Paul Berry, the Huffington Post's CTO since 2007, is one of the best regarded tech leaders in New York. After helping build one of the biggest news sites in the world, Berry announced this week that he's leaving AOL soon to focus on two new ventures: A social startup called Rebel Mouse and an incubator called SoHo Tech Lab to goof around with a bunch of different ideas and see what works.
I caught up with Berry this week to learn more about his experience growing HuffPost and what he's planning for his new projects. Following is a lightly edited transcript of our conversation.
ReadWriteWeb: I think a lot of people don't realize how big Huffington Post is and what a technical challenge that can be. What's a current snapshot?
Paul Berry: We're 120 million unique visitors a month, 31-day view by Google Analytics. We're at 1.7 billion pageviews, still growing fast. To give an indicator of the velocity, at acquisition [about a year ago], we were 55 million uniques and about 700 million pageviews. So just by sheer volume of traffic and audience, those are big numbers.
The other piece is the complexity of my CMS, and sort of how wide and deep the technology is. The team that I was leading as CTO of the Huffington Post Media Group, I had product, design, and engineering for the Media Group. There are a bunch of domains that are powered by the technology. When I started at Huffington Post, it was metaphorically day two. We were 3 million unique visitors and 70 million pageviews a month and there were three of us in the tech team. The team that Tim Dierks takes over as the new CTO is about 220 people.
And these 220 people are...
That includes a lot of designers and product and project managers. The core of Huffington Post... we had some innovations in how we would put the team together that were built out of a combination of our own character and culture and out of necessity. I was born in Mexico City, my wife is Bulgarian. International, I always knew, would mean a great deal to me. And in the last ten years and in previous jobs, I started to work out: How can you truly put together a dynamic global team? That was vital to Huffington Post.
The election year growth was driven by figuring that out. It was pretty stressful - we had no money. I couldn't just buy another server. And we had so much to accomplish. And what everyone wants from their tech team is to pull an all-nighter every single night. But you know that's not sustainable, so you know as much as you want it you can't have it. You can actually do it by playing that timezone game and passing batons. That was insanely vital to all of our growth at HuffPost. Literally HuffPost has people on every continent in every time zone. Eastern Europe and Latin America, India, Vietnam, Sri Lanka, Philippines.
What were some of the technical challenges you had to deal with?
Scaling was always a point of pride that we never talked about. And we never talked about security. If you're spending a lot of time talking about security, it's because you've gone through a horrible Gawker hack type of moment, and it's terrible. You do internally talk about security, and you have a security team, and you do a lot to make it happen. But at the board or ops level, if you're talking about security or scalability, you're generally suffering. It's a point of pride that that was never a big topic at ops or board meetings. We had very, very few moments of actual downtime.
The emergence of mobile and the emergence of HTML5 together is what's really interesting.
Personally, I think people are making a lot of mistakes in developing everything as native apps completely, when you can have a thin shell as a native wrapper around HTML5 plus responsive web design. And now you solve the problem. This really drove me crazy at HuffPost. We had so much to do, and then all these tablets kept on launching with different screen sizes and different OSes, and everything we did was native because at the time that was the way everyone was doing it.
And now what I think key companies and developers are realizing is that HTML5 and responsive web designs solves for whichever dimension and whichever OS. And you have to get really, really, really good at it before you can pull that off and still have it be a smooth app. But that's where our focus will be.
Facebook, Google, and Twitter were all fairly frustrated with the media landscape - how slow media companies were to implement stuff, how slow they were to be creative and to push the envelope. And that became the roadmap pillars: Editorial efficiency and pushing the envelope with partners. A lot of the stuff that I plan to take into the incubator and into the new company is that culture of pushing those limits.
So what are these new projects?
There's two parts to it. Both, unfortunately, I have to remain a little stealth about, or I guess a lot, annoyingly. Part of my contract with AOL allowed me to work on things during this transition. So I've actually had a team working on Rebel Mouse for a while. I'm really excited about releasing some alpha and beta stuff in recent months.
Rebel Mouse is the startup company that's well defined - it has its name and its logo and it's a really well-defined concept that we're deep into. The incubator is a way to give us space to throw a lot of stuff up on the wall. It's not meant to be a 500 Startups thing, where there's a ton of companies. It's going to be much more sharing a technology stack and a social approach. And it will be social, web, and mobile that defines the companies that we end up creating. What we'll be doing is trying with a very small but elite and awesome team to take things into prototypes that start to gain real traction and go viral, and at that point, fund those into companies that we build into really big businesses.
My definition of viral is: We don't spend on marketing and ads. And that was another point of pride at Huffington Post. We never spent on SEM, it was always SEO. We never went and bought Facebook ads, we just did really well at social. These things have to have their own organic growth, where they hit this mark where you see them growing by themselves. Then you realize we have something now that we can double down on and go raise money and built that toward a big business.
Are there any specific technologies that have been particularly useful to you at HuffPost?
When I started with HuffPost about six years ago, there was still debate about whether open source would win or not. I think that has been answered. The open source stack - whichever you end up using - you have tremendous potential. It's crazy how much has been built out the last five years. The trick has really been to keep up with those sorts of things the way you keep up with a Facebook, or a Google, or a Twitter, and their product releases.
One of the surprises has been that MySQL - when Oracle bought MySQL, everyone thought it would die - and it's actually very much alive. We use Redis ("sort of a database alternative") a lot at Huffington Post, for example. There are some of these core technology stacks and open-source libraries and etc. that we'll definitely be using at the incubator.