integrated Facebook functionality in some form, and social gaming companies on the Facebook Platform are earning hundreds of millions of dollars in overall revenues this year.
Justin Smith: What are you focusing on in your new role as CTO, and how have your responsibilities changed?
Bret Taylor: My job practically speaking hasn’t changed much yet because there’s a transition period, and the platform is one of the core parts of the company. So until we find someone to fill a lot of the stuff that I was doing on Platform, that’s probably still my 85% focus.
I was the cross functional lead for the Platform group, meaning I reviewed product direction, to some degree technical decisions, and partner strategy. It’s a big cross functional group, and we have a great team working on it and a pretty good roadmap, so I’ve been able to start working on a few other projects. Right now my focus will be on product for technology and the end user, which are very intertwined, like news feed, search, and platform. The algorithm that enables a particular UI is as important as the UI itself. Those are a mix of technologies and products, and that’s always been where my passion is. Now I’m also getting more involved with the technology, which is the main change from before.
One of the things I asked Mark a couple a weeks ago was how do you feel like the Platform has evolved today in terms of the alignment of incentives between developers, users, and Facebook?
This is one area where we’re investing a lot in technology to improve that balance. One of the things that has happened with our platform over the last six months is we’ve had increasing frustration from our longest standing game developers in particular about changes we’ve made to the platform to accomplish various user interface goals. For example, a couple of months ago we made a revision to Events, and as a part of that we put the event creation workflow at the top right of the home page. That moved requests down the page by some number of pixels, which significantly impacted the conversion rate on requests, and that was one of those instances where we weren’t thinking about the system holistically enough. We were making a very reasonable product update in isolation from the rest of the system.
There’s a few big initiatives going on right now which I think will significantly improve the sanity of our developers and improve the product. To date, we have relied on policy and enforcement to accomplish a lot of our product goals around the platform. Where are the buttons, what are the incentives you can put in place, stuff like that. They all are very solid incentives, but at the same time, it’s somewhat like the IRS tax code. The people who understand it the best can get away with the most, and the people who are new to the platform are overwhelmed with its complexity. We have evolved our thinking on that and think if we had better automated systems to detect bad behaviors and prevent the delivery of bad messages, we could reduce the number of policies significantly.
So rather than saying you’re not allowed to do X, Y, and Z with a dialog box in your game, if you’re sending useless messages from your game, we just won’t deliver them, and we’ll give you that feedback. And then you can change the way you send messages to send higher signal-to-noise content. This is something that we just haven’t invested enough in, but we now have a very large team working on spam and quality. That will touch all of our communication channels, and news feed. This is going to be a year-long project though, because we’re not going to remove the policies until we know that the system that replaces it is high quality.
The other initiative is we have a team exclusively focused on games now. Internally, we’ve always known this, but now we’re formally recognizing it, that just like photos, just like events, games are a killer app on Facebook, and a primary part of the user experience on Facebook. We have product managers and engineers who are extremely talented now working on it.
Right now we have a Games dashboard that I would say is pretty uninspired. It works, but it’s not something that I think is revolutionary. We have a team now who is responsible for making games successful on Facebook as a category, and when we make changes to our overall product, we’re going to track the effects on that category just like we track the effects on photos and events on Facebook.
Over the next 3-6 months those effects will be noticeable, I think they’ll have the effect that we won’t inadvertently affect the ecosystem by our own product changes. Then, we can slowly make our policies higher level and more “spirit of the law” instead of letter of the law because we’ll have these automated systems to enforce them in a more natural way.
Since 2007, there have been different periods of emphasis on governing the Platform via automated systems versus policy enforcement. For example, soon after the Platform launched, Facebook created allocation limits for communication channels, but in the following year invested more in increased policy enforcement. Is the final answer simply that there will be both?
There will be both, just because there are some user experience things that we’re not going to be able to enforce technically. When you’re within the Facebook chrome, there’s certain user experience things we care about. We can’t enforce policies on scammy ads technically, but it’s very important to us that people can trust apps within Facebook. But at the same time, we want the set of policies to be small and comprehensible, and something that every engineer at any company developing on the platform can understand.
My understanding right now is that some developers have an internal expert about Facebook policies, and during a product change they’ll go consult the guy that understands it. You shouldn’t need an accountant, so to speak, to follow our policies – it should be more intuitive. But it’s a process because until we get the right automated systems in place, we’ll keep the policies that our automated systems will enforce.
But as we get to that place my hope would be that our relationship with developers will be more hands off, and that those policy violations will be more the exception rather than the rule rather than the nitpicky relationship we have now – which is based on the right intentions but frustrated on both sides. We have some of the strongest engineers on the platform team working on this problem.
What are the success metrics for the new group at Facebook that’s responsible for the health of the Platform games ecosystem?
We’re focused primarily on engagement. For users who use a game, how easy is it for them to find that game and interact with their friends using that game? We’re focused a little less on distribution initially, because engagement is an area where our interests and developers’ interests are 100% aligned. If you’re a user of Restaurant City, you should be able to use that game, whereas with distribution, our interests may diverge a little bit.
The team right now is focused on improving engagement, focused on requests, bookmarks, and some features on stream. We’ve gotten feedback from our users that it’s hard to find apps they use regularly. That’s something that troubles our developers and troubles us equally, and we are developing the standard engagement metrics for games, on the order of 5-6 experiments in parallel. Based on the results of those, we’ll settle on a path. We want to roll out improvements every couple of weeks.
Once we feel like we have that in a better place, we’ll also start focusing on distribution. It is a much more formidable product challenge, because people who are not interested in games have a significantly more negative reaction when they see things like game invitations. I don’t think we have the silver bullet for that, but I think improving engagement first will be a huge positive step forward.
Was the recent change to remove notifications for applications the right move? Do you plan on introducing other distribution channels to take its place?
I think it was for that product, because it was a completely unpoliced channel. It was a tragedy of the commons, where there was enough garbage in there that it ruined it for everyone else. That was the feedback we got from developers. The channel had gotten so polluted that it wasn’t useful anymore. But there is a good chance we reintroduce channels like that that are inherently more algorithmically policed and uniformly high quality. Requests has been a higher quality channel because of the built in controls, though there are obviously improvements we can make there as well.
The idea that there are passive notifications, notifications that require action – requests, and broadcast, which is the stream, are concepts that we are still committed to. But we want to introduce them in the right way so that it can be a sustainable long term channel, rather than a free for all, which in the end hurts everybody. We’re going to be much more thoughtful when we introduce new distribution channels like that.
One of the channels that has been scheduled for updating recently but hasn’t been updated yet is the requests channel. Can you talk about what will happen there?
We are not going to change that immediately. We are in the process of updating our roadmap. This is part of the group working on games. We are doing a bunch of experiments on it first, so we can do it in a thoughtful way, because it is an extremely important channel for our canvas and games developers. We had originally said we were going to do a bigger change to that this month, but we consciously are not going to do that until we do some of our experiments around games, so I don’t think that channel will change any time in the short term.
Is the team responsible for engagement across games also responsible for monetization at all?
There is kind of a church and state separation right now. The primary goal of our platform team today is increasing the number of high quality applications and increasing engagement with those apps. We are really focused on Credits, but actually almost entirely for user experience. Games are so core to the user experience on Facebook that having the ability to transfer the money that you invested in games among games is a very important part of the user experience, and that’s the primary reason we’re investing in that today, is to make the category grow.
One of the concerns that people I talk to have about the way Credits is being rolled out is that there will be a transition period during which the product isn’t yet fully baked and enough payments partners aren’t signed up yet. Can you address the specific concerns that developers have around this?
There is a transition period inherently in rolling out a product of this scope. We do have the most high leverage payment partners today. The feedback that I’ve heard so far, and this isn’t my area of expertise per se, is largely around user experience and conversion rates related to that, and less focused on payment partners. There’s gaps in that, but it’s relatively complete.
Because we have a number of partners using it now, the product will improve for everyone really quickly because of the collective pressure. The long term vision is because there are so many large partners using it, there will be a lot of stored value in the system, which will really help the startup costs, particularly for independent game developers. Some of the best games on Facebook have been created by independent game developers, which is one area that’s been a little underserved. The existence of a high quality Credits system will really help a lot of those shops in particular.
We’re adding payment partners based on demand. If there are payment partners that aren’t supported today that are really important to your business, talk to the partner manager that you’re dealing with on the payments team, and we’ll work on it. We’ve added tons, and we continue to add more. It’s a huge priority of the team. There will be a transition period that will be kind of awkward, but in the end everyone will benefit from the system.
Do you think that third party monetization services should change the way they think about Facebook in order to be better aligned with you?
At a high level, companies that help platform developers monetize are great, and it’s great that there are companies that give developers a lot of options to monetize their product. With our Credits product, it really is a user experience thing. We’re certainly not the only platform to want a uniform currency – almost every device has something similar. Basically, users who use applications tend to use more than one application. It’s very important therefore that you don’t have silos of cash and are able to use lots of applications and have a single virtual currency to use among all of them. That just seems like the right user experience.
Certainly on Facebook.com, that’s the direction we’re going without question. There’s just very little question in our mind that that’s the right thing for our users. Other things around user experience that we’re really sensitive to are data sharing. We really don’t want sleazy ad networks, so we have limitations around data flow, but at a high level we are very supportive of third party monetization options on our Platform.
The entertainment category of apps has generally performed above expectations, but it seems like there have also been other categories that have apparently been below expectations, like enterprise or productivity. Would you agree with that?
I would. I think what happened with games was that a handful of very innovative companies created really compelling user experiences, and they were so compelling that they grew tremendously, and provided a template for how you make social games that in my mind almost created an industry. And I don’t think we’ve seen that same behavior in a lot of other verticals that we intuitively think probably will be disrupted by social platforms eventually.
That’s a lot of what we were doing at f8 with our partner program. For instance, for our instant personalization program we worked directly with partners on helping them create social experiences that other companies could use as a template of what a great social experience on the web is.
One thing that I have an intuition about is that companies that build social from the ground up, they think about it from day one, come out significantly differently from the companies that tack it on at the end. That’s why I think a lot of the biggest social companies today were startups initially, because it’s very difficult to treat social as a transplant in your product. It’s not impossible, companies have done it, and so I eagerly read Hacker News looking for the latest Y Combinator company in verticals I care about because I think there’s a good likelihood that a company like that finds the right social incentives and social interactions to make a vertical really explode socially.
That said, I think a lot of the work we’ve done to support Facebook for websites has done a lot to improve the state of things, in terms of the number and breadth of categories that are socially enabled. Starcraft 2 uses Facebook for login, and I think that’s great, because I like to play video games but I’m pretty bad at them, and playing with friends is much more fun for me than getting my butt kicked by a 13 year old in Iowa. Everything from that to almost all the newspapers I read have integrated social plugins. I don’t think we had the right plugins for them until f8. We’ve chipped away at that by expanding outside of the Facebook canvas.
In the second part of this interview, here, Taylor discusses the the state of the Open Graph Protocol and Facebook’s long term Platform vision as it relates to mobile and regional growth.